Evaluation of patient reporting of adverse drug reactions to the UK 'Yellow Card Scheme': literature review, descriptive and qualitative analyses, and questionnaire surveys

Findings from the survey and literature review

International experiences of patient reporting

Blenkinsopp et al. 24 completed a systematic review of patient reporting of suspected ADRs, which was first published online in 2006. Seven studies were included, although none involved spontaneous reporting by patients. Where comparisons were available with HCP reports, quality of reports appeared to be similar. There was some evidence that patients were more likely to report ADRs if they felt that their HCP had not acknowledged their concerns. We now report further on international experiences of patient reporting of ADRs drawing on literature from before and after this systematic review.

In Australia, patients have access to the ‘Adverse Medicine Events (AME) Line’,25 through which they can report ‘suspected AMEs, possible errors or “near misses” with their medicines’. Patients can also report to the Adverse Drug Reactions Unit (ADRU) of the Therapeutic Goods Administration (TGA). 26 The AME Line is operated by the Australian Council of Safety and Quality in Health Care and was launched in 2003. Suspected errors and ADRs are reported using two different forms; thus, reporting rates for each can be analysed separately. A recent case series of patient reports to AME relating to the use of zolpidem indicated that memory disturbances, hallucinations and dependence were more common than was previously thought. 27 An audit of the use of the AME Line showed that 43% of the 3415 calls received in a 2-year period were prompted by media publicity. 17 Females and older patients were more likely to call the line. One-fifth of ADRs reported by callers were previously unrecognised and 8% were related to complementary medicines. In total, 105 serious ADRs and drug-induced hospitalisations that callers reported had not been reported by health professionals. 17 The case series and audit were reported as conference abstracts and, as such, very limited information is available regarding their methods and conduct.

The Canadian Adverse Drug Reaction Monitoring Program (CADRMP) at Health Canada is responsible for the collection and assessment of adverse reaction reports made voluntarily by health professionals and patients. 28 In 2008, 20,360 adverse reaction reports were submitted, but these were not limited solely to medicines. 18 The majority (72%) of these reports were submitted by the market authorisation holder. Of the 16,272 reports for which a reporter was identifiable, 4851 (29.9%) were submitted by ‘consumers’ and ‘patients’, and 10,898 (67%) were submitted by health professionals. In total, 1719 (10.6%) reports were submitted solely by patients. The number of domestic reports for ADRs has nearly doubled between 2001 and 2008. 18 A quarterly newsletter is produced for HCPs and patients; the former receive mailed paper copies, whereas the latter need to access the publication online. These data are presented as annual reports, which do not include detailed methodological information.

In Denmark, patient reporting of ADRs commenced in 2004 and the reports received from 2004 to 2006 have been analysed. 19 The analysis comprised 6319 reports and 15,531 ADRs in total, with 544 (8.6%) reports submitted by patients. Highly significant differences were shown in the proportions of ADRs classified as ‘serious’ or ‘non-serious’ by different types of reporter (patient, physician, pharmacists, lawyers, other HCPs) (p < 0.0001). Patients and physicians reported similar proportions of serious reactions (46% and 45%, respectively). Compared with all other types of reporters, patients were significantly less likely to report ADRs relating to the following: blood and lymphatic system [odds ratio (OR) 0.22, 95% confidence interval (CI) 0.08 to 0.59)]; hepatobiliary disorders (OR 0.14, 95% CI 0.04 to 0.57); infections and infestations (OR 0.45, 95% CI 0.23 to 0.87); investigations (OR 0.71, 95% CI 0.50 to 0.99); and pregnancy, puerperium and perinatal conditions (OR 0.33, 95% CI 0.11 to 1.06). However, compared with all other types of reporters, patients were significantly more likely to report ADRs relating to the following: ear and labyrinth disorders (OR 2.09, 95% CI 1.01 to 4.30); nervous system disorders (OR 1.27, 95% CI 1.05 to 1.53); psychiatric disorders (OR 1.70, 95% CI 1.31 to 2.20); and reproductive system and breast disorders (OR 2.02, 95% CI 1.13 to 3.61). Differences were also shown between patient and other types of reporter regarding the types of medicines for which they submitted ADR reports. Although patients were more likely to submit reports for medicines used to treat the nervous system and sensory organs, along with antiparasitic products, they were less likely than other types of reporters to report products for blood and blood-forming organs, anti-infectives for systemic use, antineoplastics and immunomodulating agents. In addition, it was noted that patients reported nine ADRs that had not been identified by other reporters, including dysgraphia, parosmia, and thromboembolic stroke, all of which are regarded as serious. As this study focused solely on ADRs that were classified as serious, it does not reflect the full range of ADRs reported by consumers (and HCPs).

In the Netherlands, patients can report suspected ADRs directly to the Netherlands Pharmacovigilance Centre (Lareb),29 a government-run organisation. Of the 6305 reports received in 2005, 819 (13%) were submitted by patients. 29 A comparative study was conducted in the Netherlands of all suspected ADR reports from patients and HCPs, based on reports received between April 2004 and April 2007,30 i.e. the first 3 years of consumer reporting. A total of 2522 reports (5401 suspected ADRs) were received from patients and 10,635 reports (16,722 suspected ADRs) from HCPs. Patient reports were more likely to be submitted by females; the mean age of patient reporters was 48 years. The types of drugs most often reported by patients were statins, selective serotonin reuptake inhibitors, beta-adrenoceptor antagonists, anticoagulants and proton pump inhibitors. The drugs and ADRs most frequently reported by patients were similar to those reported by doctors, including the organ systems affected. No difference was found between patient and HCP reports in terms of overall seriousness, but some differences were reported between the type of reporter and different categories of seriousness, i.e. patients reported significantly more disability than HCPs (2.3% and 0.4%, respectively) and more life-threatening reactions (5.2% and 2.7%, respectively). The Lareb system accepts only electronic reports from consumers, whereas HCPs can submit online or using a paper form. This may limit the numbers of reports by consumers, especially those who have little or no access to the internet.

In addition to Lareb, a patient-run reporting scheme31 has operated since 2004. Higher numbers of reports are made to this scheme than to Lareb, but the amount and type of information recorded by the two schemes differs, and so the data are not entirely comparable. During the first 10 months of the scheme, 49% of reports related to side effects, of which 6% were severe and 30% were not mentioned on the PIL. A short report was published of a comparison of the reports associated with the use of paroxetine made with Lareb and those reported to a free telephone medicines information service, introduced in 1990, which enabled patients to consult a pharmacist about the correct use of medicines and problems related to their medicine use. 20 Out of 23,625 calls from patients, 120 suspected ADRs were reported for paroxetine compared with 89 of the 7665 suspected ADRs from HCPs. Proportionally fewer reports were made about paroxetine via the telephone service than to Lareb (0.5% vs 1.2%, respectively). However, suspected ADRs were reported sooner using the telephone service than using Lareb (mean 229 days; 95% CI 160 to 298 days). No difference was found between the two reporting systems in terms of new suspected reactions (i.e. those not included in the PIL). Nine new ADRs were identified by both systems. Each reaction was first reported using the telephone system for all nine reactions, with a mean time lag of 273 days (95% CI 89 to 458 days) between the telephone and Lareb reports. Owing to this study being published as a short report, limited methodological data were presented. More recently, a comparative study was conducted using patient reports submitted to the patient-run reporting scheme for antidepressants,32 with HCP reports submitted during the same data collection period as Lareb. This study was published as an abstract, therefore only limited methodological information was available and there was limited interpretation of the results. In total, 258 reports were submitted by patients, most (72%) of whom were female. Of these reports, 217 were associated with side effects. Significant differences were shown between the ADRs reported by patients and those reported by HCPs. For example, the patients reported more sexual problems and weight gain, whereas HCPs reported more skin, muscle and joint complaints.

In the USA, the Adverse Events Reporting System (AERS) is the national database used by the Food and Drug Administration23 to support postmarketing drug surveillance. The MedWatch33 programme is the reporting scheme for patients and HCPs to use to submit reports of serious problems associated with medicines or medical devices. All reports submitted to MedWatch are entered into AERS. The same reporting form is used for both types of reporters and reports can be submitted online or by post, fax or telephone. Reports submitted by patients are acknowledged. Patient reporters are also contacted by MedWatch if further information is needed. The scheme includes reports for serious adverse events, product problems and errors. Serious events are those that led to death, disability, congenital anomaly, hospitalisation or ‘other serious events’ and those that are life-threatening. In 2008, of the 226,647 reports received by MedWatch, 46% were submitted by patients. The proportion of total reports submitted by patients has increased since the mid-1990s. 34 Relatively detailed annual reports were available regarding patient and HCP reporting to this system in the 1990s,34 and these are now available from the AERS website. 23 These data are presented as annual reports, which do not include detailed methodological information.

Overall, considerable variation exists across national schemes in terms of the proportion of total ADR reports submitted by patients, and this is illustrated in Figure 1.

FIGURE 1.

Percentage of total reports submitted by patients by country.

Processing/coding of reports by the Medicines and Healthcare Regulatory Authority

Coding of data by the Medicines and Healthcare Regulatory Authority

Reaction outcome

Information on the outcome of the reaction is also requested in slightly different ways for patient and HCP reports. Table 3 shows the way in which the different options are coded by the MHRA.

Drugs

The MHRA codes the name of the medicines involved in a suspected ADR as they are reported on the Yellow Card (after correction of spelling errors). Thus, the medicine may be named using the generic drug substance, its brand or proprietary name, the combination name (e.g. co-codamol) or the individual constituents of a product (including, in some cases, its excipients).

An internal drug dictionary is used by the MHRA to select the most appropriate medicine or product name. The medicine/drug/product is then classified as either the ‘suspect drug’ or ‘other drug’ based on the information specified on the report. The MHRA does not code suspected drugs into hierarchical therapeutic classes within the report.

Black triangle status

Drugs with ‘black triangle’ status are those that require intensive monitoring by the MHRA to further assess their risk–benefit profile. This occurs, for instance, if the product contains a new active substance or a new formulation or combination of a previously licensed product or if a significant new prescribing indication has been added to the product’s licence. The time period for black triangle status varies. The status is usually reviewed after the product has been on the market for 2 years, although it may remain until the safety of the drug is more fully established. The MHRA allocates a black triangle flag to drugs specified on the Yellow Card report, according to the drug’s status at the time of the report. HCPs are made aware of black triangle drugs through publications such as the British National Formulary (BNF),38 but, currently, there is no system for informing patients.

Partial data entry

All data from online reports are imported into the database. For paper and telephone reports, all data are entered in to the database if the report is classified as ‘serious’ (as described above), but data from other reports are not always entered in full. The minimum information recorded for these other reports is the patient identifier, the suspect drug(s) involved and the coded MedDRA reaction terms. Hence, the data on the database may not always include, for example, the free-text description of the reaction. Such reports are flagged as ‘partially classified’ on the database. Where more complete information is required for assessment of reports that have been partially classified, full data capture is carried out and the partial classification flag is updated.

Narrative field

When free-text descriptions of the reaction are entered by the MHRA in to the database, they are transcribed in a narrative field, generally verbatim. Sometimes, however, if certain elements have been coded in other fields, the narrative may be abridged, for example to remove personal information, dates or information which can be recorded in structured fields elsewhere in the database. The narrative field is important in situations in which the MHRA wishes to undertake detailed assessment of the potential causality and/or impact of a suspected ADR.

Data provision by the Medicines and Healthcare Regulatory Authority

The MHRA provided the Drug Safety Research Unit (DSRU) with anonymised data on all patient and HCP reports received between 1 October 2005 and 30 September 2007. Reports from the pharmaceutical industry were excluded.

The data were provided electronically in the form of several encrypted Microsoft excel 2003 spreadsheets (Microsoft Corporation, Redmond, WA, USA). A case number was provided as a means of identifying unique patients and linking data between the different excel spreadsheets. The data fields available are shown in Box 1.

Characteristics of those experiencing an adverse drug reaction

In total, 26,129 ADR Yellow Card reports were received from the MHRA over the 2-year study period. Of these, 5180 (19.8%) were patient and 20,949 (80.2%) were HCP reports. Significantly more Yellow Card reports were made for female patients, whether reported by the patient or via HCPs (both p < 0.001) (Table 5). The median age of patients, as reported by either patients or HCPs, was similar (p = 0.06).

Method of reporting

The reporting method used was not documented for 22.7% of reports from patients and 29.3% from HCPs. Excluding these reports, the most frequent method used to report an ADR was the paper Yellow Card form for both reporter groups (79.0% of patients and 87.7% of HCPs); the internet was the next most frequent method (17.6% of patients and 12.3% of HCPs), followed by the telephone (3.5% of patients and 0.03% of HCPs). There was a highly significant association between the method of reporting used and reporter type (p < 0.001).

Reactions

Patients reported 20,358 ADRs in total, whereas HCPs reported 44,429 ADRs. Patients reported a significantly higher number of suspected ADRs per Yellow Card report than HCPs {median [interquartile range (IQR)] of 3 (2 to 5) vs 2 (1 to 3), respectively; p < 0.001 (Table 6)}. Almost one-half (45.2%) of HCP Yellow Card reports contained only one ADR compared with 21.6% of patient Yellow Card reports; only 3.3% of HCPs reported over five reactions per report compared with 21.8% of patient reports (p < 0.001 overall).

The most frequent LLT reaction reported by patients was nausea (n = 458; 2.2% of all patient reactions reported), followed by headache (n = 440; 2.2%) and dizziness (n = 334; 1.6%) (Table 7). Nausea (n = 987; 2.2%) and headache (n = 758; 1.7%) were also the two most frequently reported LLTs on the HCP reports, followed by vomiting (n = 647; 1.5%).

Table 8 presents the number of patient and HCP reports that had at least one of each type of reaction grouped according to the System Organ Classification of MedDRA. More patient reports had mention of a nervous system disorder problem (41.5%) than those of another organ system. This was followed by problems categorised as ‘general disorders and administration site conditions’ (39.8%), problems that were also the second most common organ system affected according to the HCP reports (23.1%). The most common category in the HCP reports was skin and subcutaneous tissue disorders (23.2%).

All of the age- and gender-adjusted ORs between patient and HCP reports by SOC were statistically significant except for vascular disorders, infections and infestations, and injury, poisoning and procedural complications (see Table 8). In general, patients tended to report more ADRs in each SOC, apart from cardiac disorders, in which case patients were significantly less likely than a HCP to report a relevant ADR.

The median (IQR) word count (excluding reports with zero word counts) used to describe the suspected reaction was significantly higher for patient reports [45.0 (22.0 to 74.0)] than HCP reports [15.0 (9.0 to 26.0); p < 0.001]. There was a significant difference between the median word count used to describe the reaction in each method of reporting (Table 9).

Seriousness of reaction

Similar percentages of both patient and HCP reports contained at least one reaction term coded as ‘dictionary serious’ by the MHRA (patients 58.3% vs HCPs 58.8%; p = 0.58).

Over one-half (55.5%) of HCP reporters stated that they considered their patient’s reaction serious.

Of the three subtypes of serious ADRs that were comparable between the two types of reporter, HCPs reported a higher proportion of each event than patient reports (caused hospitalisation 18.8% vs 12.9%; life-threatening 11.1% vs 6.2%; caused death 2.6% vs 0.7%; p < 0.001 for each).

Nearly one-half (44.8%) of the patient Yellow Card reports said that the suspected ADR was bad enough to affect everyday activities, whereas 15.4% said it was uncomfortable or a nuisance and 2.6% said it was mild or slightly uncomfortable. Some patients appeared to have completed a HCP version of the Yellow Card or had a follow-up report by their HCP merged with theirs. Owing to the structure of the database these occurrences were difficult to quantify.

Drugs

A higher proportion of patient reports (16.1%) than of HCP reports (9%, p < 0.001) contained more than one suspect drug. The median (IQR) number of suspect drugs reported was 1 (1 to 1) for both reporter types, although they were statistically different, with the HCPs reporting fewer (p < 0.001). The 20 most frequent suspect drugs reported by patients and HCPs are presented in Figures 2 and 3, and substantial differences can be seen.

FIGURE 2.

The 20 most frequent suspect drugs reported by patients (by reporter type). Note: the rank of the drug for HCP reporters is shown to the right of the HCP bars.

FIGURE 3.

The 20 most frequent suspect drugs reported by HCPs (by reporter type). Note: the ordered rank of the drug for patient reporters is shown to the right of the patient bars.

The ATC anatomical classification of suspect drugs on the patient and HCP reports is presented in Table 10. The most frequent category of drug suspected of being linked to an ADR was for the nervous system, for both patient (33.2%) and HCP (26.2%) reports. This was followed by cardiovascular system drugs from patient reports (21.8%) and anti-infectives for systemic use from HCP reports (19.4%). A statistically significant difference in the percentage of type of suspect drug between reporters was shown for drugs of the nervous system; cardiovascular system; systemic hormonal preparations, excluding sex hormones and insulin; antiparasitic products, insecticides and repellents; herbals/complementary medicine (which all had higher proportions in the patient than HCP reports); anti-infectives for systemic use; antineoplastic and immunomodulating agents; blood and blood-forming organs; and, ‘various’ (all of which had higher proportions in the HCP than the patient reports).

Patient Yellow Card reports were also split into those completed by patients themselves and those completed by a representative (parent, carer, etc.) and HCP Yellow Card reports into those reported by doctors, pharmacists, nurses or another health professional (e.g. dentists). Among the patient reports, a higher proportion of patients reported drugs of the cardiovascular (p < 0.001) and genitourinary and sex hormone system (p = 0.003) than representatives who completed the patient report. In contrast, representatives reported a higher proportion of anti-infectives for systemic use (p < 0.001). Among HCP reports, there were significant differences in the proportion of drugs in each ATC anatomical category reported by the four HCP groups, except for antiparasitic products, insecticides and repellents. A higher proportion of nurses reported a suspected ADR for anti-infectives for systemic use (41.8%) or respiratory system drugs (5.6%) than any other HCP. A higher proportion of pharmacists reported a suspected ADR for a drug for the cardiovascular system (17.2%), musculoskeletal system (11.8%), blood and blood-forming organs (10.8%), and systemic hormonal preparations (2.1%). Of all the HCP groups examined, doctors reported a greater proportion of suspected ADRs for drugs of the alimentary tract and metabolic drugs (10.5%) as well as genitourinary system and sex hormone drugs (6.4%).

Time lag between suspect adverse drug reaction and reporting

Patient reporters took a significantly longer median (IQR) time than HCPs to report their reaction to the MHRA [104 (27 to 463) vs 28 (13 to 75) days, respectively; p < 0.001]. However, there was a higher percentage of missing data for the variable ‘time from reaction to report’ among patient reports (61.0% of such reports) than among HCP reports (33.2%).

The median time to report a suspected ADR for reactions occurring in the first and second year of the study period is presented in Table 11, by reaction outcome. The median time to report an ADR among patients who were still recovering or whose reaction was resolving was significantly longer in year 2 than in year 1, regardless of who reported the ADR [patient (p = 0.01) and HCP (p = 0.02) reports]. However, these data should be interpreted with caution as there were large numbers of missing data related to this issue, in both groups of reporter.

Reaction outcome

A significantly higher proportion of HCPs than patient representatives reported fatal outcomes to the reaction (2.6% vs 0.7%; p < 0.001). A significantly higher proportion of HCPs than patient representatives reported that the patient was recovering or that the ADR was resolving (28.4% vs 16.8%; p < 0.001). More patients than HCPs reported that they had not recovered or that the reaction had not resolved (36.4% vs 22.2%; p < 0.001).

Yellow Card reports involving children

One of our objectives was to specifically examine Yellow Card reports involving children. The findings are shown in Appendix 10. The median age and percentage of females was similar between patient and HCP reporters. Patient reports contained more reactions per report and there were differences in the types of reactions reported. For example, in keeping with all patient reports, patient reports for children were more common for reactions in the following SOC reaction groups: nervous system, psychiatric, gastrointestinal, metabolism and nutritional disorders.

Data collection and processing

As described in Chapter 3, data were received from the MHRA regarding details of patient and HCP reports submitted to the YCS during the 2-year period from October 2005 to September 2007. Reports from the pharmaceutical industry were excluded.

Data synthesis

Following the initial data processing described in Chapter 3, three data sets of drug–ADR pairs were compiled: one for patient reports (‘patient only’), one for HCP reports (‘HCP only’) and a combined data set for reports from both groups (‘combined’). For the purposes of the signal generation analysis the drug–ADR pair consisted of the drug name coded at the lowest hierarchical level within the ATC classification system for drug substances and the reaction term coded at the PT level within the hierarchical MedDRA dictionary.

Signals of disproportionate reporting were generated for each drug–ADR pair reported using a modification of the PPR, calculated by the formula:

where:

• A = the number of patients who had drug X and reaction Y

• B = the number of patients who had drug X and did not have reaction Y

• C = the number of patients who did not have drug X, but had reaction Y

• D = the number of patients who did not have drug X and did not have reaction Y.

That is:

For each drug–ADR pair, each patient contributes to only one of the four A, B, C or D cells, even if a report involved more than one suspect drug or reaction term. This maintains independence between the variables used to calculate the PRR, so that variance is not underestimated.

In order to improve statistical robustness of the PRR where few or no drug–ADR pairs were observed in the data set, a shrinkage method was adopted. Thus, the ‘PRR_shrunk’ was calculated using a modification of the PRR formula:48

where ‘observed’ = A and ‘expected’ = [(A + B)C/C + D].

For each PRR_shrunk, 95% CIs were calculated using the formula:

where:

Within each of the three data sets the PRR_shrunk was generated for each drug–ADR pair reported over the 2-year study period. Based on previously published arbitrary thresholds, drug–ADR pairs were flagged as SDRs, if at the end of the 2-year study period all of the following three conditions were fulfilled:46

• The PRR_shrunk was > 2.

• The number of reports for the drug–reaction pair (A) was ≥ 3.

• The lower 95% CI for the PRR_shrunk was ≥ 1.

All PRR calculations were computed within a SQL Server (2000) database.

Quantitative analysis of signals of disproportionate reporting

Causality assessment in a sample of reports

Drug–ADR pairs identified as SDRs in the disproportionality analysis were first categorised according to whether the patient reports appeared to be adding, or taking away, information provided by the HCP reports (Table 12). This categorisation provided a sampling frame for the selection of reports for an assessment of causality.

Within each of the four categories, SDRs were subgrouped according to whether the ADR term was classified as ‘dictionary serious’ by the MHRA, whether the ADR (or similar term) was documented on the SPC and whether the suspect drug was available in a form or brand classified as a black triangle drug by the MHRA. Four SDRs were then randomly selected from the SDRs in each of the 12 subcategories. It is important to note that the categories in Table 12 were not mutually exclusive; if a randomly selected SDR had already been selected for one category it was replaced by another randomly selected suitable SDR.

The sampling frame was used to select SDRs that involved no more than 10 individual patient and/or HCP reports so that a variety of drugs (including OTC products and complementary therapies) and reactions could be assessed for causality. This number was chosen to give a range of reported problems while ensuring a manageable workload for the project.

An assessment of causality was made for each identified report, by two DSRU researchers (a pharmacist and a biomedical scientist). Conflicting assessments were discussed by the two researchers and, where agreement could not be reached, a third member of DSRU staff (a medically-qualified doctor) arbitrated. The assessment involved reviewing all of the data provided by the MHRA. Each ADR was classified into one of four categories (Table 13). The process also enabled:

• the identification of likely duplicates

• the identification of limitations when making the assessment of causality for patient reports

• an assessment of whether important information provided in the narrative description was lost by coding using MedDRA terminologies.

Comparison of patient and health-care professional reports: overview

There were 5180 (19.8% of total) reports supplied by patients and 20,949 (80.2%) by HCPs. Table 14 shows the number of different reaction terms coded (at PT level), suspect drugs (at ATC code level) and drug–ADR pairs in the three data sets. The ‘patient-only’ data set contained less than one-half (approximately 40%) of all of the drug–ADR pairs reported in this study. Figure 4 shows that most drug–ADR pairs were present in either the ‘patient-only’ or the ‘HCP-only’ data set – only 10.6% drug–ADR pairs were common to both data sets. This indicates marked differences between the two data sets in the identification of drug–ADR pairs.

FIGURE 4.

Distribution of drug–reaction pairs across the patient and HCP data sets. ^aTotal: 41,001 pairs.

Comparison of signals of disproportionate reporting generated by patient and health-care professional data sets

Figure 5 shows the number of SDRs generated using the PRR_shrunk method in the ‘patient-only’ and ‘HCP-only’ data sets. A significantly higher proportion of SDRs were generated from drug–ADR pairs contained in the ‘HCP-only’ data set than in the ‘patient-only’ data set (6.7% and 3.9% of pairs, respectively; difference 2.8%, 95% CI 2.4% to 3.2%). Only 136 drug–ADRs pairs (21.0% of those generated by the ‘patient-only’ data set; 7.0% of those from the ‘HCP-only’ data set) were generated as SDRs in both data sets. For these 136 SDRs, 56 (41.2%) were detectable earlier in the ‘patient-only’ data set, 71 (52.2%) were detectable later and for nine (6.6%) there was no difference in the time to reach SDR threshold.

FIGURE 5.

Signals of disproportionate reporting generated from drug–ADR pairs in the patient and HCP data sets.

It would appear, therefore, that patient and HCP reports generate different SDRs, although this may partly be owing to differences in the way in which reactions are described and then subsequently coded at the MedDRA PT level. Further assessment of the 649 SDRs generated by the ‘patient-only’ data set revealed that, in addition to the 136 SDRs also generated by the ‘HCP-only’ data set, another 88 used a reaction term for the suspect drug similar or related to that of some SDRs generated using the ‘HCP-only’ data set. For example, the ‘patient-only’ data set identified amlodipine and ‘heart rate irregular’ as an SDR, whereas the ‘HCP-only’ identified amlodipine and ‘palpitations’ as an SDR. Nevertheless, 425 of the 649 (65.5%) SDRs generated using the ‘patient-only’ data set were quite distinct and so unlikely be generated using the ‘HCP-only’ data set.

Table 15 shows whether the SDRs generated by each data set involved reactions which were classified as ‘dictionary serious’, or involved a black triangle drug. Compared with those generated using the ‘patient-only’ data set, a significantly higher proportion of the SDRs generated using the ‘HCP-only’ data set was classified as involving serious reactions (difference in proportions 19.5%, 95% CI 15.4% to 23.6%) or a black triangle suspect drug (difference in proportions 19.8%, 95% CI 16.6% to 23.0%).

In both data sets the majority of SDRs generated were documented on the SPC, either specifically as described or using a similar or related clinical term for the suspect drug (Table 16). There was no significant difference between the two data sets in the proportion of SDRs which were regarded as not documented (difference in proportions 2.0%, 95% CI –3.7% to 7.7%).

Table 17 shows the top five most common system organ classes involved in SDRs generated by each data set and the top five anatomical drug classes involved. The same five system organ classes were involved when using each data set although their ranking differed (Table 17). Four of the five top anatomical drug classes were common to both data sets, although again the ranking differed (Table 18).

Table 19 shows the top 10 individual suspect drugs involved in the SDRs generated by each data set and Table 20 the 10 most common preferred terms for the SDRs generated. The drugs and preferred terms for SDRs generated by the two data sets tended to be different, with none of the drugs and only four of the preferred terms being the same. The drugs involved in SDRs generated by the ‘HCP-only’ data set tended to be for recently introduced (black triangle) drugs, whereas the drugs involved in the SDRs generated by the ‘patient-only’ data set were for older, more established drugs.

Analysis of signals of disproportionate reporting generated after the combination of data from patients and health-care professional data sets

Table 21 presents the number of drug–ADR pairs generated as SDRs before and after combining the patient and HCP reports and the number of drug–ADRs pairs no longer generated as SDRs in the ‘combined’ data set.

The majority of SDRs generated in the ‘patient-only’ and ‘HCP-only’ data sets remained SDRs after combining both sets of reports. However, 278 of the drug–ADR pairs generated as SDRs (approximately 11% of all SDRs generated by the ‘patient-only and ‘HCP-only’ data sets independently) were no longer generated as SDRs in the ‘combined’ data set, including SDRs involving drugs flagged as ‘black triangle’, potentially serious reactions or reactions that were not documented on the SPC of the suspect drug. Three of the drug–ADR pairs generated as SDRs by the ‘HCP-only’ data set that were no longer generated as SDRs in the ‘combined’ data set were potentially serious, involved a drug flagged as ‘black triangle’ and were not documented as a problem on the SPC (levetiracetam/drug interaction, pregabalin/overdose and rimonabant/blood pressure increased).

Of the 1753 SDRs generated in the ‘HCP-only’ data set that remained SDRs in the combined data set, 139 (7.9%) were detectable later in the combined data set, 247 (14.1%) were detectable earlier and in the remaining 1367 (77.0%) the month of detection was unchanged. This should be interpreted with caution, however, as this assumes that both patient and HCP reporting started at the same ‘day 0’, which is not true in practice (HCP reporting to the YCS commenced in 1964).

In practice, disproportionality analysis is not relied upon in signal detection for black triangle drugs. Instead, serious reactions to black triangle drugs are scrutinised on a case-by-case basis by the regulatory authorities. ‘Loss’ of SDRs, however, may be of concern for established drugs, for which disproportionality analysis is used to prioritise serious reactions for clinical assessment. Five of the SDRs generated when using the ‘HCP-only’ data set were ‘lost’ after combination of patient and HCP reports, involving established drugs, potentially serious reactions and reactions that were not documented on the product’s SPC (anastrazole/depression, citalopram/hypoglycaemia, simvastatin/drug interaction, trimethoprim/chest discomfort and trazadone/myalgia). Four of the SDRs generated by the ‘patient-only’ data set involved established drugs, potentially serious reactions and reactions that were not documented on the product’s SPC, which were ‘lost’ when the patient reports were combined with the HCP reports (citalopram/aggression, doxazosin/muscular weakness, perindopril/hypertension and prednisolone/face swelling).

The addition of patient reports to HCP reports generated an extra 929 SDRs for drug–ADR pairs that either had not been generated as SDRs when the ‘HCP-only’ data set was used (705 SDRs) or that were not present in ‘HCP-only’ data set (i.e. 224 SDRs that were only seen in the ‘patient-only’ data set) (Figure 6). Of these 929 additional SDRs, 299 (32.2%) were classified as being potentially serious, 139 (15.0%) involved black triangle drugs and 222 (23.9%) involved reactions that were not documented on the SPC.

FIGURE 6.

Additional SDRs generated after combining the patient and the HCP reports.

Of the 929 additional SDRs generated, 508 would not have been generated by disproportionality analysis in either the ‘patient-only’ or the ‘HCP-only’ data sets alone. Of these 508 extra SDRs, 185 (36.4%) were classified as being potentially serious, 95 (18.7%) involved drugs flagged as ‘black triangle’ and 137 (26.9%) involved reactions which were not documented as problems on the SPC. Ten of these additional SDRs were potentially serious, involved a drug flagged as ‘black triangle’ and were not documented on the product’s SPC (see Appendix 11), whereas 37 of the additional SDRs were potentially serious reactions involving established drugs that were not documented on the product’s SPC (see Appendix 12). Some of these SDRs were related to conditions that were prescribing indications for the suspect drug, for example diazepam/muscle spasms, chloramphenicol/eyelid oedema and donepezil/confusional state.

Causality assessment in a sample of reports

Overall, 41 drug–ADR pairs were randomly selected from the sampling frame for causality assessment, involving 199 patient reports and 184 HCP reports, to show the findings from the causality assessments performed within each category of drug–ADR pairs of interest. These have not been tested statistically owing to the small numbers within each category. The full breakdown of causality assessments is included in Appendix 13.

In Category 1, 11 drug–ADR pairs present in both the ‘patient-only’ and the ‘HCP-only’ data sets but identified as SDRs only in the ‘patient-only’ data set were selected from the sampling frame. The distribution of causality assessments performed on the 56 patient and the 43 HCP reports involving these drug–ADR pairs was similar for both reporter groups (Figure 7). Most reports were assessed as ‘possible’ with a relatively small number assessed as ‘probable’. Slightly more patient reports than HCP reports could not be assessed.

FIGURE 7.

Category 1: drug–ADR pairs present in both the ‘patient-only’ and the ‘HCP-only’ data sets; SDRs generated only by the ‘patient-only’ data set (11 pairs selected).

In Category 2, 12 drug–ADR pairs identified as SDRs in the ‘patient-only’ data set but not present in the ‘HCP-only’ data set were selected. Just under half of the 73 patient reports involved in these drug–ADR pairs were assessed as having a ‘possible’ causal relationship, with a minority assessed as ‘probable’. Compared with Category 1, however, a slightly smaller proportion was assessed as ‘probable’ or ‘possible’ and a higher proportion was assessed as ‘unlikely’ (Figure 8). In addition, almost a quarter of reports in Category 2 were regarded as ‘not assessable’.

FIGURE 8.

Category 2: drug–ADR pairs only present in the ‘patient-only’ data set; SDRs generated by patient, but not the HCP reports (12 pairs selected).

In Category 3, 12 drug–ADR pairs were selected that were identified as SDRs only after combining the patient and HCP reports. The distribution of causality assessments performed on the 52 patient and the 46 HCP reports involving these drug–ADR pairs was again similar for both groups (Figure 9). Compared with the patient reports, a higher proportion of the HCP reports was assessed as ‘probable’ or ‘unlikely’ and a lower proportion was assessed as ‘possible’.

FIGURE 9.

Category 3: drug–ADR pairs present in both the ‘patient-only’ and the ‘HCP-only’ data sets, but SDRs generated only after combining both sets of reports (12 pairs selected).

In Category 4, 12 drug–ADR pairs identified as SDRs by the ‘HCP-only’ data set were ‘lost’ when patient and HCP reports were combined and were selected for causality assessment. The assessment was performed on 84 HCP reports involving these drug–ADR pairs. Most reports were assessed as either ‘possible’ or ‘probable’, with few assessed as ‘unlikely’ (Figure 10).

A number of limitations to causality assessment were observed. These have not been systematically quantified and compared for each reporter group, but were observed when using both patient and HCP reports.

FIGURE 10.

Category 4: drug–ADR pairs generated as SDRs in the ‘HCP-only’ data set, but no longer regarded as SDRs when patient reports were combined with the HCP reports (12 pairs selected).

1. Missing information In some reports, no narrative or relevant dates were given, preventing an assessment of temporality. In some cases only a list of symptoms/events was reported. Information on past medical history was generally sparse, making it difficult to assess whether there were other reasons for the reported event(s).

2. Common or coincidental events Several events in the sample of pairs selected involved reaction terms that are common complaints in everyday life, making it difficult to determine whether these ‘reactions’ were in fact causal or coincidental, for example hair loss, cough, hypertension, weight gain, headache, diarrhoea or insomnia.

3. Possible confounding by indication This limited the ability to determine causality for a number of drug–ADR pairs, for example pregabalin/convulsion, fluoxetine/anxiety, iofendylate/arachnoiditis, olanzapine/agitation, tiotropium/dyspnoea, atomoxetine/depression and anti-malarial/vomiting or diarrhoea or pyrexia.

4. Long time to onset of reaction It is difficult to assess causality in events or symptoms that develop or are recognised by the patient many months or years after starting the medication. There may be many other unknown or unreported events/factors that may have contributed over this time period. This was observed in assessing reports of ‘loss of confidence’ with cyproterone/oestrogen, many of whom also experience depression.

5. The reporting of vague symptoms/less well defined terms These were difficult to assess, for example lack of confidence, anorexia, asthenia.

6. In some cases a reaction was reported which was specific to the drug’s formulation or brand, or the medication was suspected of being counterfeit These reactions were regarded as ‘not assessable’ for this project.

7. Multiple reaction terms/multiple suspect drugs In some reports a long list of reaction terms and/or suspect drugs was reported. It was not always possible to determine whether the information provided regarding a possible temporal relationship or outcome related to the specific reaction term or drug being assessed for causality. Patient reports tended to contain a larger number of coded reaction terms and were more likely to contain more than one suspect drug than HCP reports.

8. Ambiguity of coded terms For each of the 383 reports sampled, the drug–ADR pair of interest was cross-checked with the free-text description of the reaction provided by the reporter. In the majority of cases (314; 81.9%) the MedDRA term used to code the reaction appeared to be appropriate; in one of these cases an appropriate coding term had been used for the drug–ADR pair risperidone/musculoskeletal stiffness, but from the narrative the intended meaning of the reaction may have been an extrapyramidal side effect. In the remaining cases there was either no narrative (45; 11.7%) or the narrative did not contain any reference to the reaction of interest (24; 6.3%). Overall, however, there was little evidence to suggest that the intended meaning of the reporter’s description was lost in coding.

Finally, of the 383 forms evaluated, 18 were thought to be duplicate reports relating to the same nine patients: seven of these were detected twice in the ‘patient-only’ data set, one was detected twice in the ‘HCP-only’ data set and one was detected in both data sets. This suggests that the MHRA system to identify and merge reports about the same reaction in the same patient from different sources is not 100% effective.

Case selection

Following extensive discussions with the advisory group, and within the project team, a purposive sample was taken from a range of different categories of Yellow Card reports. The advisory group suggested sampling by reaction type as well as by drug and suggested we cover a wide range of different drugs and different types of reaction. The details of the approach taken to sampling are provided in Appendix 14. The reports were divided as follows:

• drugs most commonly reported by patients

• black triangle drugs

• drugs purchased OTC

• complementary therapies.

The original plan was to examine around 300 patient reports and 300 HCP reports for similar types of drug–ADR pairs. There were, however, marked differences in reporting between patients and HCPs (as described in Chapter 5) and this meant that it was not possible to obtain the same number of reports for each group. In addition, there were no drugs reported by HCPs that were purchased OTC and virtually no complementary therapy reports; we examined only the patient reports for these two categories.

The details of the final sample of reports are shown in Appendix 14, and are summarised in Table 22. In total, we selected 270 patient reports and 179 HCP reports. This is fewer than originally planned, but following discussion with the advisory group we judged that there were sufficient numbers of cases for the qualitative analysis.

Analysis

Description of the sample

In total, 449 reports were analysed, with 270 of these coming from patients and 179 from HCPs.

Descriptive statistics of the reports by patient (n = 230) and HCP (n = 179) reporters for the categories of ‘most commonly reported drugs by patients’ and black triangle drugs are presented in Table 23. Combining the data for these two categories, it can be seen from the table that the main differences between the patient and HCP reports were:

• Mean age [standard deviation (SD)] in patient reports was slightly lower [44.2 (16.1) years vs 47.8 (18.5) years].

• The number of drugs that the patient was reported as taking was lower for patient reports [median 2 (IQR 1 to 3) compared with 2 (IQR 1 to 4) for HCP reports].

• Median number of distinct reactions reported was higher for patient reports [5 (IQR 3 to 8) compared with 3 (IQR 2 to 5) for HCP reports].

• Median word count was greater in patient reports [58 words (IQR 29 to 121)] than in the reports from HCPs [16 words (IQR 9 to 31)].

• Just over half (53.0%) of patient reports made using the paper form compared with 81.0% of HCP reports.

There was no statistically significant difference between patient reports and HCP reports in the proportion relating to women.

The breakdown of HCP reporters was:

• doctors (65%):

  1. – GP 51%

  2. – hospital doctor 8%

  3. – physician 6%

• pharmacists (13%):

  1. – community pharmacist 6%

  2. – hospital pharmacist 5%

  3. – pharmacist – background not specified 2%

• other health professionals (22%):

  1. – nurses (11%):

     1. – hospital nurse 8%

     2. – nurse – background not specified 3%

  2. – other health professional – background not specified 7%

  3. – hospital health professional 3%

  4. – dentist 1%.

Of the patient reports, 216 (94%) were made by the patient themselves, 10 (4%) came from carers and four (2%) were from a parent.

Qualitative analysis

Symptom description

Both patients (93%) and health professionals (78%) described symptoms of ADRs. However, patients’ reports tended to be much more detailed and extensive, as reflected in the mean elaboration score (patients 1.68, health professionals 1.07) and the word count. Comparing the following two reports, one by a health professional and the other by a patient, relating to the same suspect drug but involving different patients, it can be seen that the HCP described the symptoms and what the patient had been advised to do to: withdraw from the medicine. In contrast, the patient not only listed the symptoms experienced, but also described their intensity. She refers to ‘unbearable anxiety’, describing the current stage of dose reduction as ‘the worse stage so far’, and the medicine as ‘horrendous’. The symptom of headache is highlighted by repeating it and putting it in block capitals:

Dizziness, blurred vision, light headedness and electric shock sensations following withdrawal from Seroxat. These sensations are abolished by taking half a Seroxat tablet, but then recur after I switched him to Seroxat suspension 20 mg/10 ml and advised him to reduce dose in very small steps from 0.7 ml daily. Not seen since.

GP, male, 44 years, paroxetine, paper

Since beginning to reduce this medicine, I have had terrible withdrawal symptoms, these have included: sudden changes in emotion and mood, crying, insomnia, excessive anxiety and agitation, sweating and palpitations. There have also been bouts of stomach upsets, nausea, dizziness and headaches. Since reaching an amount of 5 mg I have had to use the liquid version with a syringe and make reductions of 1 mg per month, this has been the worst stage so far and I have been prescribed medication to alleviate the unbearable anxiety that this is causing! Still 2 mg to go before I am off this horrendous medicine! HEADACHE HEADACHE.

Patient, female, 46 years, paroxetine, paper

Similarly, comparing two different reports for levetiracetam, the health professional report was very short and the patient report more elaborate and descriptive:

Movements feel dull and heavy, tiredness, nausea and shaky.

Community pharmacist, female, 71 years, levetiracetam, reporting method not specified

I couldn’t stay awake, sick, staggering, loss of balance, eyesight got worse. When I increased the evening dose, my knuckles got swollen and sore, then I had 5 weeks of sleeplessness nights with the agony of carpal tunnel syndrome in my right hand. A steroid injection helped, but I still have a sore hand. Keppra – pain in my feet, ankles, shins, knee, hips, thighs, fingers, hands and arms, I could hardly walk, also couldn’t sleep at night which made me tired during the day. Before taking the medication for epilepsy, I had no pain and could move freely. Since medication I have been in constant pain and hobble about like a crippled old woman. I can’t even sit cross legged any more.

Patient, female, 58 years, levetiracetam, paper

The patient quoted below gives an elaborate description of the history and symptoms of her hypoglycaemic reaction to insulin glargine, emphasising the unpredictability, rapidity of onset and associated dangers of the reaction:

Reaction started almost immediately. The analogue insulin made me have hypoglycaemic reactions that came upon me with little or absolutely no warning signs at all. Not only did this happen, but the window of opportunity of recognising a symptom to being too stroppy to do or have something done for me was less than 5 minutes – on Porcine I have at least 30 to 40 minutes. On top of this a hypoglycaemic reaction at night did not wake me up. Very dangerous and could lead to severe damage or even death if no one found me. This happened on several occasions, but fortunately others were with me. It did involve me going to A&E twice and having a doctor called out to house twice. Other times husband has given me 2 or 3 glucagons! I have after considerable difficulty, persuaded my Diabetic Consultant to put me back on to some porcine insulin – for the present only the neutral short acting, but as it is going so well, I am hoping to revert to the porcine long acting as well. It took me 5 months to get this to happen and that should not be the case. Stopped as soon as I switched back to porcine neutral insulin.

Patient, female, 54 years, insulin glargine, internet

In contrast, a GP report of a different patient’s reaction to insulin glargine is brief and to the point, but gives no history:

Erratic diabetic control.

GP, male, 36 years, insulin glargine, paper

Other health professional reports were equally brief:

Raised CK 350.

GP, simvastatin, paper

Major withdrawal, panic attacks tremors, eye pain, neck pain, leg pain, found it difficult to walk. Eye problems, sees floaters like looking through net curtain. Life threatening because depressed and anxious to end life.

Community pharmacist, male, 44 years, paroxetine, paper

Very few patients made short, unelaborated reports that were similar to those from HCPs, for example:

Severe depression with suicidal thoughts.

Patient, female, 30 years, Dianette, internet

On the other hand, there were very occasional elaborated reports from health professionals, such as the following, which describes the effect on the patient in great detail:

The patient was complaining that on a bad day she had palpitations, dyspnoea, nightmares, difficulty speaking, disorientation, difficulty walking with poor balance, a feeling as though she is drunk and feeling of cold alternating with episodes of sweating. The current symptoms are of difficulty focusing which varies from one eye to another on a daily basis, electric shock type symptoms shooting down her limbs periodically, poor balance, poor memory and word finding and severe insomnia. In addition, she has recently complained of heartburn and of extreme frequency of micturition. The patient tried a course of St John’s wort. This was to try and control the insomnia, electric shock and tremor symptoms. Approximately 2 weeks after starting this she had disturbing and visual hallucinations. She was agitated and felt driven to throwing herself off a roof. She discontinued the St John’s wort and did not seek medical attention at the time. Her mother observed her closely and after approximately 48 hours the symptoms resolved. Currently the patient continues with complaints of hallucinations, slurred speech and difficulty in word finding, intermittent blurred vision and intermittent dyspnoea which sounds like chest muscle spasms. She had a severe episode of abdominal discomfort last month which resulted in an attendance at the acute medical unit.

GP, female, 37 years, paroxetine, paper

Some patients stated why they were originally prescribed the offending medicine. The health professionals tended not to do this in the free-text, comments, but past medical history, which may include these details, is also recorded in coded fields as a MedDRA term in response to relevant questions on patient and HCP Yellow Cards:

I was first prescribed Seroxat as I was feeling low in mood …

Patient, female, 37 years, paroxetine, paper

I was prescribed this medication for mild sleeplessness. I became addicted to it and after 18 months of severely debilitating symptoms, the principal symptom being persistent suicidal thoughts.

Patient, female, 31 years, paroxetine, paper

Patients also described what their relatives and health professionals said about the adverse effect as if to confirm that it was real:

My daughter, my wife and a sister in a hospital said I looked dreadful in the face …

Patient, male, 61 years, simvastatin, paper

In addition to providing the expected details about the ADR, some patients also gave reasons for making the report, including a diagnosis or to find other people with the same reaction:

I am desperate to discuss my condition with someone who is prepared to listen. I don’t know where else to turn and a friend gave me the enclosed Yellow Card and suggested that I might get someone to diagnose and tell me which of the many diagnosis [sic] I am actually suffering from. I’m told I do not have the right to get advice from another doctor but I am sinking into depression now and feel there must be someone somewhere who can help me and not just leave me to it.

Patient, male, 61 years, simvastatin, paper

Electric shock sensation

Twenty-three per cent of patient reports and 14% of HCP reports about citalopram, paroxetine and venlafaxine included descriptions of ‘electric shock sensations’ or similar reactions. Patient reports of this particular symptom exemplify the tendency among patients to stress the severity of the reaction, which was described above. Compared with the professional reports, patient reports were particularly vivid, comparing this reaction to a range of other extreme experiences and stressing the intensity of the symptoms. These provide an important example of how, if such a serious and unpleasant adverse effect were to occur from another drug in the future, patients’ reports could contribute to pharmacovigilance and to understanding of the effects of an ADR:

Shocks these are like dropping out of an aircraft as a diver for the first time or a speeding car hitting a hill at speed then flying through the air. Your breath is taken away …

Patient, male 51 years, paroxetine, internet

… electrical zaps …

Patient, female, 32 years, paroxetine, internet

… ‘fireworks’ exploding in my head …

Patient, female, 50 years, venlafaxine, paper

… electric shock-like sensations also called brain shivers.

Patient, male, 27 years, venlafaxine, internet

One patient went to great length in her description of the adverse reaction:

First symptoms constant headache, extreme agitation, pressure inside head followed by sudden explosion inside head. Sensation of chemical being forcibly released from a particular area in brain, which then felt like a cold trickle down neck, across chest down spine. Neck went rigid and painful, cold shivers then feverishness, chest felt heavy, heart rate increased [although felt very calm and was not hyperventilating (this was not a panic attack)], legs then began shaking and convulsing uncontrollably for half an hour, accompanied by regular release of chemical in head symptoms subsided after an hour, but then began to reoccur approximately every 6 days in the same pattern, but with increasing severity. Arms and legs would convulse uncontrollably, arms and legs would be numb and pins and needles and loss of feeling, neck would go rigid and jaw clench, pressure inside head would rise and fall with each explosion. Symptoms would last for up to 3 hours each time. Headache would return once symptoms had subsided each time. After 3 weeks of this, I reduced the dosage of citalopram back to 10 mg, concerned that it was causing these symptoms

Patient, female, 34 years, citalopram, paper

A number of health professionals listed electric shocks as a reaction, but only one HCP in our sample described it in any more depth, quoting the patient’s own description:

… reported that her brain was fizzing like being connected to the national grid.

GP, female, 35 years, venlafaxine, paper

Impact on patients’ lives

Forty-seven per cent of patient reports discussed the impact of the reaction on their lives compared with only 12% of health-professional reports. Three types of impact were discussed: impact on relationships and social life, occupational impact and emotional impact.

Causality

Patients commonly provided information that would be useful when assessing causal links between the suspect drug and the reaction. This will now be illustrated by discussing causality under a number of headings that have been derived from a series of questions designed for assessing the likelihood of an adverse reaction being related to the use of a medicine. 49

Involvement of health professionals

Many of the patient reports commented favourably on the advice and care they had received from HCPs, particularly that received from doctors. A small minority, however, reported that their doctor had not understood them, not known about the existence of the side effect, ignored them or told them that they were wasting their doctor’s time:

… After 10 or more years on these pills I can feel my stomach isn’t right but have had to put up with it. I feel I’ve been lied to over the years even by the doctor who did my endoscopy who didn’t want to do it, I had to persevere to try and get to the bottom of this. Still haven’t. These drugs are evil.

Patient, male, 40 years, paroxetine, internet

A few of the patients who suffered from depression following taking Dianette complained that their doctor did not know that Dianette caused depression:

… My doctor suggested I take anti depressives just until my moods had become more stable but I refused. He never mentioned that my depression and anxiety could be linked to Dianette…

Patient, female, 29 years, Dianette, internet

Other patients described how they felt ignored by their doctor:

… I pointed out to doctor that leaflet accompanying medication stated there could be muscle problems. If experienced patient should discuss with doctor. I pointed out problem to my doctor and pain I was having all of which he chose to ignore. I’m told I do not have the right to get advice from another doctor but I am sinking into depression now and feel there must be someone somewhere who can help me and not just leave me to it.

Patient, female, 71 years, simvastatin, paper

One patient described a particularly unfortunate experience with their doctor over the adverse effects they had experienced from pregabalin:

Side effects started straight away, was violently sick, fit and blacked out. Dizzy head, tendency to fall over, itching, blurry eyes, swelling of the stomach, food tastes awful, mouth dry on waking and I feel like there is a heavy weight on my chest and I have to sit up to breathe. I can’t go about doing things and my husband has Parkinson’s disease. Five other women who suffer as I do and they had the same, and stopped them. I tell you this in confidence, my doctor was very rude to me, I had a couple of drugs that didn’t work, he prescribed the pregabalin. I went back to see him and he assured me of wasting NHS money and told me if I didn’t take them he would strike me off his medical list and I wouldn’t be seen again. He also wouldn’t believe me that I couldn’t use my arm, he grabbed hold of my finger, tugged my arm up into the air, let it drop down and it slammed down on to his desk, no apology. I was told I should have reported him, but I couldn’t have coped with the outcome it may have caused. I don’t see him anymore, I go to another one who is very nice.

Patient, female, 78 years, pregabalin, paper

Influence of reporting method

It was thought that the free text in patient telephone reports might differ from those reported on paper or via the internet because the report was transcribed during the patients’ telephone call by a staff member at the MHRA, who may be selective in what they chose to record. Although a few telephone reports were short and more like health professional reports, a number of them were more elaborate, mentioning the impact of the reaction on the patients’ life and included what are assumed to be direct quotes from patients, as illustrated below:

Rigidity – not complete but a degree. Trembling feverishness – didn’t take temperature, loss of sleep, panic attacks, loss of concentration, depression worse, loss of sense of taste (partial), facial movements, blurred vision – more than usual, dizziness, agitation, feeling faint, weakness, confusion. Higher level of anxiety than usual – felt chemical in origin. Loss of initiative, being ‘full of words’ in minds ear, coupled with being ‘keyed up’ – not a good combination. Taken 2 months to die down since then.

Patient, male, 51 years, citalopram, telephone

One telephone report was written in the first person:

Twenty minutes in, as though I had been drinking – couldn’t stop laughing. I do not drink. Palpitations, very drowsy had to be prodded awake, slept till 11 pm. Arms ‘heavy’, nauseous, dreadful, slurry. Next day, violently sick. Went to GP next day, didn’t give anything as has had a mild antihistamine reaction. Tearful and aggressive/argumentative. Frightened. Feeling ill. Trying to disguise slurred speech. Headache.

Patient, female, 46 years, certrizine, telephone

Study population

The study population was all patients reporting an ADR to the MHRA via the YCS between March 2008 and December 2008.

Questionnaire development

A postal questionnaire was developed to explore patient reporters’ views and experiences of making a Yellow Card report. Members of the project team worked on several iterations of the questionnaire to ensure that it covered all key areas regarding reporters’ views and experiences, and that it had face validity. The questionnaire (see Appendix 16) asked about how patient reporters learned about the YCS, how often they had made reports, details of their most recent Yellow Card report, reasons for submitting a report, their experience of reporting the ADR and their sociodemographic characteristics. Most questions used closed formats, including some scaled responses. However, open questions were included to obtain views on problems with reporting, suggested improvements to the YCS, expectations and reasons for future use of the scheme. In addition, the patient reporters were asked if they would be willing to talk to a researcher about their report in a more detailed telephone interview (see Chapter 8 for further details).

For piloting, the questionnaire was distributed to a convenience sample of adults of varying ages and educational levels to assess ease of completion and identify areas for amendment. We did not assess test–retest reliability. A final version of the questionnaire was produced after taking account of feedback.

Questionnaire administration

To maintain confidentiality, the MHRA distributed the questionnaires on behalf of the research team with a covering letter (see Appendix 17). To minimise recall bias, the questionnaires were sent out within a week of a patient’s report being received by the MHRA. Only MHRA staff knew the identity of the reporters, and they assigned a unique identifier to each questionnaire holding the code separately. Completed questionnaires were returned directly to the researchers, who sent MHRA staff regular updates of identifiers for returned questionnaires. This enabled the MHRA to identify non-responders and send reminders (see Appendix 18), while maintaining anonymity. A single reminder was sent 3 weeks after the original mailing.

Data management and analysis

Data from the questionnaires were entered into spss and validated by checking a 10% systematic sample against the original questionnaires. The medicine that the respondent suspected as causing their ADR was coded according to the ATC classification system. 39 Categorical responses were described using frequencies and percentages. The distribution of respondent age was skewed and therefore was described using the median and IQR, and compared between groups using the Kruskal–Wallis test. Pearson’s chi-squared test was used to examine the associations between the method of reporting (online, postal, telephone) and the following responses to the questionnaire: where they learned about the YCS, the ATC anatomical classification of the suspect drug, how soon they reported ADR, ease of reporting using the YCS, whose idea it was to report the ADR, and whether they expected (and would have liked) feedback on their report. Associations between two categorical variables were examined using Pearson’s chi-squared test. To minimise the chance of a type 1 error arising from multiple comparisons, a p-value of ≤ 0.01 was used to denote statistical significance throughout.

Each free-text question was analysed separately, using a thematic approach. All comments were read individually and sorted thematically using an iterative approach. For questions with large numbers of responses, two researchers initially independently read 100 responses to identify themes. Final themes were agreed and all responses were coded independently. Any discrepancies were discussed and agreed.

Respondent characteristics

The respondents’ median (IQR) age was 56.5 years (43.0 to 67.0 years], 910 (66.8%) were female and 1274 (93.5%) were of white ethnicity (Table 25). Over two-thirds of respondents had been educated beyond 18 years. The Yellow Card report was submitted by post for 814 (59.8%), online for 447 (32.8%) and by telephone for 86 (6.3%) respondents. Respondents who made an online report were significantly younger (p < 0.001) and had a higher level of education than those using another method of reporting (p < 0.001). Only 105 (7.7%) respondents had sent in more than one Yellow Card report, and 133 respondents (9.8%) had submitted a Yellow Card report on someone else’s behalf.

How respondents learnt about the Yellow Card Scheme

Almost one-half of the respondents learnt about the YCS from a pharmacy (n = 667; 49.0%), followed by their GP (n = 220; 16.2%). Of those who learnt about the YCS from other sources not listed as options in the questionnaire (n = 252; 18.5%), the most frequent response was that they had a HCP background/professional knowledge (n = 88; 35.2%). A higher proportion of respondents reporting by post learnt about the YCS from a pharmacy or GP surgery than either of those using the other two methods (both p < 0.001; Table 26). One-hundred and thirteen (25.3%) patients reporting online found out about the scheme from an internet search, compared with only 24 (2.9%) and 7 (8.1%) of those reporting by post and telephone, respectively.

Information regarding respondents’ Yellow Card reports

Table 27 presents the ATC anatomical classifications of the suspect drugs reported by the method of reporting used. Overall, the biggest percentage of reports was for a drug for the nervous system and this did not differ by method of reporting. However, there were some differences. Over one-quarter (n = 213) of respondents reporting by post reported a drug related to the cardiovascular system compared with 22.1% (n = 19) of those reporting by telephone and 17.2% (n = 77) of those reporting online (p = 0.001). A higher proportion of online users (n = 34; 7.6%) reported adverse events from drugs for the genitourinary system and sex hormones than users of other reporting methods [post n = 31 (3.8%) and telephone n = 2 (2.3%)] (p = 0.006).

Figure 11 shows that those reporting by telephone did so relatively sooner than those reporting by either of the other two methods, whereas those who reported by post took the longest time to report the event (p = 0.01). In total, 1304 (95.7%) respondents were at least fairly sure that the side effect was due to the medicine, with no significant difference between methods of reporting (p = 0.22).

FIGURE 11.

Time the report was made after the side effect was first noticed by method of reporting. Note: Pearson’s chi-squared test showed a significant association between the time the report was made after the ADR and method of reporting used (p = 0.01).

Respondents’ views on submitting Yellow Card reports

Most respondents – 1274 (93.6%) – thought that the report was fairly or very easy to complete (Table 28), but in free-text comments 216 (15.9%) noted difficulties that they had experienced. A higher proportion of telephone reporters (6.8%) felt that it was ‘not very easy’ to report their ADR compared with those using other methods of reporting [online (3.4%); post (2.5%)] (p < 0.001).

Suggestions for improvements to the form or processes were made by 307 (22.5%) respondents. These included more space (n = 31), simpler questions or language (n = 18) and larger print (n = 7). Respondents who reported difficulties, included 39 who had general difficulties owing to problems with writing/poor sight, anxiety or confusion, 15 who indicated uncertainties about what to report and 10 who felt that some questions were missing or lacked relevant options. Several mentioned the inability to use imperial units for height and weight online.

Not much room on the form, not very well set out. I found it difficult the second time to fill in the form because I was suffering from severe anxiety which was the problem.

Female, 60 years, further educational qualification

There were 39 respondents who raised specific issues about internet reporting, many of whom gave up and reported via telephone or paper:

The proforma asked for the major symptoms. When I wrote paralysis, a long list of types of paralysis appeared – it needed medical knowledge to complete. I could not move past it. I gave up and sent a written report direct to your enquiry desk.

Male, 54 years, graduate

Seven raised issues relating to telephone reporting:

I first telephoned 08081 003 352 to be told by recorded message calls could only be taken between 10–2 Mon–Fri. The hours need to be extended. The Yellow Card says you can report suspected side effects by filling in Yellow form or by calling which I did first, but the person only asked for my name and address in order to send the Yellow Card and did not ask would I like to give details over the phone.

Female, 71 years, graduate

A total of 143 respondents suggested that greater publicity of the scheme, wider availability and accessibility of the reporting forms and encouragement to report were required. Twenty-three respondents indicated that health professionals lacked awareness of the patient reporting scheme, but should promote patient reporting. A substantial number suggested posters and leaflets should be available in GP surgeries, hospitals and pharmacies. Other suggestions included forms in post offices, libraries and local government offices or their issue with medicines:

My GP and receptionist did not know how to access a Yellow form! The hospital told me that a GP had to do it.

Female, 65 years, further educational qualification

A description of the scheme with the MHRA website should be (by law if necessary) included in the manufacturers’ information contained in the medicine packages.

Male, 69 years, graduate

Why the respondent made a Yellow Card report

Most respondents (n = 1098; 80.6%) said that making the report had been their own idea, with a significantly higher proportion of the online completers (86.4%; n = 386) stating this than postal (78.1%; n = 636) or telephone (77.9%; n = 67) reporters (p = 0.001). Among those reporting by telephone, 19.8% (n = 17) said that it was their pharmacist’s idea to report the event, compared with 17.4% (n = 142) of postal reporters and 6.9% (n = 31) of online reporters (p < 0.001). Only 43 (3.2%) respondents were discouraged by someone from making a report. Of these, most were discouraged by their GP (n = 26; 60.5%). Fifty-six respondents (4.1%) stated that a HCP had refused to make a report on their behalf.

Respondent experiences and expectations of reporting

Some respondents (n = 104; 7.6%) had received help to make the report, with most assistance coming from a family member or friend (n = 56) or a pharmacist (n = 26). Almost one-third (n = 448; 32.9%) of respondents expected feedback from the MHRA and 828 (60.8%) said that they would have liked feedback. Of those reporting their suspected ADR by telephone, 42.7% (n = 35) expected feedback compared with 36.0% (n = 290) of postal reporters (p = 0.001) and 26.2% (n = 117) of online reporters. A higher proportion of telephone reporters (74.3%; n = 55) would have liked feedback than other reporters [postal 67.3% (n = 495) and online 63.0% (n = 269)], although this difference was not statistically significant (p = 0.05).

Responses to open questions indicated that 134 respondents expected an acknowledgement of their report and 112 expected to receive information about the reaction or the medicine about which they had reported. Many of those who desired feedback (323/707) wanted specific information about the frequency with which other similar reports had been received, how common the effect was or whether it was a well-known problem. However, 149 wanted to know whether any investigation or action would take place as a result of their report, including informing manufacturers or health professionals:

Would like to know how often this particular problem has been reported in the past and how well known the side effect is within the medical profession.

Male, 53 years, postgraduate

Tell me what you are doing with the information I give you and any action you plan to take to prevent others suffering as I did.

Male, 33 years, graduate

A minority (n = 24) had expected to be approached for further information or provided with advice on what action to take (n = 25); more (n = 84) wanted to receive such advice. There were 46 respondents who wanted confirmation that the medicine and symptom were causally associated and 27 expected to receive this.

Almost all respondents (n = 1302; 95.6%) would make a report again if they, or someone they knew, had a suspected ADR. A similar number would encourage other people to report through the YCS (n = 1300; 95.4%).

The most frequently cited reasons for reporting again related to the need for collection of data on medicines (97 respondents), to help others (n = 57) or to improve the quality of medicines, medicines information or care (n = 51). The high prevalence of ADRs, the distress caused and the need to record patients’ experiences were also mentioned:

I think it is important to have reports from patients themselves as part of the post-marketing surveillance. Patients’ experiences are very important.

Female, 32 years, graduate

Some comments (n = 25) explained reasons for respondents not reporting again, including the time taken to make the report, the lack of feedback received and the perception that their report has little impact:

But think some feedback should be given, otherwise what’s the point? How would I know if it was making any difference?

Female, 50 years, further educational qualification

Although 23 respondents stated that they definitely would or indeed already had encouraged others to report, 22 provided reasons for not encouraging others and a further 34 indicated conditional encouragement only:

I wouldn’t want people to report unless they feel strongly that the product caused the effect, because it could reduce the value of these reports.

Female, 26 years, graduate

General views on Yellow Card reporting

An open question ‘Why do you think it is important for people to report side effects from medicines using the YCS?’ elicited a total of 1802 specific comments from 1238 respondents, although 103 were judged to be unrelated to the question. The view most frequently expressed (by 355 respondents) illustrated an awareness of the purpose and importance of the scheme as a means of gathering data from a large population taking medicines. Respondents also cited the need to prevent others from suffering similar problems (n = 165) and to make the public or other patients aware of side effects from medicines (n = 152):

Might be small number of people scattered over country or something you think people are aware of but aren’t. Is good for someone to collate this info together to be able to see the whole picture and spot major problems with a particular medicine or any trends to stop others from suffering or make sure people are aware before they take them.

Female, 34 years, left school at 16 years

Others felt it was important to make MHRA or manufacturers aware of ADRs (n = 100), with many respondents expressing the hope that as a result medicines may be improved or PILs amended (n = 173) or products withdrawn from the market if necessary (n = 41):

This should ensure that drug companies are made aware of severe side effects so that they can issue appropriate warnings. They should also carry out further research to reduce the risks of side effects.

Female, 65 years, higher educational qualification

The view that health professionals need to be informed about ADRs and perhaps to change their practice was raised by 167 respondents. Others criticised government incentives to prescribe certain drugs and a number of respondents expressed distrust in the pharmaceutical industry, as well as in health care and government (n = 79):

Absolutely necessary to act as backup to clinical trials, plus acting as a brake on government instructions to issue pills on a one-size-fits-all basis, i.e. how many blood pressure and cholesterol tablets are issued to borderline cases in order to meet targets and gain bonuses?

Male, 72 years, left school at 16 years

For too long there does not seem to have been any method the general public can input side effects from their medication. It also means (hopefully) things can be collated, monitored, and fixed in shorter period of time.

Female, 38 years, higher educational qualification

The attitude of health professionals to suspected ADRs was raised by 106 respondents, who indicated both dismissive attitudes among HCPs and their failure to report ADRs. The need for a reporting mechanism which is independent of health professionals (n = 32) and for patients’ voices to be heard were also identified as issues (n = 49). A further 90 comments related to the importance of the patient’s perspective, indicating that side effects were, or could be, severe; may be perceived differently by patients or may be worse than the underlying medical problem:

Because I feel if I told my doctor about my side effect, he would belittle it and I would wrongly feel I was making too much of a fuss, whereas the reality of my experience of the side effect was not wrong or little. I feel scared of my GP’s response. Being able to report side effects effectively anonymously allows side effects to be known by pharmacologists, which would otherwise remain unreported or lost in a GP’s database somewhere.

Female, 43 years, graduate

So the MHRA have independent information about prescribed drugs. Sometimes GPs may not believe patients reporting side effects. It helps patients recover from side effects ‘getting it off your chest’. Patients hope that someone might take notice/make amends on patients’ suffering. People need to know they have a voice.

Female, 58 years, postgraduate

What doctors and nurses regard as a side effect worth reporting may differ from what the patient thinks. My mother has had a life changing experience that came out of the blue and is keen to ensure others don’t experience the same problem.

Female, 57 years, postgraduate

Some comments related to the need for research data on medicines, whereas others hoped that by drawing attention to problems, novel research would be initiated (n = 66). A wide range of other comments were received (n = 116), covering issues such as side effects arising from drug combinations or changes in brand, suggestions of ‘fake’ products, mistakes in production, or reactions to excipients. Criticisms of both health care and licensing were also raised:

Different (as far as I can tell) manufacturers may have different recipes. One may be okay and another not. I don’t know.

Female, 80 years, left school at 16 years

Because too many doctors prescribe inappropriately bowing to pressure from reps, patients and health authorities. In addition, a number of pharma companies tend not to look for AE data and so medicines are licensed on the basis of too small a safety database. Real life experience is the only way to build the database.

Male, 49 years, postgraduate

I think it is too easy for HCPs to go for the ‘quick fix’ if they know that a particular medication will ease a problem, but they don’t follow up often on repeat prescriptions and don’t really view subsequent problems holistically so can miss links to other reported problems. I see older people in relation to my work, they often have repeat prescriptions with many items on and some items are to deal with side effects of other meds. GPs don’t have time to follow through. I think we, as patients, should do more to raise awareness of these issues.

Female, 53 years, higher educational qualification

Attribution of symptoms to medicines

The questionnaire also explored respondents’ views about what made them think that the medicine caused the side effect. There were 1348 responses to this question, 99% of all respondents. Of these, 181 (13.3%) were not analysable as 78 listed only symptoms and 103 were comments on other issues or inadequate descriptions. Of the remaining 1167 respondents, most (689; 59.0%) provided only one piece of information to explain causal association. However, 382 (32.7%) respondents provided two pieces of information, whereas 87 (7.5%) provided three and nine respondents (0.8%) provided four.

Among these 1167 responses, 675 (57.2%) mentioned reasons relating to timing only. A further 238 (20.2%) indicated using information sources and 145 (12.3%) mentioned both. Explanations included: symptom not present before medicine started (n = 195, 16.7%), symptom began soon after starting medicine (n = 320: 27.4%), symptom reduced on stopping (n = 275, 23.6%) and symptom re-occurred on re-challenge (n = 89, 7.6%). A further 37 noted symptom changes with dose changes, 25 noted symptoms started after medicine withdrawal and 29 on changing brand. In addition, 22 indicated it was a new medicine and 78 cited other timing-related issues:

Never had it before. Didn’t have it after stopped. Was worse on higher doses. Was better on lower doses.

Male, 32 years, graduate

I had been taking Pfizer’s Istin (amlodipine) for over six months and when I suddenly could not find it anywhere I was forced to take a generic version. Within half an hour of taking this I could feel symptoms which I had never experienced with Pfizers Istin. I reported them immediately to the doctor.

Female, 63 years, higher educational qualification

The most common sources of information used to associate the symptom with the medicine were HCP (n = 186; 15.9%), and information from the internet, information leaflet or books (n = 182; 15.6%). A further 57 (4.9%) indicated other information sources:

Read about possible side effects on internet – commencement of taking medication coincided with my own medical problems. This was backed up by consultant.

Female, 65 years, postgraduate

The onset was a short time (few weeks) after commencing the drug. I looked it up in the literature and consulted several clinical friends.

Male, 77 years, postgraduate

In addition, 33 indicated previous reactions to the same or similar drugs, while 25 mentioned they or their doctor had seen similar effects in other people. A substantial number (n = 130) also felt that, for various reasons, no other explanation for their reaction was possible, including elimination of other possible causes. Other reasons/views were given by 27 respondents, including the severity of the event, interactions, specific excipients and confirmation by tests:

Only medication at the time. Several of my friends on this drug have had similar side effects.

Female, 71 years, higher educational qualification

I know that I’m allergic to E104 and discovered after suffering severe migraines that this colouring was in some of my medication.

Female, 47 years, graduate

Conduct and analysis of interviews

The semistructured telephone interviews were conducted by two researchers using an interview guide (Appendix 19) developed by the research team. The first three interviews were performed by one researcher (AG) and the remaining 24 by a second researcher (TP). The development of the guide was informed by a preliminary analysis of the questionnaire data and issues identified by the project team including:

• exploration of any difficulties in making Yellow Card reports and suggestions for improvement in the reporting system

• patients’ motivations for making the report and anticipated contribution of their report

• patients’ expectations about what would happen to their report

• patients’ satisfaction or dissatisfaction with the process of making a report

• patients’ willingness to report in future.

The areas to be explored were reviewed and revised in the light of the data obtained from the initial interviews to ensure that the data obtained were relevant to the focus of the research. For example, in the early interviews respondents tended to focus on the story of their ADR rather than their experience of reporting it. So, in subsequent interviews we shifted the focus away from the reaction towards the reporting scheme. Interviews were audio-recorded digitally, with consent, and transcribed verbatim. The interview transcripts were analysed by an academic member of the research team (CA) and the analysis checked by another academic (EM). Data were analysed for both anticipated (based on the questions asked) and emergent themes using the constant comparison method. The researcher first read the interviews and noted the main themes. The data were then categorised into the major themes, which were finding out about reporting, recognition of adverse effects, reason for reporting, involvement of others in decision to report, how patient reports might differ from health professional reports, ease of reporting, what people expected to happen following making their report, advertising the scheme, and reporting again and encouraging others to report. The researcher then identified from the data a number of subcategories for each of these themes. For example, under the major theme of ‘reasons for reporting’ a number of subcategories emerged from the interview data, including altruism, solidarity and pharmacovigilance. Quotes are used to illustrate the themes, and for each quote the following information is provided: interview number, gender and age of reporter and mode of reporting, for example ‘interview 12, female 54 years, paper’.

Results

Twenty-seven telephone interviews were carried out; 23 interviewees had originally completed paper reports, two internet and two telephone reports. Two reports were done by parents on behalf of their children. A summary of the characteristics of the interviewees is shown in Table 29 and more detailed information on each interviewee is shown in Table 30. All of the interviewees were of white-British ethnic origin.

Recognition of adverse effects

Most interviewees were clearly able to link the ADR with taking a particular medicine, including making temporal links:

I took this one capsule of the antibiotic and er I was talking to a friend on the phone and er within half an hour my tongue started to swell and I had to ring off because I couldn’t speak … and then I had mucous streaming from my nose and mouth. Oh it was awful, I felt sick and broke out into a rash all over my body which itched and felt hot.

Interview 2, female, 72 years, paper

… that was the Friday evening then on the Saturday morning I started to itch and noticed that I had a rash that was fairly extensive, pretty well covered in a very fine rash and didn’t feel very well at all and in the evening I just went to bed early. I didn’t take any more of the medication, I only had the one dose I think and it was because of the rash and feeling really sick and I think I felt very cold, I was very shivery.

Interview 11, female, 57 years, paper

For another reporter’s daughter, Stevens–Johnson syndrome started within 12 hours of starting penicillin.

Patients also noted a ‘dose response’. For someone who had taken pregabalin for pain, the reaction started the next morning after taking the first dose and became worse on the second dose. This patient recalled a previous similar experience:

Researcher: So how long had you been taking your medication before you realised that there was a problem?

Interviewee: Oh gosh erm, er a matter of days. Erm I sh’ it should be on the form.

Interviewee: Yes, it appeared following the morning after taking the dose the night before?

Interviewee: That’s right. Yeah.

Researcher: And then got worse after the second dose.

Interviewee: That’s right yeah because I’d had a similar reaction on, on meloxicam.

Researcher: Right so you phoned your GP who told you to …

Interviewee: Stop, stop!

Interviewee: within 24 hours I was feeling a lot better.

Interview 4, female, 69 years, paper

An interviewee who had been encouraged to fill in the report by a pharmacist at the store where she worked had been taking azathioprine for a flare-up of Crohn’s disease for 2 weeks when she started feeling sick. The doctor had asked her to take the azathioprine again as a re-challenge to ensure it was causing the ADR:

So I started taking that and about maybe two weeks into taking it and I started feeling sick. Reading on the instructions it just said that it can cause nausea but these are sort of like, when you read these instructions it’s maybe, you know, one in however many people will have these … Yes, so I started feeling sick as I say and I thought right, that will be fine, I’ll just sort of ride through it and it just got worse until in the end I was being sick, violently sick. So I went back to my GP rather than the hospital because it was just every so often that I went to the hospital and he said, ‘Right let’s stop taking it.’ So I stopped taking it on a particular Wednesday and then he said, ‘Leave it a full week and then start taking the drug again and then we’ll know whether it’s that that’s actually making you be sick’. So I started taking it on the Wednesday morning and I remember taking it at half past 8 just before going into work and by 9 o’clock I was being violently sick again and I was being sick about every 5 minutes.

Interview 9, female, 47 years, paper

Another man had been taking a prostate medication and had experienced recurrence of symptoms on re-challenge:

I think it was one day, two days and it was most pronounced, the symptom was pronounced and definite and I was very surprised. Stopped taking it for I think a few days, repeated it and got exactly the same results much to my amazement. It was more pronounced, most definite and repeatable.

Interview 23, male, 61 years, paper

One man was convinced that his tablets for hypertension were causing the problem as all his adverse effects stopped when he ran out of tablets while on holiday, only to come back when he started to take extra tablets:

Well I had no problems up to start of taking the tablets and when I went abroad and ran out of the tablets everything returned to normal … But it was only when I came back, started retaking the tablets that the symptoms started to reappear and when I started taking the extra tablets the symptoms, they got extremely bad.

Interview 24, male, 66 years, paper

Why interviewees had made the report

Involvement of others in decisions to report adverse drug reactions

A number of interviewees had been encouraged to report by their pharmacist. One woman who had been taking statins for 6 months was discussing her thinning hair with her hairdresser who asked her if she was taking any medicines. She then went to her pharmacist who confirmed the hairdressers’ diagnosis. This is the only example in our sample of a layperson identifying an ADR:

She said, ‘Are you on any new medications?’ But of course that was something I hadn’t considered and I told her about the statins and she said, ‘That’s funny’, and this girl that does my hair, she said, ‘My mum stopped taking statins for the same reason because she thought her hair was falling out’. So I filled it in. It was my pharmacist when I went to get a repeat prescription and she said, ‘Oh simvastatin’, and I said, ‘It’s making my hair fall out’. And she said, ‘Oh you should report that’, she said, you know, ‘they’re keen to know of any contra-indications sort of thing.’

Interview 7, female, 48 years, paper

Others said that they made their report because they did not think that they could rely on their GP to make a report on their behalf or could not rely on it being accurate. These interviewees felt that their report would compensate for their GP’s lack of commitment to reporting:

Yes. Because I’ve used it before. I’d reported pregabalin and meloxicam both of them I’d reacted badly to. I thought well if people don’t report you don’t know erm and I couldn’t rely on GP reporting.

Interview 4, female, 69 years, paper

If it had just been going to the doctor and reporting it I wouldn’t have gone to the doctor and reported it but the fact that this card was available, I filled it in. I gather from what I’ve read that a lot of GPs don’t bother reporting side effects anyway. I mean that’s just hearsay, I mean that’s what I’ve read but I don’t know if it’s true or not. Maybe that was the case 5 years ago, maybe it isn’t now.

Interview 23, male, 61 years, paper

How patient reports might differ from health professional reports

Three interviewees commented about how a patient report might differ from a HCP report, compensating for the perceived shortcomings of health professional reporting:

I think by the time it’s filtered through, through various other people it doesn’t actually go in the form in which the patient actually reports it … to, to make mine fairly specific erm and it was fairly specific and quite, and fairly detailed but I’m sure that anybody else would simply, a lot of health professionals would simply specify some general adverse symptoms and that would be it.

Interview 4, female, 69 years, paper

I think it is important that people do get the opportunity to report their own understanding of a problem which may be different from a professional’s understanding of a problem. And that, you know, some people just prefer to do things themselves than to have to go through an official course, you know, to have to go to the doctors and ask them to report it or, I think if people knew the reporting system was there and available more then they would use it.

Interview 11, female, 57 years, paper

I think it’s far better than reporting it to your GP because, you know, you have more time to sit and think about it and it doesn’t get filtered in the process if you know what I mean, you know what actually happened because you get so little time with the GP, you wouldn’t have time to think it through and put all the exact details down.

Interview 23, male, 61 years, paper

Ease of reporting

Although some of those interviewed found it straightforward to report, a number of difficulties were mentioned including lack of space on the paper form. Further details are not given here, as the issues concerning ease of reporting are covered more fully in Chapters 7 and 9.

What interviewees expected to happen after making their report

Eight interviewees had not expected a response to their report, although were pleased when they received one:

Nothing. I expected absolutely nothing and I was very surprised when I was informed that they’d received it and it had been noted and I thought well this is progress, most of the time things that you send off to official bodies just sort of disappear into the ether and you never hear about them again. So I was quite chuffed to hear back, feedback was the word I was looking for and yeah that was fine.

Interview 12, female, 54 years, paper

Well I did get a letter thanking me for the card, which I didn’t expect and that was nice because it made me feel that actually someone is taking notice there.

Interview 15, female, 43 years, paper

Ten interviewees said that they had expected to receive an acknowledgement and or some action:

Well I don’t know, I suppose maybe an acknowledgement that somebody had received it but nothing particularly further than that other than it might be of some use to somebody who was monitoring different medications.

Interview 11, female, 57, paper

Others showed some understanding of the pharmacovigilance process, but were cynical as to whether anything would actually change as a result of their report:

Then I expected them to collect them and if there were more than a certain number of people report the side effect then it would get flagged up as a common one in the NHS or, I don’t know whether anything actually goes back to the manufacturers or if there are any studies that might come out of, you know, even if there might be something new that might put up a red flag, they might think well maybe we should look at this and find out what’s happening there and instigate some further research. That would be my hope but I tend to think that once something’s licensed and they’re making money then they’re not really interested in going back and doing further research

Interview 15, female, 43 years, paper

What was I expecting to happen? Very little, I suppose I expected, I hoped there’d be an acknowledgement because you’re never sure when you send something off it’s going to get there but I didn’t expect any benefit at all or anything to happen to me, I was just reporting what happened for some database somewhere really. I felt I wanted to be sure someone actually got it as an input.

Interview 23, male, 61 years, paper

One had not heard anything and felt ignored:

I take it from the fact that I’ve heard nothing is that I’m a maverick and they’ve ignored it, don’t worry about it, it’s not a significant problem so let’s not worry about it.

Interview 25, male, 64 years, paper

Advertising the Yellow Card Scheme

There was a general feeling among the interviewees that people ought to be aware of the scheme and not find out about it by chance, like they had. There was also a feeling that people should be encouraged to report as some people thought that they might not be believed:

… but one thing I do want to say is could the MHRA advertise more? The more they advertise it, the more people will be aware and the less people will suffer side effects.

Interview 1, female, 74 years, paper

Well I think the Yellow Card could be advertised better so people know what’s going on. I only heard about it by chance from the radio.

Interview 2, female, 72 years, paper

Well I’m not sure that the little Yellow Card is the best way of bringing it to people’s attention. I think we’ve got some in our surgery but I think on the whole people are reluctant to report any side effects just in case they’re not believed.

Interview 10, male, 75 years, paper

The interviewees had a number of ideas as to how the scheme might be better advertised to the public. Interviewees suggested putting information on or in the pharmacy bag or on the medicine label:

Another thing I think you could do, you know on your pharmacy bag, you get another bag to put your tablets in, it could even be printed on the paper bags they put your medication in … Or even put a Yellow Card in with your medication to just say if these tablets have unforeseen, upset you in any way after two or three days, get back in touch with your doctor and tell them that you’re going to fill this form in or go back into the chemist and say, ‘Right they’re not agreeing with me, I’ve been in touch with the doctor, I want to fill this form in.’

Interview 8, female, 66 years, paper

In fact I would go so far as to say that in every single box of medication … there should, or information about the Yellow Card.

Interview 12, female, 54 years, paper

She also suggested running a television advertising campaign:

I mean why don’t they even do a TV ad. I bet a sort of TV ad when people are sat there in front of their tellies watching Coronation Street or something, I don’t know, look at everybody out there that takes medication, ‘Do you know there is a scheme that’s called the Yellow Card de da, de da, de da’ and make people aware that it’s out there.

Interview 12, female, 54 years, paper

Reporting again and encouraging others to report

The majority of interviewees were very positive about reporting again or encouraging other people to report ADRs:

Yes I think it’s a really good thing, I still think most people aren’t aware of it or really understand, I think lots of people who are on medication don’t even know what the YCS is or were aware of it even when it was just the GPs could fill them in…There were people I suggested it to.

Interview 15, female, 43 years, paper

A number of interviews had already actively promoted the YCS to others as a matter of course:

I run a shop in a village and I’ve heard of people who’ve had side effects and I’ve explained what the Yellow Card System is and I don’t know if it’s sinking in or not but it helps, it’s helpful. It’s not guaranteed that it helps for everything or they even fill it in.

Interview 22, male, 52 years, paper

And I run a support group in [place name] for thyroid patients, when they’re having problems I’ll always mention the Yellow Card to them as well.

Interview 21, male, 72 years, paper

Because I can tell you, I have been in touch, with friends, colleagues, from the different organisations I belong to. I do a lot of charity work, you know, and people in the village here you know and it is amazing how many people are on prescribed drugs which are giving them dreadful side effects. I am trying to persuade them to get in touch you know on the YCS. So their information is recorded and some sort of investigation can be carried out, you know.

Interview 1, female, 74 years, paper

Some saw the importance of reporting for pharmacovigilance:

Well yes, I mean I suppose each individual is probably different aren’t they, maybe what one person suffers maybe the other person doesn’t … But, you know, occasionally there’s a pattern maybe there’s a lot of people have had the same thing …

Interview 17, male, 49 years, paper

Yes unless the public know about it people are going to be worried. I think it’s important to report as many side effects as possible so that they can list them then when people take the drug they know what to expect and they don’t think it’s their health.

Interview 18, female, 50 years, paper

Recruitment

We planned to recruit six focus groups with up to eight subjects in each. For logistical purposes participants were recruited from only the Nottingham area and recruitment took place between June and November 2008. Although we recruited from only one area of the country, we aimed to recruit a broad demographic sample of people. Originally, we had planned to include patients who believed they may have experienced side effects from medications, but had not previously filled in a Yellow Card report. We decided, however, that this would make recruitment difficult and that there would be some advantages to widening our selection criteria to all adults. For example, this allowed us to recruit younger people, who might not have had side effects from drugs, but who might be more familiar with the use of online reporting systems. Three main approaches were used to recruit participants with a range of demographic characteristics.

Recruitment posters and leaflets were displayed in general practices, pharmacies, social centres and in the University of Nottingham. In addition, an article about the research study was placed in the local press. Finally, the pool of simulated patients utilised by the University of Nottingham Medical School for the purposes of supporting medical student training were invited by letter. Those interested in participating in the research were invited to contact the research team directly by telephone. Also, the simulated patients had the option of returning a consent form in a stamped addressed envelope provided.

All of those volunteering to participate in the research were offered a choice of dates for attendance at the focus group/usability sessions. The dates were confirmed by letter. Sessions were run on different days of the week and times of the day, to enable participants to attend at a time of their convenience. Volunteers were all offered a £25 inconvenience allowance. Sessions were conducted at the University of Nottingham where computer facilities were available. Refreshments were offered prior to the commencement of the focus groups.

Focus groups

The topic schedule for the focus group was developed by the research team, informed by the literature and project objectives. It included the participants’ thoughts about the YCS, whether they felt they might utilise the scheme and how they felt the scheme might be improved. Finally, each participant was asked which method of reporting they felt that they would prefer to use. After completing the usability tests, participants were asked whether their views had changed.

A pilot focus group with members of the research team was conducted to confirm clarity and appropriateness of wording. As a result, the schedule was revised. The final version is shown in Appendix 20.

Before the focus group/usability session began participants were asked to complete a short questionnaire asking for basic demographic data and experience of using computers. These data were used to inform which of the usability tests participants were allocated to.

At the start of the focus group/usability session there was a 15-minute presentation, by a member of the research team, explaining the YCS and the research project. We felt it necessary to give a presentation to provide context for the focus group discussions and for the subsequent usability testing. At the end of the presentation, there was an opportunity for participants to ask questions. They were then asked to complete and sign a consent form to document their agreement to participate in the research and to the audio-recording of the focus group/usability session. The signed consent form was returned to the research team. Participants retained a duplicate copy. Once consent had been obtained from each participant, the focus group was conducted. Recordings of the focus group/usability sessions were transcribed verbatim. The focus group data were analysed thematically (by CA and AG), identifying the major predicted and emerging themes and ideas from the data.

Usability tests

Characteristics of participants

Overall 40 participants took part in seven focus groups/usability sessions (22 recruited via the posters, 11 from the simulated patients and seven from the newspaper article). Table 31 shows the characteristics of participants.

Over two-thirds of the participants were female and most (74%) were aged 50 years or over. Despite widely publicising the project there were few men or younger people.

Two of the participants were hearing impaired and one had to use a magnifying glass with the large print materials in order to overcome his visual impairment.

Thirty-seven people completed simulated online reports, 36 people completed paper reports and eight people completed telephone reports (81 reports in total).

Awareness and value of the Yellow Card Scheme and direct patient reporting

Of the 40 participants, two had previous knowledge of the YCS: one was a retired dentist, who knew of the scheme through his profession and another had heard of it on a radio programme.

After hearing the initial presentation introducing the YCS, all of the participants felt that it was a very worthwhile scheme and they could see the benefit of asking patients to report ADRs directly to the MHRA:

It’s very sensible, you don’t know that the general practitioners are going to complete the forms even if you go back to them, but this way they will get more information I suppose. Whether they can use the information or not I don’t know, or whether they get too much information or not, I don’t know. But at least they are getting information.

Female, 60–69 years

Several of the participants also commented that they had reported side effects from medications to their doctor previously and had felt that the doctor was disinterested or dismissive. They believed that direct patient reporting would avoid information being reported to the MHRA through a professional lens.

In two of the focus groups there was some discussion relating to the quality of data that would be reported by patients and the possibility that this could be influenced by varying levels of literacy and education.

Although most of the participants felt that it was a positive step to allow members of the public to make reports to the YCS, a few felt that the GP should still be informed about the potential adverse reaction. Three participants stated that it should remain their GP’s responsibility to report such ADRs to the YCS, as this would be the most appropriate way of screening out those problems that were minor and not worth reporting. One participant expressed concern that direct patient reporting could result in the patient not discussing the issue with the GP. Others saw direct patient reporting as an option in situations where the GP did not agree to report a suspected ADR. Nevertheless, several participants felt that it was a professional responsibility of HCPs to report ADRs. Some participants recognised that direct patient reporting provided a mechanism to report adverse events from medicines bought from pharmacies.

Identifying adverse drug reactions

One issue raised in all of the focus groups was the difficulty of determining whether the symptoms experienced were a side effect of the medicine, due to the patient’s illness, or unrelated to their health problems. In addition, several participants discussed the potential complexity of deciding which drug was causing an ADR if the patient was taking multiple drug therapies:

I think my problem would be on the complexity of it all. I wouldn’t be able to really say which drug it was coming from. I find it useless information to give as I have a mishmash. If you give them the wrong one it could be wrong.

Female, 70–79 years

Determining the severity of the adverse reaction was also discussed. There was great concern over what would constitute a side effect worth reporting; whereas some participants felt they would be keen to report any potential side effect, many feared that reports of minor or unimportant side effects would ‘waste’ the MHRA’s time. There was general agreement that participants would only report side effects that were serious enough to be debilitating:

I wouldn’t bother reporting it unless I thought it [the ADR] was serious.

Female, 18–29 years

If it was just a headache or something I wouldn’t make a report, but something life threatening I would.

Female, 40–49 years

Participants also raised the issue of potential inter-patient variability on determining of the severity of ADRs:

How do I know, that, you know, my headache which I think is 8 on a scale of 1 to 10 is the same as yours (nodding at another participant) or yours?

Male, 50–59 years

The participants saw potential issues relating to the consistency of reporting of the severity of an ADR from members of the public, with no effective ‘filtering’ by a prescriber with greater medical knowledge.

Issues regarding reporting

Participants were informed that they would only receive an acknowledgement of their report from the MHRA. Concern was raised about this. Several participants felt that the lack of detailed feedback would potentially discourage them from completing a report:

If you are still suffering from the side effect, are still ill, er then I am not sure I would think it was worth the effort required to complete the form if I wasn’t going to get any advice or something, perhaps some personal feedback.

Female, 40–49 years

The majority of the participants felt that there ought to be more feedback provided to the reporter than a simple acknowledgement. It was suggested that this feedback might simply to give the reporter an idea of the frequency of the side effect reported for the drug involved. The participants felt that reporters would want to know how their experience of a side effect related to that of other people reporting side effects from the same drug. It was felt important that the reporting forms and information provided about the YCS made clear that feedback on an individual’s specific case and medical condition would not be possible. This would avoid disappointment or anger when reporters did not receive a more detailed response.

Advertising of the Yellow Card Scheme

None of the participants had seen a poster or leaflet relating to the YCS, despite some of those involved in the focus groups being individuals who attended regularly both GP and outpatient clinics and received regular prescriptions from their local pharmacy. A high level of concern was expressed by participants about this apparent lack of promotion of the scheme.

Advertising the YCS was discussed in some detail, with participants mentioning that there needed to be much wider awareness in order to ensure that the maximum volume of reports were obtained from the patient reporting system. Without extensive advertising, there would be a limited response from the public:

If you don’t know about it you can’t do anything can you? … I think the surgeries or pharmacies or whoever being more proactive in displaying all of this with posters and what have you so that the general public really do know what it is for. I don’t think that there is any other way that the general public are going to learn about it.

Male, 20–29 years

Several suggestions were made on how to improve awareness of the scheme among those most likely to suffer from ADRs. Simple measures, such as ensuring that posters are displayed in all pharmacies, general practice surgeries and hospital outpatient departments were seen to be a good method for raising general awareness. To target those receiving prescribed medicines, it was suggested that there could be a leaflet included in the bag with repeat prescription items:

Mmm … would it not be better that any time you collect any medication or something from your chemist, then one of these little Yellow Cards are put in separately. Because then it’s, it’s in your face as soon as you open your bag …

Female, 50–59 years

However, some felt that the leaflet might be treated as ‘junk mail’, and so would be an expensive and inefficient method of raising awareness. Another suggestion was to include information about the YCS on the bottom of each PIL provided within the medicine packaging by the manufacturer. Participants suggested that even if people do not read these leaflets routinely, they were much more likely to go back and read them if they suspected that they were suffering from an adverse reaction. Therefore, this would be an appropriate place to site information on the YCS.

Another suggestion for raising awareness included advertising on radio stations, particularly those targeted at the older population, as this section of the population takes the largest proportion of prescribed medicines. Attempts could also be made to have the YCS discussed on popular television programmes that deal with viewers’ medically related issues.

Following on from the need to raise awareness of the system, participants felt there was also a need to ensure easy access to the reporting forms in surgeries and pharmacies, with forms being clearly displayed in both settings. The participants also felt that HCPs could play a more proactive role in encouraging reporting by patients, and suggested that GPs and pharmacists could hand out the reporting forms to all patients they saw.

Paper reporting

The form was generally well received, with many of the participants commenting positively:

This is an excellent form and with my own information I could er … fill it in easily.

Female, 60–69 years

Two participants with visual impairment found the font size too small. Several participants commented that they would not bother reading the information about the YCS provided within the integrated information leaflet and reporting form; they would just start filling in the form. It should be noted, however, that the scheme had been explained to participants in the focus group directly before they undertook the usability tests.

In some of the scenarios, details were provided about a number of drugs that were being taken by the fictional individual and some of the participants found it difficult to fit all the required information onto the form. Although the form suggests that further information can be provided on a separate sheet, participants were generally not keen on this:

Having to attach extra sheets to give all the information would annoy me, so I wouldn’t bother!

Male, 70–79 years

I’m frustrated to have to add additional sheets.

Female, 70–79 years

And some felt that this part of the form was tedious and long winded:

I have lost the will to live (comment made when entering all details of the drugs).

Female, 50–59 years

Some individuals commented that they would struggle to remember the date on which their various drugs had started and finished and they would need to look up this information.

The layout of the form was commented on by some participants. Details about the reporter are asked for at the end of the report after details of the ADR and the severity of the side effect. This did not feel natural to some of participants, who felt that detailed ADR information should be requested after more basic information has been collected.

Whereas all the scenarios included the height and weight of the fictional character experiencing the side effect, some participants commented that they would not necessarily be able to provide these details, particularly their weight.

When participants tried to insert their completed form in to the reply-paid envelope, which is also part of the leaflet, they were disappointed to discover that the form did not slide neatly into it; instead, the form had to be folded:

The envelope is the same width as the form, so it doesn’t go in without folding. And the adhesive is not very good.

Male, 18–29 years

One participant commented that the use of a first-class reply-paid envelope made her feel that the MHRA was taking the whole system very seriously.

It was suggested that access to the system would be improved by making the paper-based and online forms available in languages other than English:

It might be useful to er you know, provide the service in other languages, er not only English.

Male, 60–69 years

Telephone reporting

The telephone reporting system was popular with all of those who used it. The interactive nature of the conversation with the person staffing the telephone was highlighted as a factor that made the process much easier for the reporter. Participants expressed concern that such a convenient method of reporting was only available for a limited number of hours, thereby limiting access. They were pleased to be told that the reporting line has a direct-dial number which does not require reporters to enter a queuing system with an automated response menu:

I might be put off by having to pay for the phone call … the time restriction of the service might put me off too.

Female, 40–49 years

The only issue encountered with the telephone system occurred to one participant who had impaired hearing. He found that he was unable to hear the person taking the report because of background noise; the test had to be abandoned because of this. However, the participant stated that at home he would not have encountered this problem and he did not consider this to be an inherent problem.

Online reporting

The online reporting system received mixed comments from participants. There were a few who found the system easy to navigate, but many found it very challenging, even though they used the internet regularly and completed forms online.

When reporting the nature of the suspected ADR, there are a number of drop-down menus (dictionaries) that provide mixed-response options. However, the complexity of the terms used within these menus confused a number of participants. For example, entering an ADR term of ‘rash’ produced over 70 medical terms for rash (such as ‘rash desquamating’, ‘rash morbilliform’). Participants were often unable to understand these terms, and many participants felt forced to choose the first term on the list (which is always a general term) or a term at random.

A number of participants did not see the drop-down menus appear, as they were concentrating on the next part of the form that they had to complete. This resulted in lack of further detail being provided.

There appeared to be no facility to save data at each stage of completing the online form. At least two participants had problems with losing information after taking some considerable time completing sections of the form. Having wasted their time, these participants told us that they would not have finished completing the form in real life.

One comment made by a number of participants related to the different ordering of questions on the online and paper forms. A consistent approach would make it easier for users who may use the paper form to collate their information, but then choose to submit the data using the online system.

The online system also asks for extra information to that requested on the paper form. It is important to consider whether this additional information is required and, if so, whether it should also be collected on the paper form.

Analysis

Data were analysed using spss for Windows version 15.0. Data were described using frequencies and percentages. Original BMRB codes for ethnic background were combined into five major ethnic groups: (1) white Irish, white British and any other white background; (2) mixed African, mixed Caribbean, mixed Asian and any other mixed background; (3) Indian, Pakistani and any other Asian background; (4) African, Caribbean and any other black background; and (5) other ethnic background. The chi-squared test was used to compare the sociodemographic profile of our sample with UK general population figures of 2008. Similarly, the chi-squared test was used to examine sociodemographic variables across level of awareness of the patient reporting to the YCS. Multiple logistic regression was used to identify independent patient demographic predictors of awareness of the YCS. Mann–Whitney or Kruskal–Wallis tests were used to compare median convenience of reporting scores across sociodemographic factors. A p-value of ≤ 0.01 was used to denote statistical significance throughout.

How do patients describe suspected side effects, and how does the richness of patient reports compare with those of health-care professionals?

Patients gave detailed descriptions of suspected ADRs, attributed reactions to specific medicines and provided information useful for assessing causality. Patient reports often had richer narratives than those of HCPs and rarely provided irrelevant information or ambiguities. Patient reports also often contained detailed information about the impact of the suspected ADR on the patient’s life, thus providing insights that there were comparatively rare in HCP reports.

These findings are in keeping with analyses of consumer reports of reactions to paroxetine52 compared with reports from HCPs to the YCS. 52 The study authors noted:52

Our analyses suggest that reports from patients – in their own words – communicate essential information which professional reporters can never be expected to provide. In this case, patients provided reports that were much richer in their descriptions of behavioural phenomena and feelings than the YC [Yellow Card] reports, and often much better at explaining the nature, significance and consequences of adverse drug effects.

Medawar and Herxheimer52 also identified the potential for some terms used by patients, such as electric shock sensation associated with paroxetine, to be coded as less-specific terms, such as paraesthesia. In our study, we did not formally check the text from patients against coded reaction terms, but this information was readily available when selecting the cases for qualitative analysis. Our impression was that symptom descriptions were coded accurately and comprehensively. Nevertheless, if an entirely new symptom not listed in the MedDRA dictionary was to emerge in relation to a medicine, delays would occur in detection if there was an overemphasis on coding at the expense of careful reading and consideration of reports. Also, it is important to recognise that reports from either patients or HCPs will lose their richness when converted to symptom codes. Therefore, detailed reading of reports is still an extremely important task for regulatory bodies, particularly in relation to drugs undergoing intensive monitoring.

In keeping with our findings, an earlier literature review of patient reporting of ADRs24 concluded that none of the countries with patient reporting systems had identified poor quality of patient reports as an issue. Van Grootheest and colleagues53 state that the quality of the report should be distinguished from the quality of the clinical judgement. In terms of the quality of the report, they identified the following elements needed to make a report useful: description of the adverse reaction, time of onset of the reaction, the drug involved, concomitant medication, duration of drug use, including dates of initiation and discontinuation, and any other relevant information. Most of the patient reports in our qualitative study contained this information. Quality of the judgement, relates to whether patients can distinguish drug-related complaints from other problems, for example disease-related complaints. 53 Many of the patients in our qualitative study appeared to be able to differentiate between symptoms of disease and ADRs, sometimes explaining their reasoning for this judgement.

The pharmacovigilance impact of suspected adverse drug reaction reports from patients and health-care-professionals

Spontaneous reporting of suspected ADRs by HCPs is a valuable pharmacovigilance tool1,54 and has contributed to the identification of major safety issues. 1 There is general acknowledgement that patient ADR reporting is the right thing to do and that it will enhance pharmacovigilance,6,53,55 although the specific value of ADR reports from patients has not been clear.

Although proponents have highlighted potential benefits from patient reporting of ADRs,30,56,57 others have expressed concerns. Many of these concerns have been highlighted already, including the capability of patients to distinguish between suspect ADRs and problems associated with the underlying condition being treated by the suspect drug;58 unclear ADR description;58 large numbers of reports of minor and well-known ADRs adversely affecting the ability to identify new drug safety issues;1 duplication of ADR reports;1 resources available to regulators;1 the potential influence of campaigning groups on signal generation;1 and perceived undermining of HCP status. 59

Our qualitative findings should reassure those who worry that patients may not be able to give useful information for detailed pharmacovigilance assessment, such as the detailed reading of narratives required to assess the potential causality and impact of suspected ADRs. Other issues relevant to pharmacovigilance include:

• seriousness of reports

• outcome of the ADR

• timeliness of reports

• role of patient reports in the generation of new signals.

Other observations regarding patient and health-care professional reports

Awareness of the Yellow Card Scheme among patients and members of the public

Our national omnibus survey of 2028 adults showed that although one-quarter of the respondents had experienced a side effect from a medicine at some time, fewer than 10% had heard of the YCS and only three people had used it. Those who were aware of the scheme were more likely to be from upper or middle socioeconomic groups and to have completed education post secondary school. Most of those who had experienced a side effect had told a HCP about it (85.5%).

Our questionnaire survey of 1362 patients who had reported an ADR via the YCS found that the most likely source of information about the scheme was from a community pharmacy (49.0%) or general practice (16.2%). Most patients reported the ADR(s) by post, with a third reporting online.

Our finding that a higher proportion of respondents reporting by post learned about the scheme from a pharmacy or GP surgery probably reflects the main source of the forms. From our telephone interviews we know that some patients had seen posters about the scheme or displays of Yellow Cards in pharmacies. Some respondents commented that there appeared to be limited awareness of direct patient reporting among health professionals. Given that responses to our omnibus survey indicated that most of those experiencing a drug side effect told their HCP about it, it is important that HCPs are aware of patients’ ability to report ADRs directly and that they pass this information on to patients.

There are likely to be benefits from increasing public awareness of the YCS. These include increasing the volume of patient reports and reducing delays in reporting. Suggestions from participants about how to increase awareness of the YCS included advertising campaigns and provision of information when patients are issued with their medicines, for example by providing a leaflet with the drugs or information about the scheme on PILs. There may be a need to target publicity among minority ethnic groups and less educated members of the public, given their under-representation among reporters.

The MHRA had advertising campaigns about direct reporting to the YCS in 2005 and 2007. Information is also available on the MHRA website,14 including a video about patient reporting that can also be found on YouTube (www.youtube.com/watch?v=jExeaOpDRbE). There is some evidence from our findings that some patients heard about the YCS through television and radio, whereas others came across the MHRA website when searching the internet for information about their suspected ADRs. These results highlight the need to use multiple media for disseminating information about the scheme. Changes in awareness of the YCS following new publicity campaigns could be monitored by repeating the omnibus survey.

Patient and public views and experiences of reporting to the Yellow Card Scheme

Our questionnaire showed that although, in response to a closed question, most patient reporters found that it at least ‘fairly easy’ to make a report, some thought that the reporting systems could be improved. More detailed suggestions for enhancements arose from our usability testing sessions. In addition, the omnibus survey provided support for continuing with the three different methods of patient reporting.

Success in addressing the objectives of the evaluation

Overall, we have been successful in addressing the main objectives of the evaluation and answering the research questions set out in Table 1 in Chapter 1 and in our original protocol (Appendix 23). We had only partial success in addressing our original plans for the following aspects of the work (more detailed explanations are given further below):

• In the descriptive analysis of reports from patients and HCPs there were few categories on which we could directly compare the seriousness and outcomes of suspected side effects.

• In the signal generation analysis:

  1. – It was difficult to assess whether new signals generated by patients were for events that are expected to be of more concern to patients than HCPs.

  2. – With only 2 years’ worth of data there was limited scope to make a valid judgement on whether patient reports generate new signals sooner.

• When selecting cases for the qualitative analysis of Yellow Card reports:

  1. – We identified and analysed 449 reports, but did not reach our original suggested target of 600 reports.

  2. – There were very few reports relating to OTC medicines or complementary therapies.

• For the focus groups and usability testing, originally we had planned to include patients who believed they may have experienced side effects from medications, but had not previously filled in a Yellow Card report. We decided, however, that this would make recruitment difficult and widened our selection criteria to include all adults.

Literature review and survey of international experiences of reporting schemes

Although the literature review was conducted systematically, the searches were not exhaustive. Although only studies published in English were included, it is unlikely that any major comparative study (of patient and HCP reports) has been omitted, because most countries with longer established patient reporting schemes have already published reports or studies in English.

The online version of a major review of patient reporting of suspected ADRs was published in 2006. 24 Our review has identified a number of papers published since then, including two large studies comparing patient and HCP reports from Denmark19 and the Netherlands. 30 Other studies we identified were of variable quality, with some being available only as conference abstracts.

The questionnaire was sent to 47 different countries using the pharmacovigilance contacts listed by the WHO Uppsala Monitoring Centre website (www.WHO-UMC.org). Nevertheless, there was a low response rate, despite a reminder e-mail being sent. From the responses received, however, and from the ‘15-country survey and literature review’ on direct patient reporting of ADRs,15 as well as the WHO survey of pharmacovigilance schemes in low- and middle-income countries,16 it is clear that patient reporting has been introduced to an increasing number of national pharmacovigilance schemes.

There is no single source of information showing whether a country has patient spontaneous ADR reporting. To allow for sharing of information and future comparisons, it would be helpful to have a centralised resource of all national patient reporting schemes, similar to that held by the WHO Uppsala Monitoring Centre. Nevertheless, when comparing information on patient reporting from different countries it needs to be recognised that there may be differences in the types of problems that can be reported, the available methods of reporting, the types of data collected and other context specific factors.

Analysis of reports of suspected adverse drug reactions from the UK Yellow Card Scheme

This is the first study in the UK to investigate large numbers of reports of suspected ADRs from patients and HCPs. Our descriptive analysis allows these reports to be compared with large retrospective analyses of other national reporting schemes. In addition, we undertook signal generation analyses, which enabled us to answer a number of questions raised in the literature regarding the pharmacovigilance impact of patient reports. Furthermore, our qualitative analyses of extracts from patient and HCP Yellow Card reports have provided useful information on the richness of patient reports, how patients describe ADRs, the impact of ADRs on patients’ lives, the clinical significance of patient reports and likely causality.

One strength of our study has been collaboration with the MHRA. This enabled us to obtain and analyse all ADRs reported over 2-year period, giving access to a database of > 26,000 reports including just over 5000 from patients. As a result, our study was larger than previously published major studies from Denmark19 and the Netherlands. 30 This has also provided us with a sizeable database for signal generation analysis and for the sampling of reports for qualitative analyses.

We worked carefully with the MHRA to ensure that, wherever possible, any anomalies in the database were corrected. This meant that, as far as we can tell, we had virtually complete information on key variables, such as the drugs and reactions reported by patients and HCPs. Nevertheless, many reporters failed to complete all fields and so data were missing for some variables, including indication for the drug, reaction outcome and time to report a reaction. Gender and age was missing for 3.6% and 10%, respectively, of all reports.

We noted several differences between the patient and HCP Yellow Cards. The most notable difference related to the perceived seriousness of the suspected reaction; only three of the six options were comparable. Nevertheless, we were able to directly compare the seriousness of suspected ADRs reported by patients and HCPs using the MHRA classification of seriousness based on the MedDRA dictionary.

For reasons outlined in Chapter 3, we were not able to calculate the frequency of duplicate reports in the database of patient and HCP reports. We found some evidence to suggest duplicate reports, but these appear to be uncommon; probably not at a level to invalidate the pharmacovigilance assessments.

The main limitation of our disproportionality analysis, was that the data were accumulated over only a 2-year period, soon after the launch of patient reporting. Reports from the pharmaceutical industry (normally around 40% of all reports received by MHRA each year) were also excluded, as the aim of the project was to compare direct reporting by patients and HCPs. Furthermore, as data from HCP reports received in the YCS before the study period were not included in our data set, it was not possible to analyse SDRs generated over many years. In addition, this meant that we could not investigate properly whether the introduction of patient reporting leads to earlier or later detection of signals.

In the database we used for our studies, patient reports comprised around one-fifth of all reports. In reality, the proportion of patient reports in the YCS database will be much smaller, as the data have accumulated since 1964. Hence, the contribution of patient reports to quantitative signal generation is likely to be smaller, although it will increase as patient reports accumulate.

We wished to investigate the relative contribution of patient reports to signal generation, and so we wanted every patient report to contribute as much information as possible, especially as it was clear that patient reports were different to HCP reports in terms of suspect drugs and reactions reported. The majority of the signals generated involved reactions that are not classed as ‘serious’ by the MHRA. In practice, these reactions would not routinely be prioritised by their signal detection system (unless they involved certain events on the MHRA’s ‘alert’ list or those involving a child). Nevertheless, we have shown the contribution that patient reports made to signals for reactions that are classed as ‘serious’ by MHRA. It is important to note that the signals ‘lost’ or ‘gained’ in this study were intended as examples to illustrate the potential effect of adding patient reporting to the spontaneous reporting database. It is difficult to say whether one ‘signal’ is more important than another. This would depend on a number of factors such as the level of use/exposure of the drug in the population, the type of population using the drug, the drug’s indication, the tolerability of the reaction. A detailed impact analysis for each of the drug–reaction pairs in question would be required in practice, but was beyond the scope of this project.

Another possible limitation of our signal generation analysis was that it was based only on the proportional reporting ratio method (see Chapter 5 for further details). There are several methods for disproportionality analysis and different thresholds for filtering potentially important reactions. In order to maximise the size of the data sets in this study we analysed drugs and vaccines together, rather than analysing these separately as the MHRA would do. Ideally, our results should be replicated by the MHRA using source data.

In our main qualitative analysis, we included many patient and HCP Yellow Card reports, looking at a range of purposively selected drug–ADR pairs. There was, however, sometimes an imbalance in the number of reports made by patients and HCPs available for comparison. We had planned to analyse around 600 reports in total, but the sample generated was smaller than anticipated (270 patient reports and 179 HCP reports) because although the database appeared to contain a very large number of suspect drugs and reactions that met our sampling strategy criteria, specific cases within the database were replicated many times if multiple drugs and multiple reactions had been recorded. Having completed the time-consuming process of selecting reports (which took several months) we did not have time to go back and select additional cases. Nevertheless, having undertaken the qualitative analysis of 449 reports, we judge that it is unlikely that our findings would have changed substantially with 600 reports.

We focused our attention on the medicines most commonly reported by patients and those that at the time of the study were classified by the MHRA as ‘black triangle’ drugs. For these groups of drugs we had relatively large amounts of data. For reports of medicines that were purchased OTC and complementary medicines, we had to focus on patient reports only because there were so few reports from HCPs; even then, the numbers of reports were few.

Our content analysis of reports was necessarily subjective and could have been biased because the researcher knew whether the reports were from patients or HCPs. Nevertheless, the work was checked carefully (by AA) throughout the process and very few misinterpretations, or inaccuracies in data entry, were apparent. For some categories, such as text indicating the impact of the suspected ADR on the patient, the contrast between patient and HCP reports was so marked that they clearly represent genuine differences. For other categories, such as text indicating temporal relationships between medicines and suspect ADRs, differences were less marked and did not take account of information available in other fields on the Yellow Card report database.

When considering different types of report, there was some concern among the research team that the meaning of patients’ concerns may be changed when telephone reports are transcribed at the MHRA. Reassuringly, the qualitative analysis indicated that telephone reports were not very different from internet and paper reports, often including information that was assumed to be direct patient quotes.

Our in-depth qualitative analysis of reports was, by its nature, subjective and could have been biased because we knew who the reporter was. Nevertheless, we checked the data carefully, rereading the reports several times and sought out divergent cases to counter possible biases.

Questionnaire of patient reporters and telephone interviews

Our questionnaire was the first large national study to follow-up patients or their representatives reporting via the YCS. The response rate of 68% was reasonably high; probably helped by distribution of the questionnaires directly by the MHRA. Although, we had no information on the characteristics of non-responders, the age and gender of respondents to our questionnaire was similar to those of all patient reporters in the MHRA database used for the retrospective analysis of Yellow Card reports. Therefore, it is likely that the findings are generalisable to all patient reporters at the time of our study. The high number of responses to open questions shows that reporters were willing to express their views in detail; this information has given us valuable insights into many aspects of patient reporters’ experiences of the YCS.

These more in-depth comments were supplemented by telephone interviews, which allowed us to further explore issues regarding ease of reporting, motivations for reporting and promotion of the scheme. Of the 27 interviews, we managed to interview only two people who had reported by telephone and two who had reported via the internet, so our results are biased towards those who reported using the paper Yellow Card. This imbalance occurred because of difficulties arranging interviews with patients who had reported using the internet. In addition, because of circumstances beyond our control, the data were analysed by a third person (CA) who had not done the interviews. This may have led to misinterpretation of data, although the researcher listened to the tapes and repeatedly reviewed the transcripts. The proportion of women interviewed (70%) was slightly higher than the proportion of women in the questionnaire survey (66.8%) and the median age of interviewees was also slightly higher (60 years vs 56.5 years). It should be noted, that the interviewees are likely to have been a highly motivated group of people who not only completed a Yellow Card report and a questionnaire, but also who volunteered to take part in the interviews; not surprisingly, their views about ease of reporting were different from those in the usability study.

Usability study and focus groups

The participants recruited for the focus groups and usability testing covered a range of demographic characteristics, with a similar proportion of women (67.5%) to that found in our other studies, but they were slightly older and there was a higher proportion from minority ethnic groups (15%).

The scenarios for the usability tests were ‘hypothetical’. This may have introduced some artificiality as the participants may not have been familiar with the details contained within these scenarios. It might also have lengthened the process of completing the forms by any of the three methods. Conversely, the process may have been shortened as dates for starting and finishing the medicines were included in the scenarios, whereas in real life participants would have had to recall this information, look it up or ask someone else.

For practical reasons we recruited from the Nottingham area and decided to widen our selection criteria for the focus groups and usability testing to all adults. Even so, it was a challenge to recruit 40 participants and it is unlikely that we would have achieved this number if we had restricted entry criteria to people who had experienced ADRs. Although it is possible that we would have obtained different findings using members of the public from other parts of the country, we doubt whether any important additional issues would have been detected from the usability assessments.

Nevertheless, some of the findings from this part of the study are different from our studies of patient reporters, and this may, in part, have been owing to the sampling. For example, whereas some of our focus group participants say that they would have ‘given up’ trying to report because of difficulties they had experienced with the online reporting system, it is possible that people who had suffered a suspected ADR would have been more persistent.

National omnibus survey

The sample of > 2000 members of the British public was broadly representative of the general population, although they were different to patient reporters to the YCS. The response rate was very low, a common issue with this type of survey. We were aware of this issue at the time of planning our study, but judged that the advantages of being able to conduct a national survey at relatively low-cost outweighed the difficulties of using more rigorous techniques. Although it is possible that another type of survey might have revealed some different findings to those observed in the omnibus sample, we doubt that other methods would indicate a substantially higher level of public awareness of the YCS.

One advantage of the omnibus survey was that it could be repeated at regular intervals and at low cost to track public awareness.

Implications for patient reporting

In the authors’ opinion, the following approaches may help to improve the timeliness and value of patient reporting for pharmacovigilance; increase the number of reports from patients; and improve patient experiences of reporting:

• increase publicity for patient reporting by:

  1. – advertising campaigns targeted at the public and HCPs

  2. – encouraging proactive efforts by HCPs, particularly in community pharmacy and general practice, to publicise and promote patient reporting

  3. – providing information on patient reporting at the point of dispensing, ideally by incorporating this information within PILs

• provide further guidance to reporters on what information to report (this may help strengthen the usefulness of both patient and HCP reports)

• increase patient awareness of medicines where the MHRA is undertaking intensive monitoring, for example by highlighting this on PILs

• change the design of paper reports and the online reporting system, in line with suggestions made in this report, and increase the number of hours during which telephone reports can be made

• provide general feedback to patient reporters on what the MHRA does with reports

• explore possibilities for providing specific feedback to patients in relation to the medicines and suspected ADRs that they report.

To aid future comparisons of reports submitted by patients and HCPs, it important that similar information is collected from both groups, particularly with respect to categories of seriousness.

According to patient accounts, some HCPs seem to be unaware that patients can submit their own ADR reports, and some appear to be dismissive of patients who report suspected ADRs. Education at undergraduate and postgraduate level might help address these issues.

Recommendations for research

In order of priority these are investigating:

1. the pharmacovigilance impact of patient reporting in a long-term study, including the identification and tracking of regulatory action taken as a result of the contribution of patient reports, such as the addition of new ADRs to PILs, SPCs and the BNF

2. the optimum approach to signal generation analysis of patient and HCP reports. It is important to establish, in practice, the true extent to which important signals may be identified or missed by combining data from patient and HCP reports for data mining purposes and the feasibility of analysing the data separately by reporter group, in addition to analysing all reports combined

3. the burden of ADRs in terms of impact on patients’ lives, and evaluating the extent to which patients’ views and experiences of the seriousness of ADRs concur with those of regulatory bodies such as the MHRA

4. the knowledge and attitudes of HCPs patient reporting of ADRs, and evaluating approaches aimed at addressing any learning needs identified

5. the value of using patient reports of ADRs to help other patients and HCPs who are seeking information on patient experiences of ADRs

6. the impact of increasing publicity and/or enhancements to reporting systems on the numbers and types of Yellow Card reports from patients. This would be especially important if patients were to be provided with information on drugs undergoing intense monitoring, for example black triangle drugs.