Predictive clinico-pathological features derived from systematic autopsy examination of patients who die with A/H1N1 influenza infection in the UK 2009-2010 pandemic

Article history

This themed issue of the Health Technology Assessment journal series contains a collection of research commissioned by the NIHR as part of the Department of Health’s (DH) response to the H1N1 swine flu pandemic. The NIHR through the NIHR Evaluation Trials and Studies Coordinating Centre (NETSCC) commissioned a number of research projects looking into the treatment and management of H1N1 influenza. NETSCC managed the pandemic flu research over a very short timescale in two ways. Firstly, it responded to urgent national research priority areas identified by the Scientific Advisory Group in Emergencies (SAGE). Secondly, a call for research proposals to inform policy and patient care in the current influenza pandemic was issued in June 2009. All research proposals went through a process of academic peer review by clinicians and methodologists as well as being reviewed by a specially convened NIHR Flu Commissioning Board.

Declared competing interests of authors

none

Copyright statement

© 2010 Queen’s Printer and Controller of HMSO. This journal is a member of and subscribes to the principles of the Committee on Publication Ethics (COPE) (http://www.publicationethics.org/). This journal may be freely reproduced for the purposes of private research and study and may be included in professional journals provided that suitable acknowledgement is made and the reproduction is not associated with any form of advertising. Applications for commercial reproduction should be addressed to: NETSCC, Health Technology Assessment, Alpha House, University of Southampton Science Park, Southampton SO16 7NS, UK.

2010 Queen’s Printer and Controller of HMSO

Background

From April 2009 to January 2010, the pandemic of A/H1N1 influenza affected the UK. There were > 30,000 infections and 457 deaths (all ages). Reports from other countries had indicated that certain comorbidities were associated with a higher risk of death from H1N1 infection, and there was a need to identify these factors in the UK population as knowledge of them could lead to improved treatment in the current epidemic and reduced mortality in future epidemics.

Objectives

To gather all the available clinical pathology information from autopsies performed on patients dying with known or suspected influenza A/H1N1 infection, across the UK. To evaluate comorbidities present in these deceased patients; correlate them with the H1N1-related pathology and treatment-associated pathology, determine their relative contributions and estimate the significant features associated with death.

Methods

To obtain the autopsy reports, standard request letters were sent by e-mail to all histopathologists in the UK on the Royal College of Pathologists list, all the coroners’ jurisdictions in England, Wales and Northern Ireland, and to procurators fiscal in Scotland. The letters asked for autopsy reports of the autopsied deceased who included: those with H1N1 infection, proven before or after death, and those in whom swine flu was unproven but most likely to have been present; those in whom H1N1 was a minor pathology, as well as those in whom it was the immediate cause of death; those whose cause of death mentioned ‘swine flu’, ‘swine influenza’ or ‘H1N1 infection’; and those of any age from infancy to old age.

Results

Sixty-eight autopsy reports were received: 19 children (0–15 years) and 49 adults (16 + years). All but two autopsies were medico-legal, and only two (3% of the total) were consented. This sample thus represents 15% of the known 457 deaths from H1N1. Median age for children at death was 6 years, for adults it was 41 years. Deaths in children were associated with congenital diseases (47%, 9/19), particularly of the heart and central nervous system. The autopsied children were not obese. Death in adults were associated with pregnancy (three cases in the study, but nationally 12/457 H1N1-associated deaths were noted), obesity (50% of adults had a body mass index ≥ 30 kg/m²) and chronic respiratory disease (12%, 6/49 adults). Diabetes did not emerge as a risk factor for death, but learning difficulties did. Nearly all the deaths (94%, 64/68) were a consequence of H1N1 infection in the respiratory tract. In more than one-third (41%, 28/68) of the deaths, bacterial secondary infection was the significant complication; the pneumococcus was the most common agent identified (25%, 7/28).

Limitations

This review is an incomplete medical study of what happened during the epidemic, and the small sample number (68 reports from 457 deaths) limits further speculation. We have no true measure of whether the cases selected for autopsy are representative of the total deaths in terms of pathology and comorbidities.

Conclusions

The major comorbidities associated with death from H1N1 infection were obesity, chronic respiratory disease and pregnancy. Young age at death was confirmed. Congenital disease in children and learning difficulties in adults were also important, but diabetes was not. This methodology of gathering data for research has potential for use in other public health questions, but is dependent on the co-operation of the medico-legal services. These results reinforce the need to enquire further into the pathogenesis of severe and fatal H1N1 disease, and the circumstances of clinical presentation and rapid evaluation in a time of epidemic influenza.

Funding

The National Institute for Health Research Health Technology Assessment programme.

Research objectives

• To rapidly gather all the available clinical pathology information from autopsies performed on patients – adults and children – dying with known or suspected influenza A/H1N1 infection, across the UK.

• To evaluate comorbidities present in these deceased patients; correlate them with the H1N1-related pathology and treatment-associated pathology, determine their relative contributions, and estimate the significant features associated with death; and identify pre-mortem diagnostic confusions contributing to death.

• To evaluate the effects of treatment in these patients, including any adverse events following mass H1N1 vaccination.

• To post this information, anonymised, in real time, on a website available to health-care workers, followed by formal peer-reviewed publication.

In the event, the last two objectives were dropped. While the approval for the project was being sought (it came through at the end of November), it became evident that the numbers of deaths was not going to be as large as feared initially; thus real-time posting of information was not a priority. Secondly, during the epidemic and after, no evidence of any AEFI emerged, nor any side effects of the two anti-influenza chemotherapies used in the UK [oseltamivir (Tamiflu^®, Roche) and zanamivir (Relenza^®, GlaxoSmithKline)].

It should also be noted that, despite the anecdotal reports in newspapers of such cases, this study could not identify and analyse deaths where the presenting illness had been presumed to be H1N1 infection, but the actual clinical pathology turned out to be something different – and perhaps where the patient might have survived had H1N1 not distracted the carers. By definition, finding these cases systematically would have posed difficulties, and this was not in the original remit of the study.

Also the study did not include review of the histopathology samples taken at autopsy – which would have presented logistic issues – only review of the written autopsy report.

Time line for H1N1 morbidity and mortality in the UK [Health Protection Agency (HPA) data]:

• first case April 2009

• total reported infections > 30,000

• total deaths in UK = 457

• epidemic ends January 2010.

How many autopsies would there be?

The proportion of deaths with H1N1 infection that would have an autopsy was, of course, unknowable until the epidemic matured. It was assumed, from previous discussions on emergency planning since 2006 [Emergency Preparedness Clinical Liaison Advisory Group within the Department of Health, the author is the representative from the Royal College of Pathologists (RCPath)] that, in the initial phase of attack, there would be many autopsies carried out while clinicians, virologists, pathologists and coroners familiarised themselves with the clinical pathology and other issues around evaluating persons suspected or known to have died from H1N1 infection. Thereafter, the autopsy rate would lessen as the perceived need declined – through more efficient case detection in life. It was also assumed that the core medico-legal need for coronial autopsy, i.e. identification and inclusion/exclusion of unnatural death (the main requirement of the medico-legal autopsy), would continue.

Secondly, it was not known what proportion of such autopsies would be medico-legal versus hospital/consented. This affected how the autopsy reports would be obtained, as those from consented autopsies would come under NHS standards for ethics and confidentiality, whereas those from medico-legal autopsies do not, being carried out independently of the NHS and Department of Health.

A guesstimate that 10% of H1N1 + deaths would be autopsied was made. It should be noted that in England and Wales, 22% of all deaths are subject to a coronial autopsy. 1 It was anticipated that because pandemic H1N1 would be a novel infection, categorised initially as Hazard Group 3 on the 1–4 scale of risk by the Advisory Committee on Dangerous Pathogens, pathologists and coroners might be reluctant to commission autopsies unless strictly necessary. Best practice guidelines on performing H1N1 + autopsies were promulgated in advance, on the websites of the RCPath (www.rcpath.org) and the Association of Anatomical Pathology Technologists (www.aaptuk.org).

Collecting autopsy reports

Following consultation with the:

• National Institute of Health Research Health Technology Assessment programme

• RCPath

• Coroner Society of England and Wales (CSEW)

• Crown Office and Procurator Fiscal Service (COPFS)

• Centre for Maternal and Child Enquiries (CMACE)

• HPA Swine Flu Unit

• National Research Ethics Service (NRES),

a sweep for autopsy reports relating to H1N1 deaths took place in the form of an e-mail to all active histopathologists in the UK (England, Wales, Northern Ireland, Scotland) to discover whether such autopsies were actually being performed, and how many were in children versus adults and were medico-legal versus consented. This produced a very positive response, encouraging the subsequent steps:

1. A later e-mail to all active histopathologists in the UK – with the letter seen in Appendix 1 attached – to request copies of the autopsy reports by post, fax or e-mail.

2. The CSEW sent a similar e-mail (see Appendix 2) and a hard copy to all the coroner jurisdictions in England, Wales and Northern Ireland, with an official endorsement to collaborate by providing direct to the author the copies of the autopsy reports by post, fax or e-mail.

3. The CSEW also forwarded the e-mail to the COPFS, who also forwarded it to the forensic pathologists on their list, so that procurators fiscal in Scotland could collaborate.

4. Discussions with the CMACE which was, separately, undertaking a review of H1N1-related maternal deaths; this provided three autopsy reports, anonymised.

5. Discussion with the HPA Swine Flu Unit at Colindale also took place to identify possibilities of case-sharing of reported deaths of/with H1N1 infection, and possible information sharing about those who had had an autopsy. These conversations proved fruitless as the HPA did not subsequently respond, and nothing more happened with the HPA for study data collection (however, see Chapter 3, Secondary bacterial infection of the lung).

Note: to encourage maximal participation and return of reports, the case definition for an autopsy report to be submitted was deliberately limited to one of:

• H1N1 infection was mentioned in any part (I or II) of the cause of death sequence by the pathologist

• H1N1 was proven to be present in pre- and/or post-mortem material.

In the event, all the submitted reports were on subjects proven, by standard virology polymerase chain reaction analysis, to have been infected with H1N1.

The study was not intended to, and did not, investigate those cases where H1N1 had been clinically guesstimated in life, without virology confirmation, where the autopsy revealed different pathologies.

Confidentiality

Much time was spent during the study development period discussing how to manage patient confidentiality and anonymity. However, this aspect did not impede the inflow of autopsy reports. Only 2/68 (3%) of cases were hospital/consented autopsies (one adult, one child), for which it is an NRES requirement that the autopsy reports be submitted anonymised.

Nine of the 48 adult medico-legal case autopsy reports arrived anonymised (mostly from Scotland and Northern Ireland); all the others and all the child medico-legal case reports came with names and addresses indicated. The single consented child autopsy report was anonymised; that of the single adult consented case was not anonymous as the author performed the autopsy.

The causes of death in this cohort of patients

The causes of death were derived from the autopsy pathology. The data were from combined gross and histopathology examinations, plus the results of further bacterial and other investigations (all the cases had or had had H1N1 infection proven before and/or after death). These data were considered alongside the clinical and H1N1 virology data contained in the autopsy reports.

Because there is no universal standard for medico-legal autopsy practice, report formats, data and quality content, the results of this analysis are necessarily biased. Case definitions for primary H1N1 lung disease and secondary bacterial infection are partly subjective, but the best that can be obtained from the data. Reporting of comorbidities is also a minimum data set as, if the pathologist deemed a particular condition not relevant or interesting, or did not notice it at all, it would not be in the report.

Remarkably, the histology taking rate in this set of patients was high. For children, who are traditionally investigated fully, it was 100%. For the adult cases, 86% of the autopsies included histopathology (lung in all cases, plus other organs as deemed appropriate by the pathologist in consultation with the coroner). This contrasts with the 19% histology taking rate for non-homicide coronial autopsies found in the NCEPOD report into coronial autopsies, which sampled deaths in 2005. 2

Evidently, the H1N1 epidemic in the UK was new, and pathologists and coroners believed it important to be more certain as to cause of death than, in other circumstances, they normally would be.

In addition, many cases (denominator unclear) had tissue samples tested in bacteriology and virology laboratories for co-infections – particularly bacterial lung infection. This enables presentation of the actual causes of death in four categories (Table 3):

1. H1N1-related influenza pneumonitis without secondary infection

2. probable H1N1 influenza pneumonitis, with later death under intensive care, where the autopsy necessarily will not be able to distinguish H1N1-only pathology from H1N1 plus another infection

3. secondary bacterial lung infection following from H1N1 infection that has damaged the airways and lung parenchyma

4. H1N1 infection around the time of death but incidental to the death.

Thus, H1N1 infection was responsible for the great majority (64/68 = 94%) of deaths in the cohort studied. About half of all the patients studied died of direct H1N1 infection of the airways and lung parenchyma; most of the rest had bacterial secondary infection in the lung.

In the autopsy reports, mention of specific anti-H1N1 chemotherapy was erratic. Thus there is no consideration here of possible toxic side effects, and no calculation of any clinical benefit provided by these therapies in persons with H1N1 infection.

Primary H1N1 influenza pneumonitis

The virus directly damages the lower airways and alveoli in a characteristic fashion, well-described in pandemic influenza. The two main lesions are necrotising bronchiolitis and acute lung injury (ALI; also termed diffuse alveolar damage). These can be identified only where there is histopathology of the lung, not from naked eye inspection.

From the 58 autopsy reports where these lesions were positively identified, or could be inferred from the microscopic descriptions, nine (16%) showed necrotising bronchiolitis, and 38 (66%) showed ALI. All the patients with necrotising bronchiolitis also had ALI, as has been noted in the Brazilian H1N1 autopsy study. 3 The third lung pathology noted in Brazil, alveolar haemorrhage in association with ALI, could not be quantitatively assessed from the UK autopsy reports. The ALI prevalence in UK H1N1 cases is necessarily a minimum estimate for the reasons indicated above; a proportion was associated with secondary bacterial infection.

Secondary bacterial infection of the lung

The number of patients with pneumonia/bronchopneumonia in addition to H1N1 infection was 28/68 (41%). The specific infections identified result from culture of lung and/or blood in life or at autopsy; the aspergillus infections were identified histologically. The infections are indicated in Table 4.

The total number of Streptococcus pneumoniae infections was six definite, plus one possible (gram-positive cocci on histology), giving a proportion of 7/68 total deaths (10%) or 25% of the deaths with secondary bacterial infection. Group A streptococcal infections numbered four, and there were two each with Staphylococcus aureus and Haemophilus influenzae. All of these specific infections were distributed across children and adults.

These prevalences are minimum estimates given the varied degrees of the investigations at autopsy and of the standards of autopsy reporting. But they do indicate a very significant burden of secondary bacterial pathology on top of H1N1 infection of the airways and lung. Other studies have noted similar proportions of bacterial co-infections in their autopsy series: 38% in Brazil,3 55% in New York, NY, USA,4 26%–29% in the US Centers for Disease Control’s study of referral autopsy case material. 5,6 But each of these studies, including this UK study, has used different methodologies, so they are not strictly comparable.

The UK data do address and inform on a specific question that was raised by the Department of Health in preparation for future H1N1 epidemics: ‘What proportion of deaths in H1N1-infected patients might be preventable by advance vaccination (“Pneumovax” against the pneumococcus) and by optimum antibiotic treatment during illness?’. The question was brought to the author by Dr Martin Utley (University College London conducting work on behalf of the Department of Health’s Health Protection Analytical Team). This study indicates that the pneumococcus was the most common organism identified in children and adults.

Although positively identified in only six cases, the possibility has not been excluded that the pneumococcus was the agent responsible for all the other bacterial infections for which a different specific agent was not identified. Thus, potentially, the pneumococcus could be relevant in up to half (14/28) of all deaths where secondary infection of the lung occurred.

Other comorbidities in H1N1-associated death

Assessment of comorbidities was the prime aim of the UK autopsy study. Bacterial co-infection in the lung has been considered already. The other comorbidities identified in children (up to 15 years) were broadly different from those noted in adults.

Learning difficulties and mental health problems

These have not been noted in other series. The high prevalence of congenital CNS conditions among the dead children, plus the clinical information in the adult autopsy reports concerning ‘learning difficulties’ and psychiatric conditions (see Table 6) suggest that this group of patients is vulnerable to severe H1N1 infection. The mechanisms might include (a) an associated impairment of immune defences to infection, and (b) problems in accessing health care.

Specific other pathologies

A few other specific conditions are worth mentioning as arising from the data accrued during this autopsy report of H1N1-associated deaths.

Future research questions

These results reinforce the need to enquire further into the pathogenesis of severe and fatal H1N1 disease, and the circumstances of clinical presentation and rapid evaluation in a time of epidemic influenza.

• Why are disabled children, pregnant women and obese adults particularly at risk of death?

• How effective are the specific antiviral drugs given to patients?

• When should they be optimally given to presenting patients?

• Given the importance of secondary pneumococcal lung infection, what better preventive measures can be instituted?

• How better can patients over-diagnosed as H1N1 be systematically identified, so that diagnostic protocols can be refined and thus fatalities remediable be reduced?

Disclaimers

The authors have been wholly responsible for all data collection, analysis and interpretation, and for writing up their work. The HTA editors and publisher have tried to ensure the accuracy of the authors’ report and would like to thank the referees for their constructive comments on the draft document. However, they do not accept liability for damages or losses arising from material published in this report. The views expressed in this publication are those of the authors and not necessarily those of the NIHR or the Department of Health.

[full title: NIHR project 09/84/18 – Predictive clinicopathological features derived from systematic autopsy examination of patients who die with A/H1N1 (pandemic flu) infection]

[full title: NIHR project 09/84/18 – Predictive clinicopathological features derived from systematic autopsy examination of patients who die with A/H1N1 (pandemic flu) infection]