Systematic review of the links between human resource management practices and performance

Focus on high involvement and skill acquisition

The Department of Health (www.dh.gov.uk/en/index.htm) repeatedly singles out the crucial role of skill development, involvement, team working and morale to the modernisation of the NHS. In the words of the Department of Health’s website, ‘Successful human resources management improves efficiency through a culture that supports and develops its staff, allowing the workforce to share in the organisation’s objectives’ (www.dh.gov.uk/en/Policyandguidance/Humanresourcesandtraining/Buildingpeoplemanagementskills/index.htm).

The evidence offered to support the Department’s view that HRM is vital to the NHS is the finding that patient mortality rates improve in acute hospitals where an HRM director is a member of the Trust Board. In this, the Department of Health is mirroring a more general emphasis on employee development and involvement as key to maximising the human resource (HR) contribution to strategic objectives of organisations. This high-involvement management and human capital approach to HRM, sometimes referred to as ‘the HRM approach’, has been at the centre of the literature for the past two decades. For consistency, we refer to it as ‘the high-involvement management approach’, while acknowledging that not all of those who are referring to modern HRM necessarily place empowerment at its core, but rather emphasise the development of human skills and knowledge. The term ‘HRM’ is used whenever a potentially more all-embracing concept of personnel management is being considered.

The past 20 years have seen increasing attention given to both the general notion of HRM and the more specific high-involvement management approach as a major contributor to organisation performance, even in capital or technologically intensive industries. High-involvement management is assumed to contribute both in its own right and as a support for such other modern management practices as total quality management (TQM) and just-in-time (JIT). So, in the health context, high-involvement management adds to the human capital of the workforce and helps support new modes of organisation, technology and patient delivery. Its importance derives from an assumption that some forms of personnel management – and, particularly in the current context, approaches founded on involvement and development – have greater effects on the development, skills and motivation of employees than others.

Much of the attention given to the high-involvement or commitment HRM system has been advocatory. Contrasting it with a traditional control approach (sometimes known as Taylorist or Fordist), academics1–3 have championed high-involvement management as a progressive form of management that can remove past restrictions to both economic efficiency and the achievement of high-quality performance, thereby sharpening the competitive edge of Western organisations and improving public sector delivery. They contend that there is a unique set of practices or approaches to management that will, regardless of the context, outperform all others. Within the strategy literature, the resource-based theory of the firm4–6 highlights how human resources, and the knowledge embodied in them, may be decisive for long-term competitive advantage. In the operations management literature7–10 the emphasis is on how HRM practices and the knowledge and skills of workers are decisive in exploiting lean production11 and other modern management methods (e.g. see Dean and Bowen12 on TQM).

Research evidence

The last decade has witnessed a stream of research assessing high-involvement management and its associated concepts. The basic hypothesis underlying this work is that high-involvement HRM systems will have positive effects on organisational performance. As academics have sought to move beyond advocacy of high-involvement systems, they have focused on testing this hypothesis. A spate of research studies in the 1990s were motivated by this, with many surmising that high-involvement HRM systems do perform best. 13–20 Several overviews21,22 concluded that, on the basis of the first few studies, the universalistic hypothesis is supported. The increasing use of the term ‘high-performance’ model to describe the set of practices may imply that the matter is indeed settled, much as a drug might be named by the disease it is known to cure.

The initial overviews of these studies22,23 tended to present them in a homogeneous way, taking for granted that they are studying the same phenomenon and that the results are broadly the same, concluding they had successfully demonstrated the link between the human capital approach and key organisational outcomes. These reviews have led some to conclude that the main issue now is to explain the link between high-involvement management and performance, i.e. to assess the mechanisms between them or get inside the ‘black box’ between the HRM system and outcomes. A second issue might be the extent to which the link will be found in all contexts, and particularly in service industries such as health care. The majority of the early studies were in manufacturing, and it is significant that one of the few early studies in services24 found evidence of a contingent relationship between the human capital approach and performance, i.e. that the impact of human capital will vary between contexts. In the past few years, more studies have either been conducted singly in the service sector or included that sector within the scope of the study,15,25–27 and the results have been mixed. Studies specifically on health care are emerging, for example West et al. 28 have investigated the link between HRM and patient mortality.

Assessing the evidence

Notwithstanding the above, an increasing number of writers question whether the portrayal of the studies as providing conclusive evidence for a universal link between HRM and performance is premature, and that there is a clear need to delve more deeply – both into the studies and into the reality of high-involvement management and its links to other aspects and methods of management. There are a number of reasons for extending the debate.

First, the results of the various studies are neither as clear-cut nor as uniform as some have concluded (see Wood,29 particularly table 1 for a summary of the main results; and Wall and Wood30,31). Within studies, there is unevenness in the findings between performance measures; while some results point to universal effects, others do not. Moreover, in many studies only the universal hypothesis has been tested so one cannot rule out the contingency argument that the effects of high-involvement management on performance are contingent on a third factor, either a dimension of the context or the strategy of the organisation, even when a positive link between high-performance management and performance has been found.

Second, it is not always clear whether high involvement or performance management is being defined simply as the combined use of best practice in each domain of human resources, as a synergistic set of practices, or by a more fundamental managerial orientation. 32 In fact, the discussion so far has begged the question of how we define and identify a system. Most studies have defined an HRM system a priori, often with little theoretical justification, as the emphasis has been on testing the high involvement–performance link, rather than first investigating the relationship between practices or the nature of any systems. Researchers have measured HRM on the basis of differentiating organisations by their usage of a set of practices, the precise means varying between studies. Then they correlate these differences with performance measures. The problem of simply aggregating practices to provide an overall measure of HRM is that each practice is treated implicitly as if it were equally important. With only a few exceptions,33,34 studies have provided insufficient information about which individual or subset of practices has the strongest effects and which may be marginal or even irrelevant. In one case13 where this was done, and it was shown that only some of the practices affect performance, and, moreover, that their strength varied between industrial context, though the authors still framed their conclusions in terms of an overall (high-performance work) system yielding superior performance. Aggregating practices does not allow one to test whether the effects of one practice are enhanced by the use of another, i.e. whether they are synergistically related, and begs the question of whether they tend to be used together. 32,35

Third, there is considerable variability across studies in the practices included either in the model of high-involvement management or the set of practices used to test its link to performance. Some of the differences are terminological, perhaps reflecting disciplinary biases or a quest to differentiate one’s wares. But a core difference within the literature can be identified. On the one hand are those authors who see the core of high-involvement management as changes in work organisation, job design and employee involvement methods (particularly idea capturing), with the other practices acting as supports to help motivate and equip people to work in a new, more flexible and proactive way. On the other hand are those who do not prioritise the task system. They may then concentrate more on skill acquisition, following an emphasis on the resource-based theory of the firm, according to which genuine competitive advantage reflects an organisation having a unique capability or set of resources. Therefore, primacy is placed on the competencies of individuals, rather than whether they are empowered. Alternatively, they may treat high-involvement management as entailing the use of the whole gamut of sophisticated personnel management methods, and conceive them as primarily operating through people’s commitment or sense of whether they are being treated fairly and consistent with their psychological contract. The different approaches need not affect greatly the practices researchers include in their studies, but the inclusion of job design and work organisation practices is one major source of diversity between the studies, as Wood and Wall31 show. The other major difference is incentive systems, with some studies treating them as part of high-involvement management or at least performance management, whereas others do not.

Fourth, there has been insufficient empirical exploration of the mechanisms or intervening variables that might explain any link between HRM and performance. Of course it must be acknowledged that there is little point in investing resources in exploring the reasons for a link until it has been established. Yet, labels such as ‘high commitment’ and ‘involvement’ imply certain mechanisms that are concerned with the extent of commitment or engagement on the part of the work force, which is mirrored in the importance given to morale in statements about the modernisation of health care, and these could be explored more than they have been. Much of the theoretical justification that precedes the empirical studies also implies that skills, knowledge and learning might have an independent effect on performance as important as employees’ attitudes.

Some studies have included intervening variables, which have either centred on intermediate HR outcomes (e.g. labour turnover in Huselid18) or on commitment and satisfaction. 36,37 Several authors hint at, but do not explore in depth, other mechanisms; for example, Guest and Conway38 imply that procedural justice may be important. In addition to morale and staff development, an important emphasis in statements on the modernisation of health care is what might be termed the strategic integration of individuals into the organisation, so that they both understand and share the goals that achieving patient care entails and, consequently, orientate their behaviour towards common visions. Studies outside the narrow confines of the HRM studies, particularly building on the job redesign and lean production literature, point to the importance of the kinds of attitudes that advocates of total quality management and other modern management methods highlight, including quality consciousness,39 continuous improvement orientation40 or flexible work orientation. 41

Finally, the majority of studies have concentrated on a narrow range of HRM practices in isolation, the exception being those that have included TQM in their analysis,20,31,34 but there is a general failure to assess the effectiveness of HRM and related practices relative to both key elements of management, such as leadership, and practices further afield, such as R&D expenditure.

In addition, despite the differences in focus and measures, most of the studies share a basic research design, which has a number of weaknesses. These include:

• The use of cross-sectional designs, which limits the ability to make causal inferences (e.g. to decide whether HRM practices promote performance, or rather better performance encourages greater investment in HRM).

• A reliance on a single data source (e.g. a chief executive or HRM director) for information on HRM practices and performance, which may result in measures of unknown reliability and common-method bias (see, for example, the debate between Huselid and Becker42 and Gerhart et al. 43 on the reliability of measures based on single respondents).

• Many have small samples, low response rates and concentrate on a limited range of sectors.

However, even the most searching reviews that have highlighted problems in the studies have concluded that there is nonetheless sufficient promise in the body of evidence to consider major investments in the area. 30 It is thus timely to invest in detailed assessment of what we know.

Developing research questions

Getting the question right is ‘the most important step in doing a review … [because] poorly focused questions lead to unclear decisions about what research to include and how to summarise it’. 48 The first stage is to develop specific research questions for the review.

The initial steps for this project involved breaking down the research objectives into their specific component questions. The aim was to provide questions that would form the basis of the series of systematic reviews that followed. The process of formulating precise, answerable questions from the research objectives was achieved through consultation within the team and with the expert advisers.

The six research objectives (that is, the five that were originally commissioned and the sixth added following consultation with experts) were broken down into 10 specific questions, as shown in Table 1.

Once this had been done, each research question was then typically broken down into the components of ‘population’, ‘intervention’, ‘comparators’ and ‘outcome’; this ‘PICO model’ provides a structure for devising research questions, constructing search strategies and defining the inclusion and exclusion criteria for the review. 47 It is not necessary for all of the elements of the PICO model to be present in every question. Where a PICO model has been used in these reviews, it is detailed in the specific methodology for that chapter.

Scoping exercise

The scoping exercise defines key issues for a review as clearly as possible in terms of the population, interventions and outcomes to be studied, and the inclusion and exclusion criteria to be applied. Together with the definition of the research question, this constitutes the ‘problem formulation stage, which involves the development of key research questions, construction of the definitions of the key concepts … and the establishment of inclusion and exclusion criteria for the review’. 60

For each review, a series of scoping exercises was undertaken to identify the best search terms with regard to the balance of specificity (relevance) and sensitivity (inclusivity). The nature of the HRM literature made the identification of specific intervention and outcome terms particularly challenging. Chapter 3 details the process of identifying HRM practice terms, and the specific intervention and outcome terms used by the respective reviews are given in Chapters 6–8.

Developing the protocol

The protocol provides an explicit outline and plan to be followed by the review. Conventional systematic review methods require that a protocol describe clearly defined populations, interventions and outcomes for the review, and provide details of the literature-searching procedure, quality assessment, and data extraction and synthesis to be conducted. 47 As the outcome of the scoping process, the protocol specifies the exact content and processes of the review, from which there should be no deviations. It facilitates the performance of the review by itemising exactly what the review does and does not include, and specifying the methods of quality assessment and synthesis. It should also state any language or geographical restrictions that might lead to bias in the review. 61,62

It is increasingly recognised that protocols for reviews in disciplines where the literature is a less known quantity, i.e. potentially heterogeneous, qualitative, and lacking the tradition of a recognised hierarchy of study design and quality assessment criteria,54,55,60,63 may be required to be broad, flexible and open to change. In short, as Eakin and Mykhalovskiy64 suggest, the question is to be viewed as a compass rather than an anchor, continually being enhanced and refined until completion of the review. Management is typical of such a discipline. 56 In such cases, it may be necessary to revisit and redefine interventions and outcomes in response to findings, and conduct iterative and purposive searching of the literature to address questions that arise or gaps identified in the literature. 52,58,60,63 This is often the case where the overall intent of the review is interpretive, not aggregative, and where the terminology and concepts are neither secure nor predefined. 65 The starting point, defined by the protocol, can therefore be broad and inclusive, rather than narrow and exclusive, in contrast with that required by more traditional systematic reviews.

In order to manage the diffuse HRM literature, simple, broad protocols were developed for each search. Protocols were reviewed regularly during the search process to help shape the eventual body of literature included in the data synthesis.

Identifying literature

After the scoping exercise, which is naturally exploratory, imaginative and purposive, the approach adopted is required to become more systematic, transparent and reproducible. A literature search for a review aims ‘to provide as comprehensive a list as possible of primary studies, both published and unpublished, which may be suitable for answering the question posed by the review’. 47 Systematic review methods almost unequivocally advocate an extensive and multimethod approach to identifying relevant literature. The intention is to identify all published and unpublished literature of relevance to the review that satisfies the review’s inclusion criteria. The principal means for achieving this is the searching of electronic databases. Search strategies follow the PICO model, which, as mentioned previously, breaks a search query down into four separate elements: population, intervention, comparator and outcome, a now-standard approach within health care. 66 Critical in this, was a time-consuming process designed to identify both the interventions (HRM practices) and the outcomes (whether intermediate outcomes, such as job satisfaction or final outcomes, such as reduction in mortality). The resulting lists and classifications inform most of the subsequent questions. This process is described in more detail in Chapter 3 (HRM practices) and Chapters 5 and 7 (intermediate and final outcomes).

Identifying longitudinal studies

A further component of the review was restriction by study design. This was used for sections of the review that sought to establish a causal link between interventions, in this case HRM practices, and either intermediate or final outcomes. While longitudinal studies in themselves do not prove causality, it is widely recognised that longitudinal analyses do provide better evidence for causation than cross-sectional analyses that use the same variables. 67,68 Indeed, contemporaneous systematic reviews of HRM have reported that they are limited in their ability to draw conclusions by their dependence on cross-sectional data. 69 The project team decided to limit the searches to longitudinal study designs only, as the intention was to identify those studies that could demonstrate causal links between HRM practices and outcomes, rather than simple associations. It is believed that this literature would offer evidence with greater capacity for answering the questions set by this review. No limitations of language were applied in order to avoid known biases produced by language and location. 61,62 There were no limitations by date other than those of the databases themselves.

Operationalising the search

For much of this particular review, only terms pre-specifying an intervention and an outcome were required. Searches aimed to optimise use of free-text searching and keyword terms, along with synonyms and variant spellings. This improves the chances of identifying all literature relevant to a topic, and is particularly necessary within social sciences where indexing and abstracting quality falls short of standards encountered within health care. 70

Literature searching requires more than interrogation of electronic databases with search strategies based on the protocol. It should use other formal methods, such as citation tracking and scanning the reference lists of selected papers, and hand searching of relevant journals. Informal methods of study identification can also be utilised, such as personal communications from peer reviewers or expert advisors and serendipity. Supplementary formal and informal methods may account for many, or even a majority, of studies included in non-quantitative reviews. 52,53,60,63 Such additional approaches are especially valuable if concepts are imprecise or if the literature is widely dispersed and inconsistently indexed.

In conventional review methodology, literature searching is refined by the scoping process and then carried out once. 47 However, where concepts or interventions are evolving or cannot be easily defined, supplementary, purposive literature searches may be performed. This may reflect an increasing understanding of the topic or the subsequent identification of specific questions raised by the findings of the review. 52,63 Searching becomes an iterative rather than one-off process, the so-called ‘spiral approach’ as described by Grant et al. 71 This was often how we had to proceed in this review, where pre-defining outcomes could have limited the findings. However, such a process is still systematic and consistent with the principles of systematic review methods. All search strategies and sources for all included papers are documented to make the process transparent, auditable and reproducible.

Developing the search terms

Scoping searches were performed using two of the less ambiguous categories from the list of HRM practices generated for the review: pay and training. Even relatively non-sensitive scoping searches of the principal health and psychology databases (MEDLINE and CINAHL) produced large numbers of research studies, which, although they used these terms, were not the focus of the review and would provide no data on HR performance linkages. Additionally, the scoping searches indicated that even larger numbers of citations would be retrieved by the more sensitive searches planned for the full review. Table 2 provides an illustration of the number of hits from three databases using standard and more sensitive versions of the pay and training filters. The number of potential HRM categories and practices was very large. A complete and formal process of reciprocal translation to produce a more clearly defined list would have been an extremely resource-intensive task and would have produced an unmanageable number of results from any searches performed. On account of the multifaceted nature of our study, with its multiple questions, it was decided that the approach of leading the review by specific HRM interventions was not viable. The original idea of structuring this review by HR intervention was therefore rejected.

Problems encountered using particular HRM terms required an alternative approach to the intervention terminology. The team tested the viability of constructing an intervention search filter using only generic HRM terms, such as human resource, HRM, HR and high-performance work practices or management, and database keywords, such as ‘personnel management’. This removed the problem posed by constructing strategies using potentially non-HRM-related terms, such as ‘training’, ‘communication’ or ‘pay’, which retrieved large numbers of studies and were, arguably, as generic and non-specific as this overarching terminology. Scoping searches using this approach produced manageable numbers of potentially more relevant studies and, therefore, this method was adopted. The resulting HRM search strategy is discussed more fully in Chapters 5–7, and example searches are provided in Appendices 6 and 7.

Search process

Literature search strategies were developed and performed by a systematic review information specialist on the following databases: Applied Social Sciences Index and Abstracts (ASSIA), British Nursing Index (BNI), Business Source Premier, Campbell Collaboration, Cochrane Central Register of Controlled Trials (CENTRAL), Cochrane Database of Systematic Reviews (CDSR), Cumulative Index to Nursing and Allied Health Literature (CINAHL), Database of Abstracts of Reviews of Effectiveness (DARE), DH-Data, EMBASE, Health Management Information Consortium (HMIC), International Bibliography of the Social Sciences (IBSS), King’s Fund database, MEDLINE, NHS Economic Evaluation Database (NHS EED), National Research Register (NRR), PREMEDLINE, PsycINFO, ReFeR, Social Sciences Citation Index (SSCI) and Science Citation Index (SCI). Search strings were modified for the different databases to take account of the different keywords and thesauri they use. Searching of electronic databases using pre-designed search filters was supplemented by reference tracking of selected studies. Searches were conducted in May/June 2006. Table 3 provides details of the search filters and databases used for different areas of the review.

Use of generic HRM terms is not ideal in identifying relevant studies. Papers that examined an HRM practice, such as employee feedback, could be missed if no mention of human resources, personnel or a synonym is made in the title, abstract or database designated keywords. However, this was the only way to operationalise HRM as a search strategy, given the extensive problems of using single practice terms. This search string was modified for different databases recognising the different keywords and thesauri they utilise. Additionally, the ‘HRM’ searches were supplemented by reference tracking and citation searching.

Study selection

This stage aims to identify articles that ‘help to answer the questions being addressed by the review’. 47 Study selection should be systematic, replicable and free from bias. Sifting is informed by criteria reflecting the population, intervention, outcome, study design, language and other criteria defined in the protocol. Titles and abstracts of retrieved articles are sifted by two reviewers or by one reviewer after an acceptable inter-rater reliability score is achieved. 47 Where a question remains over whether or not to include a study it is referred to a third reviewer (or a consensus is reached by two reviewers).

Conventional systematic review methods advocate the application of a quality threshold in determining study inclusion. 47 The aim is to evaluate the internal validity of studies by assessing selection, performance, measurement and attrition bias. The intention is to include only those studies that achieve the highest standards of methodology and reporting. Many tools or checklists have been developed for the critical appraisal of study quality based on study design, but each has limitations and there is no accepted standard set of tools. 72 Nevertheless, such tools are principally designed for specific types of medical research; surveys and case studies, for example, are less easily assessed, although quality assessment checklists have been developed. 58

However, there is a case for not excluding studies on the basis of quality, but simply to use assessment criteria to weight the findings of the studies in the review. 52,54,55,63,73 Quality assessment may act as a guide to interpreting findings and determining the strength of inferences, as well as enabling weighting of study results. It has been argued that qualitative studies cannot and should not be appraised formally,54,57 and that quality assessments may be evaluating limitations of reporting rather than methodology. 74 Quality assessment thus depends on the type of studies being reviewed. Controversy remains on whether or not to exclude ‘weak papers’. 75 In this review assessment of quality is used to moderate the messages from included research, and not as a quality threshold.

This review had already imposed a longitudinal criterion on the studies selected for review – a fairly high threshold for studies in the HRM field (in which cross-sectional, self-report design is common). Further weighting of studies by study design was applied and this is reflected in the relative weight given to the various studies in the data synthesis.

Data extraction

This is the process ‘by which reviewers obtain the information they need from what is reported by primary investigators’. 47 The aim is to collect from the selected studies the most important data for answering the research question. A data extraction form is piloted on several studies to ensure that reviewers are recording the same data and to develop forms of optimal length so that all key data are recorded. 47 In this review, data are primarily extracted and presented in the form of evidence tables.

Data synthesis

Synthesis aims to ‘draw together, contextualise and interpret the findings from the separate [studies]’. 53 Methods of synthesis are determined by whether data are qualitative or quantitative, and by the questions being asked of the data. 75 Meta-analysis is ‘the quantitative synthesis of data, in which evidence is pooled using statistical techniques’. 75 A narrative review seeks to ‘capture and describe, rather than “average out” the heterogeneity between studies’, and its choice as a method of synthesis may be ‘predicated on the diversity and complexity of the field’. 53 Several methods of synthesising qualitative data are being developed, such as thematic analysis, grounded theory, meta-ethnography and realist synthesis. 75 The method applied to each review is determined by the heterogeneity of the included studies and their data. In this review, synthesis is primarily descriptive and quantitative in recording the studies that meet specific characteristics, and is accompanied by a narrative synthesis. 76 There were very few studies in this review that addressed the same basic research question (i.e. examining the same relationship or set of relationships) precluding the use of meta-analysis for most intervention groups. Narrative synthesis was retained to ensure consistent synthesis across all interventions and enabling comparisons to be made.

Objective 1

Objective 1 is a mapping exercise involving both the identification of the range of HRM practices documented in the literature and the HRM practices actively reported in use reported by management practitioners. It is necessary to triangulate findings from these two sources in order to identify any gaps between research and practice.

• Question 1.1 What HRM practices are described in the literature?

Method Content analysis was performed on six, recent major reviews of HRM practices, supplemented by additional literature identified for Objective 6. Practices were then categorised under broad headings of HRM, using accepted definitions of these categories, for example work design, performance management.

• Question 1.2 How widespread is the use of HRM practices in UK organisations?

Method Evidence was drawn from the best available surveys of HRM practices in the UK. These surveys were identified from the expert knowledge of members of the project team, personal communications from colleagues, and systematic searching of electronic databases and relevant websites. Surveys were included if they contained UK data. They were then weighted according to specific criteria. Results were tabulated.

• Question 1.3 How can the fidelity with which an HRM practice is implemented be measured?

Method A critical review was undertaken of research on implementation fidelity and relevant studies from the field of HRM. Studies were identified by systematic searching of electronic databases and informal methods, such as serendipity. A model, a measure and a guidance document were developed, based on this literature review.

Objective 2

• Question 2.1 What are the most reliable measures of the intermediate outcomes identified for this review?

Method The intermediate outcomes to be assessed by this review were identified by content analysis of the relevant literature and a Delphi process involving HRM experts. The existing review and theory literature on the measures of these intermediate outcomes was examined to identify candidate measures. This literature was identified from a combination of systematic searching of electronic databases and informal methods, such as expert knowledge.

Objective 3

• Question 3.1 What is the evidence on the intercorrelation of intermediate outcomes with each other?

Method Intermediate outcomes covered by this question were identified as described under Objective 2. The existing review and meta-analysis literature was reviewed to identify correlations between individual intermediate outcomes. This literature was identified by a combination of systematic searching of electronic databases and reference tracking.

Objective 4

• Question 4.1 What is the evidence for the impact of intermediate outcomes on non-health final outcomes?

Method Once again, the intermediate outcomes covered by this question were devised as for Objective 2. The existing review and meta-analysis literature was identified and reviewed to evaluate correlations between individual intermediate outcomes and final outcomes, such as performance, turnover and absenteeism. This literature was identified by a combination of systematic searching of electronic databases and reference tracking.

• Question 4.2 What is the evidence for the impact of intermediate outcomes on patient care outcomes?

Method Intermediate outcomes were again identified as for Objective 2. The aim was to produce a synthesis of the longitudinal literature for each intermediate outcome as it relates to the patient care outcomes identified for this review. This literature was identified by a combination of formal methods, such as systematic searching of electronic databases and reference tracking.

Objective 5

• Question 5.1 Do any HRM practices have a significant impact on the intermediate outcomes identified for this review?

Method Again, each intermediate outcome identified under Objective 2 was examined in a systematic review of the longitudinal HRM literature. This literature was identified from Objective 6, Questions 6.1 and 6.2, which included longitudinal research examining the impact of HRM practices on various intermediate outcomes. A qualitative, narrative synthesis of the literature was produced.

Objective 6

• Question 6.1 Which HRM practices have an impact on performance outcomes?

Method A systematic review was undertaken of longitudinal research examining HRM practices and both intermediate and final outcomes. Studies were identified by the systematic searching of electronic databases and reference tracking, and by informal methods such as expert advice. The literature was mapped to the category framework developed in response to Question 1.1 and a qualitative, narrative synthesis of the literature was produced.

• Question 6.2 Which HRM practices have an impact on patient care outcomes?

Method A systematic review was undertaken of longitudinal research examining HRM practices and their impact on patient-care outcomes. Studies were identified by the systematic searching of electronic databases and reference tracking, and by informal methods, such as serendipity. The literature was mapped to the category framework developed in response to Question 1.1 and a qualitative, narrative synthesis of the literature was produced.

Against this methodological backdrop, the specific details of each review are described in the corresponding section of the report and justification is given for methods used.

Step 1

The research team and expert advisers generated a list of 13 headings. Eleven of these were broad HRM categories under which it was anticipated that the majority of individual HRM practices could be grouped, which could then be used as a framework to shape the reviews. Initial HRM practice terms are presented in Table 4.

It was recognised from the outset that HRM practices were unlikely to all fall into neat categories. In addition to the 11 broad categories of HRM practice identified, two further categories were used to group individual practices:

• The ‘Miscellaneous’ category was used to group practices that were not subject to direct manipulation by the HR department, or were too broad to be influenced by HR policies/practices alone, but were of potential interest to the review for their impact on organisational or individual performance. Examples include bullying/harassment, socialisation, etc.

• A sizeable number of HR policies/practices, 29 in total, were considered to be part of HR maintenance functions (i.e. administration) rather than proactive HR practices aimed at enhancing performance. Examples include equal opportunity, dispute resolution policies, and health and safety policies.

The classification of HRM practices into broad categories was reviewed by the expert advisers. As a result, the terms were refined according to the most appropriate HR category as defined and used in the literature and additional HRM practice terms were suggested, which had not been identified in the review of reviews:

• The higher level concepts category was renamed ‘bundles’ of practices. Over the last 10–15 years there has been a growing body of research on ‘bundles’ of practices. This category was therefore intended to capture research that focused on combinations of HRM practices that were intentionally designed to be used in concert as well as for multiple practices opportunistically used together.

• Articles in the ‘Miscellaneous’ category were reappraised and either recategorised under one of the main headings or excluded from this assessment of the main HRM practices described in the literature.

Step 2

In a parallel process, a list of HRM practices was compiled from six major recent reviews of the HRM literature. 30,35,77–80 This process identified 245 practices.

A process of reciprocal translation81 was applied to the 245 practices identified by the review of reviews, with the intention of building a consensus about the main HRM practices covered in the literature. Reciprocal translation involves the comparison of themes across papers and an attempt to establish whether there is equivalence in the themes of one paper with those from another. The aim is to ensure that a key theme captures similar policies or practices from different papers. In this exercise, the reciprocal translation was approached by two of the researchers independently attempting to compare HRM practice terms across review papers and group them under one of the broad HRM categories listed in Table 4.

Initially, a substantial number of practices did not appear in any of the categories or could be allocated to more than one. For example, appraisal-based pay can be seen as a part of the broad HRM category of appraisal/performance management and of compensation and reward systems. Likewise, introducing autonomous work teams can be seen as both an employee involvement strategy and a work design approach.

The process of reciprocal translation made it apparent that the same term could be used to describe different practices in different papers (e.g. ‘goal setting’ to mean performance management in one context and employee involvement in another).

At this stage in the research, it became apparent that the diffuse nature of the HRM literature, in particular the inconsistency of terminology and the blurred boundaries of some of the concepts, meant that the broad categories identified through this process would not be a suitable as the sole basis for the subsequent review of evidence. Ultimately, a combination of broader search terms and study design criteria were used (which are described in Chapters 2, 5 and 6).

Step 3

The systematic reviews undertaken to meet Objectives 3–5 generated over 450 papers (see Chapter 5 for a description of these searches). The individual HR practices, policies or interventions that had been the focus of these studies were mapped on to the list of HR practices generated by Steps 1 and 2. This was done in order to identify any gaps in the list of HR practice terms and further refine the categories. The exercise revealed that:

• A large body of literature in the health sector focused on specific HR practices not commanding much attention in the non-health sector – specifically ‘working hours’, ‘staff levels’ and ‘skill mix’. These practices were included under the ‘staffing’ practice category along with selection and recruitment.

• All TQM practices identified (a separate category on the original list) were, in fact, training interventions, so were included under the broad ‘training’ heading.

The final list of 10 broad HRM categories is as follows:

1. work design

2. staffing

3. training and development

4. compensation and rewards

5. communication

6. family friendly

7. single status/status harmonisation

8. employee representation, involvement and participation

9. appraisal/performance management

10. bundles.

Appendix 2 contains a list of these broad HRM categories with definitions. Appendix 3 presents the specific HRM practices identified in the literature grouped under these 10 broad categories.

Aim

As part of Objective 1, the review sought to document evidence on the extent of use of HRM practices within UK organisations or workplaces. The intention is to:

• provide a comprehensive report of the percentage of organisations using the HRM practices identified for this review and to identify gaps in the survey literature

• relate findings of the review of the research literature to actual practices used within UK organisations. For example, research may demonstrate that a practice has a significant beneficial impact on job satisfaction or patient outcomes, while survey data may indicate that this practice is not actually widespread.

The first aim is addressed in this chapter and the implications for the overall conclusions of the review are considered in Chapter 11.

Methods

Findings from the highest-quality and most up-to-date HR surveys were used to generate these data. Surveys were identified using several strategies. The Workplace Employment Relations Survey (WERS) and other survey organisations were identified by members of the project team. Searches were conducted to identify appropriate publications on the websites of relevant organisations, including: Community Innovation Survey (CIS), Continuing Vocational Training Survey (CVTS), European Survey on Working Time and Work–Life Balance (ESWT), PASO (Panel Survey of Organisations), Employee Direct Participation in Organisational Change (EPOC Project), and Workplace and Employee Survey (WES) Canada. Searches of the SSCI and King’s Fund electronic databases were used to identify additional surveys, particularly for the UK health sector. Search strategies used combinations of the following terms and their variants: survey(s) and practice(s) and either human resource(s), staffing, work or personnel. To be included, the survey data had to relate to UK workplaces.

Surveys were ordered in a hierarchy, based on the following quality assessment criteria. The survey reported data that:

1. were weighted and representative of UK workplaces (i.e. data were collected or adjusted so that the resulting figures do not simply represent large organisations or organisations from particular sectors or industries)

2. were based on the unit of the workplace (this can be a self-contained entity or a unit of a larger organisation) rather than an individual employee

3. included the UK public sector and/or health sector.

Data from surveys that were weighted were preferred to data from surveys that were not representative of UK workplaces but simply reported data based on the unit of the workplace. Data from such surveys were, in turn, preferred to those that simply reported UK private or public sector data. If data on a practice were available in a survey that was weighted and representative of UK workplaces then these data were used in preference to data on that practice from a more limited survey.

Results

Health sector-specific surveys

The most comprehensive surveys within the health sector are those conducted by the Healthcare Commission (HCC). Whilst providing extensive information about staff experiences of working within the NHS, the annual staff survey differs considerably from the HR surveys reviewed here in that:

• It is targeted at the individual employee, rather than the HR function.

• It asks questions about individual staff perceptions or experiences rather than the absence or presence of specific HR practices.

• The nature of the data means that findings are reported at the individual level (i.e. percentage of all staff responses to a particular item) for the NHS as a whole, or by type of Trust [e.g. all Primary Care Trusts; all Ambulance Trusts, etc.], or by Strategic Health Authority, or by individual Trust, rather than at the organisational level (i.e. percentage of Trusts/organisations reporting the presence of an HR practice).

Comparing such data with HR practice measures is extremely problematic. For example, the proportion of employees reporting that they have had an appraisal in the last 12 months may not bear a strong relation to the proportion of NHS organisations with appraisal policies in place, or to the existence of an appraisal policy in any specific organisation. Accepting these limitations, it is possible to interpret some of the NHS Staff Survey findings as a proxy indicator for staff experience of particular HR practices in the NHS.

Analysis of data from the NHS Staff Survey focuses on 26 key areas (key scores). These are a mixture of attitudinal and experiential measures, the latter providing an indication of the percentage of NHS staff who have experienced a particular HR practice (i.e. a proxy indicator).

Where a proxy indicator for staff experience of an HR policy is available from the National NHS Survey 2007,103 this has been indicated in the right-hand column of Table 6. Findings from the survey are summarised later (see Table 17) and discussed below (see Data from the NHS Staff Survey 2007, below).

Data from the NHS Staff Survey 2007

Data from the NHS Staff Survey103 is presented in Table 17. Caution is needed when considering these data in relation to the data on HR practices in general. The NHS data shows the proportion of NHS staff who report work experiences that are consistent with the HR practices considered here. It is not, however, indicative of the number of Trusts in which such HR policies/practices exist, nor does it help address other important questions raised in this section, such as the percentage of the workforce covered by a particular HR practice in a given organisation.

Accepting the limitations of the NHS Staff Survey103 data in relation to the focus of this chapter, it can be considered as providing proxy indicators of staff experiences in areas. This is of particular interest where organisational level HR practice data are not available.

Specifically the NHS Staff Survey103 data indicate that 39% of NHS employees work in a well-structured team environment. This contrasts with data from the survey of HR practices, which indicate that 72% of all organisations surveyed have some employees working in teams and 79% of health sector organisations make some use of team working.

With regard to training, 54% of employers surveyed on HR practices had funded or arranged some training over the previous 12 months. No breakdown is available for health sector organisations, but 94% of NHS employees report taking part in some form of training over the previous 12 months. Eighty-one per cent of employers from the HR practice surveys provided job specific training and 80% health and safety training. From the NHS Staff Survey, 77% of staff reported receiving job relevant training, 71% health and safety training.

There are no health sector-specific breakdowns available for the wide range of communication practices covered by the HR practice surveys. The NHS staff survey indicates that 23% of NHS employees feel that senior management communicate effectively with staff.

No breakdown is available for the percentage of health sector employers offering family friendly/work–life balance HR practice; however, the NHS Staff Survey103 indicates that between 2% and 32% of NHS employees make use of one of a range of specific practices in this area, with 73% of all employees reporting that they use flexible working options.

Eighty-nine per cent of health sector organisations report having an equal opportunities policy in place, but no further breakdown of specific HR practices in this area is available. Thirty-nine per cent of NHS employees report receiving some form of diversity training. Levels of training in age, disability, gender, race, sexual orientation or religion-related equality/diversity range from 14% to 18% of employees surveyed.

With regard to employee involvement, there is some evidence from HR practice surveys in relation to health sector organisations. This mainly relates to union membership and joint consultative committees (Table 14). No breakdown is available for the percentage of health sector organisations using other employee involvement HR practices (such as suggestion schemes or problem-solving groups). Findings from the NHS Staff Survey103 indicate that 23% of NHS employees feel that senior managers try to involve staff in important decisions and 31% feel that senior managers encourage staff to make suggestions to improve services.

With regard to direct participation, 48% of NHS staff surveyed felt that they were involved in decision-making, and/or consulted about decisions that affected them.

Finally, with regard to performance management, 89% of health sector employers reported using performance appraisals (compared with 71% for all organisations). From the NHS Staff Survey, 61% of respondents said that they had received an appraisal in the previous 12 months. Interestingly, this drops to 24% when factors indicating a well-structured performance appraisal were taken into account. Fifty-two per cent of staff (as measured by the NHS Staff Survey103) reported that they agreed a personal development plan (PDP) as part of their appraisal.

Intermediate outcomes

Although there are a multitude of variables that are implicated in this process, some of the more important proposed mental, emotional and attitudinal states are employee well-being, an effective psychological contract, perceived organisational support, fairness and justice, motivation, job satisfaction, commitment job involvement and organisational climate. 156 An organisation’s HR architecture will be closely tied to these states. For example, satisfaction of employees, perceptions of fairness and trusting management–employee relationships may be influenced by HR practices that provide employees with opportunities for participation and autonomy. Intrinsic motivation could well be influenced by work design interventions, such as job enrichment and empowerment. Firms offering job security and organisation pay programmes, such as profit-related pay, can engender employee perceptions’ of organisational support, and their identification with, and commitment to, the organisation.

However, a lack of clarity remains about the degree of overlap between the constructs describing employee states and their causal ordering (e.g. Does satisfaction precede commitment, and under what circumstances?; Is job satisfaction a consequence of job involvement or is the relationship reciprocal?). Also the relationships between any one individual’s feelings, attitudes, motives and performance will not be straightforward. 157 As Ostroff and Bowen155 point out, ‘Lower performance is only one possible response to dissatisfaction. A dissatisfied employee could file a grievance, engage in sabotage, ask for a transfer, try to improve performance or fail to pass on important information. 158 In contrast, a satisfied employee could work harder, engage in more citizenship behaviours, improve skills or make suggestions for improvements in work procedures’ (p. 228). But although the relationship between satisfaction and performance at the individual level may be not be large, the collective effects for the organisation as a whole may be enhanced due to the overall cumulative impact of the responses of satisfied or dissatisfied employees. 155

Productivity-related employee behaviours

How do employee mental, emotional and attitudinal states get converted into organisational performance? The assumption is through their influence on employee behaviours salient to effective organisational performance, i.e. when these mental, emotional and attitudinal states are favourably directed then employees are more likely to display performance-orientated behaviours. 156 Three pertinent sorts of employee behaviour are task performance, withdrawal behaviours and organisational citizenship behaviour (OCB). 154

Task performance behaviours include job-related tasks and activities comprising the employee’s formal role.

Withdrawal behaviours refer to absenteeism and turnover.

Organisational citizenship behaviours include components of so-called prosocial behaviours, and refer to constructive or cooperative gestures that are not mandatory, but which contribute to organisational effectiveness. 159 OCB refers to work-related behaviour that is outside traditional task performance, in-role job performance and the formal organisational reward system, but is beneficial to the organisation. 160 Typically, employees who engage in OCBs are those who go the extra mile for their organisations and therefore contribute to its effective functioning. This discretionary work performance has been conceptualised in a number of ways (e.g. contextual performance, prosocial organisational behaviour, extra role performance), but the research literature has focused most prominently on OCB. The most popular conceptualisation of OCB is a five-dimension model proposed by Organ,159 consisting of altruism, courtesy, conscientiousness, civic virtue and sportsmanship.

Although much of the work on OCB takes the position that there is a clear boundary between in-role and extra-role behaviour, the distinction is not as clear as much of the literature suggests. 161

Research has suggested that employees who are satisfied with their jobs and committed to their organisation, and feel that they are treated fairly, are more likely to engage in OCBs. 162 Consequently, OCBs have recently emerged as a potential important linking mechanism between employee attitudes, such as commitment and satisfaction, and organisational outcomes. Any single incidence of citizenship may have a minor or moderate effect – ‘however, across multiple acts of citizenship from a single employee, and across multiple employees, the aggregate impact should be more substantial’ (Ostroff and Bowen:155 227). However, so far, arguments for an association between employee OCB and organisational performance are conceptual rather than empirical.

In summary, despite the paucity of evidence on these relationships (evidence that will be reviewed in later chapters), it has been consistently theorised that satisfaction, motivation and other mental, emotional and attitudinal states of employees are key factors in determining appropriate productivity-related employee behaviour, which ultimately contribute to organisational performance.

A large number of variables may serve as intermediate outcomes linking HRM practices and final performance outcomes, from job satisfaction to organisational commitment, and from burnout to organisational support. Given the vast range of potential intermediate outcomes available from the research, and the limitations of project time and resources, it was necessary to select a limited number of intermediate outcomes.

Motivation

Work motivation can be generally defined as the direction, intensity and persistence of work behaviour;168 i.e. what drives a person’s choice of what to do, how hard they try and how long they keep trying. Motivation is clearly critical to performance. Indeed, work performance is often used as a measure of a person’s motivation,169 although there are many other factors that influence performance other than motivation (e.g. ability, resources, employee collaboration and cooperation). There are many motivation theories that have been advanced to help understand behaviour at work but it is often difficult to know which theory is helpful in any individual case. 169 Some theories are more relevant to the notion that HRM practices effect employee motivation, such as goal-setting theory, equity theory and job design theory.

Job satisfaction

Job satisfaction is the most widely researched concept in organisational psychology and organisational behaviour. Locke170 estimated that over 3300 studies on job satisfaction had been conducted up to 1973. Judge et al. 171 then identified a further 7856 studies on job satisfaction since 1973 using the PsycINFO database. Studies and papers have been written about its definition, meaning, specific dimensions of job satisfaction, its antecedents and its consequences. Its popularity stems from the key role it plays in many theories and models of organisational practices and processes, individual attitudes and behaviours. Job satisfaction is important as an end in itself for employees and also because of proposed relationships with individual and organisational effectiveness (e.g. turnover and performance). It is a widely held belief that ‘a happy worker is a productive worker’ (although there is considerable debate about the validity of this statement), and job satisfaction is one way of measuring happiness at work.

Locke170 defined job satisfaction as ‘a pleasurable or positive emotional state resulting from the appraisal of one’s job or job experiences’. Thus job satisfaction involves affective (emotional) states or how one feels about one’s job, and cognition, deriving from appraisal of the situation. Although there is some debate about the emphasis placed on affect and cognition, it is probably most helpful to think of job satisfaction as the interplay of cognition and affect, or thoughts and feelings. 171

Job satisfaction can be thought of as an overall assessment of one’s job but which comprises various job facets, such as satisfaction with pay, recognition, promotion, co-workers, supervision and the work itself. In addition, the concept is sometimes separated into intrinsic satisfaction, which involves satisfaction with supervision, co-workers and the work itself, and extrinsic satisfaction that targets features extrinsic to task activities, such as pay and promotion.

Organisational commitment

Employee commitment to an organisation has been defined in a variety of ways but is most popularly operationalised as the level of attachment and loyalty to an organisation among its employees. 172 The concept has been particularly popular in the HRM literature as many scholars advocate that the traditional relationship between employer and employee, based on control, must be replaced by a high-commitment approach. 1

Mowday et al. 173 who were influential in the early work on commitment, characterised it as a strong belief in and acceptance of the organisation’s goals and values, a willingness to exert considerable effort on behalf of the organisation, and a strong desire to maintain membership in an organisation. They have developed a widely used scale, the Organisational Commitment Questionnaire (OCQ), to measure these elements. But the definition and the measure have been criticised for conflating commitment with outcomes such as effort and propensity to stay.

Meyer and Allen174 have suggested definitions and measures based on three broad themes. Affective commitment describes an employee’s liking for an organisation and emphasised identification and involvement in the organisation. Employees with a strong affective commitment continue employment with the organisation because they want to do so. 175 Continuance commitment refers to an awareness of the costs associated with leaving the organisation. Employees whose primary link to the organisation is based on continuance commitment remain with the employer because they need to do so. Finally, normative commitment reflects a feeling of obligation to continue employment. Employees with a high level of normative commitment feel that they ought to remain with the organisation.

The HRM literature has, not surprisingly, concentrated on the affective aspect of commitment because of the assumed associated positive organisational gains. The general theory of organisational commitment predicts that high commitment should result in greater motivation and performance, lower absenteeism and lower labour turnover,172 with consequent organisational benefits.

Occupational/professional commitment

Much of the theoretical and empirical work on commitment has focused on the organisation; however, it is clear that workers have multiple commitments at work (such as work itself) and the organisational unit (such as the team, department or branch). An important commitment domain outside the employing organisation is commitment to the profession or occupation as well as the family.

As with organisational commitment it is the state of psychological attachment that is important in the occupational commitment literature. 176 Occupational commitment is the individual’s psychological attachment to, and identification with, his/her occupation, based on an affective reaction to that occupation. 177

The strength of employees’ occupational commitment may have important consequences for organisations because of its potential links to retention and work performance. Also, the salience of occupational commitment may be on the rise as a means of coping with the uncertainty of extensive organisational change. 178 Workers may increase their identification with their occupation on the basis that they retain more control over that aspect of their working life.

Although there has been debate over the possible inherent conflict between occupational commitment and organisational commitment, more recent work argues that the two forms of commitment are not incompatible, but rather they are positively associated. 177

Work engagement

Work engagement refers to ‘a positive, fulfilling, work-related state of mind characterized by vigour, dedication and absorption’. 179 Vigour is characterised by high levels of energy and mental resilience while working, a desire to invest effort in one’s work and persistence in the face of adversity. Dedication refers to a strong psychological involvement in one’s work, together with feelings of pride, significance, enthusiasm and inspiration. Finally, absorption refers to total concentration on, and pleasant immersion in, work.

Although the concept is relatively new, work engagement does overlap with established concepts also included in this review as intermediate outcomes. Vigour is conceptually similar to motivation, and the dimension of dedication shares conceptual space with those of job involvement and commitment. 180 Engagement has been defined as the opposite of burnout,181 as, contrary to employees suffering from burnout, engaged employees are energetically and effectively connected to their work, and able to cope well with the demands of the job. 182

Engagement is influenced by the availability of job resources that reduce job demands, help achieve work goals and stimulate learning and development. 183 Examples of possible resources are training and coaching, performance feedback and participative management.

Burnout

Burnout is defined as ‘a persistent, negative, work-related state in normal individuals that is primarily characterised by exhaustion, which is accompanied by distress, a sense of reduced effectiveness, decreased motivation and the development of dysfunctional attitudes and behaviours at work’. 184 It is a psychological strain resulting from chronic work stress, such as job demands.

The concept of burnout was initially associated with human services work such as health care and social work, where employees are engaged in ‘people’ work, for example, providing patient care. It is usually characterised as a three-dimensional syndrome. Le Blanc et al. 185 describe the dimensions in relation to burnout in nursing staff in three stages:

• First, high emotional demands in interpersonal relationships with patients lead workers to feel emotionally exhausted.

• Second, as a coping strategy they may try to protect themselves by detaching from their patients by treating their patients in an indifferent and cynical way. 186,187 This detached attitude towards patients is called depersonalisation.

• Finally, as a result of this dysfunctional attitude, workers are unable to perform adequately, and the quality of their care is impaired. In turn, this may lead to a decline in their feelings of personal accomplishment or professional efficacy.

More recent research has argued that burnout exists outside the realm of human services. Consequently, the concept has been adapted to apply to all employees, by replacing the depersonalisation dimension with cynicism, which reflects indifference towards work in general, not necessarily towards other people. 188

Burnout may have negative consequences not just for the individual (such as health problems), but also in relation to organisational outcomes such as absenteeism and performance.

Job involvement

The concept of job involvement has been the subject of a large volume of research for over 40 years. Although it is subject to some definitional confusion, Brown189 notes that most research has followed Lawler and Hall’s190 definition of job involvement as ‘psychological identification with one’s work’ and ‘the degree to which the job situation is central to the person and his [or her] identity’. The more recent concept of employee engagement treads on similar territory.

Multiple outcomes are often attributed to high job involvement. Job involvement has been considered important for promoting employee motivation,191 goal-directed behaviour, personal growth and satisfaction, and consequently performance. 190,192,193 For example, Janssen194 describes it as ‘the key that unlocks motivation’ to exert extra effort. Increasing job involvement has been advocated as a means to ‘enhance organisational effectiveness and productivity by engaging employees more completely in their work and making work a more meaningful and fulfilling experience’. 189 Developing enriched jobs is regarded as an important means for stimulating job involvement. 192,195

Turnover intentions

Morrell et al. 196 describe turnover as ‘voluntary cessation of membership of an organisation by an employee of that organisation’. Turnover intention or intention to quit is the psychological state that precedes this cessation.

Psychological contract

The relationship between employees and employers is central to HRM. The concept of the psychological contract explores this relationship by describing the mutual obligations or exchange relationship perceived by the employee to bind him or her with the organisation. 197 The individual’s psychological contract reflects his of her understanding of each party’s obligations, derived in part from HRM practices, and also verbal statements and actions each party has made.

Psychological contracts have important implications for employee attitudes and behaviours, especially when employees perceive that obligations or promises explicitly or implicitly made by an employer have been unfulfilled. Contract perceptions are associated with attitudes such as mistrust, employee behaviour (including effort and contribution), frequency of absence, and intention to stay or leave. 198

Organisational justice

Organisational justice is one of the most popular research topics in HRM, organisational psychology and organisational behaviour. 199 The literature has ‘grown around attempts to describe and explain the role of fairness as a consideration in the workplace’,200 and much of the research in this area starts from the premise that employees’ judgements of fairness are important in determining organisational outcomes such as job satisfaction, organisational commitment, intention to quit and OCB. 201 What is just or unjust is based on individuals’ perceptions of fairness.

Organisational justice research has primarily been concerned with distributive justice and procedural justice. Distributive justice deals with perceived fairness of the distribution of outcomes, such as pay and promotion decisions, whereas procedural justice focuses on the fairness of the process by which outcomes are determined. 202 The process for making organisational decisions may be equally or more important to employees as their perception of outcome fairness.

Interactional justice,203 as more recently advanced in the justice literature, relates to the quality of the interpersonal treatment people receive when procedures are implemented, and can be seen as taking two forms. 204,205 First, interpersonal justice, which pertains to the degree to individuals feel they are treated with respect, politeness and honesty in the process. The second, informational justice, reflects employees’ perceptions about whether they feel that they are provided with adequate explanations for the use of procedures or for outcome distribution.

Organisational support

Perceived organisational support refers to employees’ global beliefs concerning the extent to which the organisation values their contributions and cares about their well-being. 206,207 Employees view their favourable or unfavourable treatment as in indication that the organisation favours or disfavours them. Perceived favourable treatment can be around issues such as fairness, organisational rewards and favourable job conditions (e.g. pay, promotions, participation). Favourable treatment contributes more towards perceived organisational support if the employee believes that they result from the organisation’s voluntary actions, as opposed to external pressures such as government regulations. 206,207

Employees who perceive themselves to be supported and valued are more likely to respond in a favourable manner. Organisational support theory proposes three psychological mechanisms underpinning the consequences of support. 208 First, on the basis of reciprocity, organisational support should produce an obligation in employees to care about the organisation’s welfare and goal attainment. Second, the support and respect provided by the organisation should satisfy socioemotional needs, leading to employees psychologically identifying with the organisation. Third, perceived organisational support should increase employees’ beliefs that the organisation recognises increased performance. These processes should result in positive attitudes (e.g. commitment and satisfaction) and behaviours (e.g. reduced turnover, increased performance).

Perceived organisational support has been found to be related to, but distinct from, organisational commitment, procedural justice and job satisfaction. 208

Organisational climate

Central to most, if not all, models of organisational behaviour are employee perceptions of the work environment, referred to generally as organisational climate. 209 Primarily understood as an intervening variable between the context of an organisation and the behaviour of its members, and attempting to understand how employees experience their organisations, the concept has inspired many descriptions and operationalisations. The dominant approach conceptualises climate as employees’ perceptions of what their organisation is like in terms of its routines, practices, procedures and rewards. 210

An initial assumption of theory and research in this area was that social environments could be characterised by a limited number of dimensions. Typical dimensions were role stress and lack of harmony; job challenge and autonomy; reward orientation; leadership facilitation and support; work group cooperation, friendliness and warmth. 211,212 However, over the years the number of climate dimensions identified as targets of assessment has proliferated to cover numerous dimensions of organisational life. Confusingly, some of the dimensions that are frequently used in climate research overlap or match other constructs considered in this review, such as perceived organisational support, and work design variables, such as autonomy and role conflict.

An important distinction has been made between psychological climate and organisational climate. 213 Psychological climate is based on individuals’ perceptions of the organisation, but when individuals in the same unit (e.g. team, department) or organisation agree or share their perceptions, organisational climate emerges, represented by the mean score across all employees within a particular unit or organisation. The rationale behind aggregating individual data to a unit level is the assumption that organisational collectives have their own distinct climate.

Culture and climate are similar concepts since both describe employees’ experiences of their organisations. Organisational climate, according to Schneider,214 represents the descriptions of the things that happen to employees in an organisation, they tell us what is happening, whereas culture comes to light when employees are asked why these patterns exist. The question is answered in relation to shared values, common assumptions, and patterns of beliefs held by organisational members and it is these which define organisational culture.

Schneider214–216 eschews the use of general multidimensional measures of climate and argues for a facet-specific climate approach where climate has a focus and is tied to something of interest. Schneider suggests that the dimensions of organisational climate will differ depending on the purpose of the investigation and that general measures of organisational climate will contain dimensions that are not relevant for each specific study. This line of argument has encouraged the development of measures with a particular strategic focus such as service217 and safety,218 which are predictive of relevant outcomes (e.g. customer service, safe behaviours).

Research has suggested that climate perceptions are associated with a variety of important outcomes at the individual, group and organisational levels. These include leader behaviour,209,219 absenteeism and turnover,220 job satisfaction,221–223 commitment,224 individual job performance,225,226 and organisational performance. 36

For the purposes of this review, the focus was initially on general multidimensional measures and performance outcomes. However, very little was found in this area so the criteria were expanded to include facet-specific measures.

Within HRM and performance models, climate is often conceived as a linking mechanism between the HR system and collective employee thoughts, emotions and attitudes. 154 That is, the HR system influences climate, which shapes employee states, such as satisfaction and motivation, and subsequently employee behaviours and performance.

Inclusion criteria

• Longitudinal study design.

• HRM practice (e.g. work design, training, appraisal).

• One of the intermediate outcomes identified in Chapter 5 (e.g. job satisfaction, burnout, organisational commitment).

• The intermediate outcome must apply to health professionals.

• The study must test the causal relationship between the practice and the intermediate outcome.

Exclusion criteria

• Unpublished dissertations.

Summary of evidence

References identified by the search were scanned by two reviewers for studies that satisfied the inclusion criteria. Intermediate outcomes are categorised with the corresponding numbers of studies as shown in Table 22, below.

Job satisfaction

All of these studies consider the impact of HRM interventions or practices on job satisfaction among health professionals. A narrative synthesis is preferred here despite the homogeneity of some clusters of studies in this sample. No systematic reviews of this relationship in the health sector have been found.

Burnout and morale

All studies consider the impact of HRM interventions or practices on burnout and morale among health professionals. No systematic reviews of this relationship in the health sector have been found.

Turnover intentions

All studies considered the impact of HRM interventions or practices on health professionals’ turnover intentions or intention to withdraw or quit. No systematic reviews of this relationship in the health sector have been found.

Organisational commitment

A single study considered the impact of HRM interventions or practices on organisational commitment among health professionals. No systematic reviews of this relationship in the health sector have been found.

Climate

A single study considered the impact on HRM intervention or practice on the cultural or working climate among health professionals. No systematic reviews of this relationship in the health sector were found.

Protocol development

Protocols were produced for the review question ‘Which HRM practices have an impact on patient care outcomes?’ but much was left open because of the lack of clear definitions and because the literature itself was an unknown quantity. The literature was so diverse and heterogeneous that nothing was fixed beyond the broad categories of HRM practice (Chapter 3) and the patient outcomes described above, and the plan for the systematic search of the literature by various methods. There was to be no exclusion of studies based on formal quality assessment. However, longitudinal study design was a criterion for inclusion in the review and served as a form of quality assessment. The details of the data extraction and synthesis remained open and flexible. Such flexibility is permitted by evolving systematic review methods, especially when dealing with complex literatures. 52,56,58,60,63

Search strategy

Literature search strategies were developed and performed on the following electronic databases: ASSIA, BNI, Business Source Premier, Campbell Collaboration, CENTRAL, CDSR, CINAHL, DARE, DH-Data, EMBASE, HMIC, IBSS, King’s Fund database, MEDLINE, NHS EED, NRR, PREMEDLINE, PsycINFO, ReFeR, SSCI and SCI. Search strings were modified for the different databases to take account of the different keywords and thesauri they use. Examples of the search filters and strategies can be found in Appendices 6 and 7. The searching of electronic databases using pre-designed search filters was supplemented by reference tracking of selected studies.

Study selection

A pilot sift was conducted by two reviewers to test the study selection criteria form and to assess inter-rater reliability. Titles and abstracts of studies identified by both electronic database searching and reference tracking were sifted using the following inclusion criteria.

Studies that:

• were longitudinal in design

• examined an intervention which could be described as an HRM practice, such as training, work design or performance appraisal

• evaluated the impact of this intervention on a patient care outcome.

Studies were excluded if they were unpublished theses.

All of the identified articles were sifted by one of the research associates and half by a second reviewer. The k-statistic for inter-rater reliability across the double-sifted articles was 0.9. This complies with the standards required by conventional systematic review methodology. 47 In cases where there was disagreement, the two reviewers attempted to reach a consensus. If they could not, a third reviewer took the decision about inclusion. There was no formal quality assessment of the studies at this stage. The only quality criterion applied was that the study needed to be longitudinal. Papers were included only if they had a longitudinal study design, an HR intervention and a stated outcome. Papers were excluded if they did not satisfy these criteria or if they were student theses or dissertations.

Data extraction

A data extraction form was piloted by two reviewers. This process indicated the high degree of heterogeneity between the studies. As a result, the team designed a first-level form to extract only the minimal amount of key data: study design, intervention, fidelity measure, outcome(s), outcome measure(s), and to identify any potentially relevant additional studies. This form served two purposes. First, it enabled the team to gain a quick overview of the studies’ interventions and outcomes in order to test their potential for synthesis. Given the number and apparent heterogeneity of the studies, more in-depth data extraction would have been time-consuming and potentially of little added value. Second, reference tracking was used to identify additional potentially relevant studies. The new forms were piloted by all six members of the team and questions were resolved at team meetings. The data extraction was then completed for each study by a single reviewer. Double data extraction was rejected because of the number of selected studies and the limited potential for synthesis. This process provided a necessary overview of the available literature.

Data synthesis

Despite the apparent heterogeneity of the studies, a second stage of data extraction was then undertaken by the research associate leading on this question (CC) when it became apparent that some studies demonstrated sufficient homogeneity to be grouped together for synthesis. This additional data extraction compiled information on country, the unit of analysis, staff involved in the intervention, sample size, length of the study, bottom-line results, staff outcomes and whether staff variables had been used as moderators or mediators.

Study design was the principal method of quality weighting studies in the synthesis. RCTs were considered to be the best-quality studies because this design limits the scope for bias in the study. 48 The next best studies had a quasi-experimental design: a research design that resembles a RCT but has no random assignment. These were followed by prospective studies and, finally, retrospective studies were considered to be the longitudinal design most vulnerable to bias. This hierarchy was applied in the synthesis. For example, the findings of a RCT are given greater weight than those of all other study designs examining the same intervention or outcomes, and the findings of a quasi-experimental study are given greater weight than a prospective or retrospective study. As far as possible, the review used authors’ own descriptions of their study’s design in order to limit potential bias in the labelling of studies as particular designs by reviewers in the project team.

This provided all the data required for a worthwhile summary and synthesis of the studies.

Work design

All of these studies consider interventions or practices changing or affecting the content of a job or the way it is to be done. The term ‘work design’ is rarely used in the literature relating to HRM in the health sector and is not used to describe any of the interventions in this sample of longitudinal research. However, the interventions fall under the heading of ‘work or job design’ because they all involve the redesign or extension of job content. The absence of any existing systematic review or meta-analysis of health sector studies described as work design may be explained by difficulties in creating a clear, structured definition of work design or operationalising the concept as a search strategy. It may also have been prevented by the potentially limited number of relevant studies and the heterogeneity of their interventions and outcomes, as demonstrated by this sample. In fact, there is only one meta-analysis covering job enrichment, an element of work design, in the non-health sector. 281

Staffing level and working hours

These studies consider research into two elements of staffing practice: staffing level and staff working hours. The staffing level literature considers whether increased ratios of staff–patients improve patient care outcomes or whether decreased ratios have an adverse effect on these outcomes. This ratio may relate either to numbers of staff or to grades of staff, such as registered nurses (RNs). There is one systematic review of studies of staffing levels in the health sector. 295 Two of the longitudinal studies described below have been included in this previous systematic review. 296,297 However, the majority have not been included in existing systematic reviews and therefore add to the review literature on this topic. There are no systematic reviews or meta-analyses on staffing levels in the non-health sector HRM literature.

The working hours literature considers both the variable of total numbers of hours worked and interventions that affect the number of hours worked by health professionals, especially reductions or rescheduling. There are two meta-analyses that include studies of health sector working hours, reflecting the sizeable number of studies in this field. 298,299 However, none of the studies described below has been included in these existing reviews. This sample therefore adds to the review literature on this topic. There is only one meta-analysis in the non-health sector HRM literature reviewing studies on a similar subject. 300

Introduction of a new type of worker to existing teams or services

The staffing category includes skill mix at work, which involves ‘practices affecting enhancement (increasing depth of job by extending roles of skills), substitution (working across professional divides or exchanging one type of worker for another), delegation, innovation (creating new jobs by introducing a new type of worker), transfer of services, relocation of staff and services, and liaison (using specialists in one sector to educate and support staff in another)’. 318 The following synthesis focuses on studies researching innovation: the introduction of a new type of worker to existing teams or services. There are five recent systematic reviews or meta-analyses on staffing in the health sector workforce, reflecting both the sizeable number of studies and the growing interest in this field. 295,319–322 However, only one of the studies described below has been included in any of these systematic reviews. 323 There is no meta-analysis in the non-health sector HRM literature reviewing studies on a similar subject.

Substitution of one type of staff for another

The following synthesis focuses on studies researching substitution. There are five recent systematic reviews or meta-analyses on staffing in the health sector workforce, reflecting both the sizeable number of studies and growing interest in this field. 295,319–322 However, only three of the studies described below have been included in any of these systematic reviews,296,334,335 so this sample also clearly adds to the review literature on this topic. There is no meta-analysis in the non-health sector HRM literature reviewing studies on a similar subject.

Transfer or relocation of staff and services

The following synthesis focuses on studies researching transfer or relocation of services. There are five recent systematic reviews or meta-analyses on staffing in the health sector workforce, reflecting the sizeable number of studies in this field and the growing interest in this field. 295,319–322 However, only one of the studies described below has been included in any of these systematic reviews. 335 There is no meta-analysis in the non-health sector HRM literature reviewing studies on a similar subject.

Training

All of these studies consider interventions that involve some form of training or education for health staff and its impact on patient and staff outcomes. Studies have only been included if the entire practice was training or if training was a primary element of an HRM practice. A number of studies have been excluded because education or training was only one element of a multifaceted intervention in which none of the other components could be described as falling within the field of HRM. 361,362 There are five existing systematic reviews or meta-analyses of training and development interventions in the health sector. Three examine forms of continuing medical education for physicians and their impact on physician performance and health-care outcomes. 363–365 The two other reviews examine systems of continuing professional development to enhance the recruitment and retention of specific health professionals. 299,366 There are also three meta-analyses of training practices from the non-health sector. These focus on elements of training that are very different from those examined for health, such as altering organisational behaviour,367 mentoring368 and evaluating the effectiveness of organisational training. 369 None of the studies reviewed below appear in any of these systematic reviews. A narrative synthesis has been undertaken due to the general heterogeneity of the studies in this sample.

Compensation and payment

These studies consider payment systems, compensation and reward practices that seek to affect patient and service outcomes. A narrative synthesis is preferred because of the very small sample of studies identified. However, this topic has been the subject of seven systematic reviews. Five of these reviews have compared the impact of different types of remuneration for physicians on a range of patient and service outcomes. 382–386 The remaining reviews consider the impact of target payments on immunisation rates and financial incentives on retention rates among rural physicians. 387–388 No health professionals other than physicians have been considered by these reviews. None of the studies described below appears in any of these systematic reviews. There are also five reviews focusing on compensation and rewards in the non-health sector, such as financial rewards, social rewards and profit sharing. 367,389–392

Employee involvement

All of these studies consider interventions or practices that seek to involve staff in decision-making. The absence of any systematic review or meta-analysis of studies of this intervention in the health sector may be explained by difficulties in operationalising the concept of employee involvement for the purposes of systematic review. It may also have been prevented by the potentially limited number of relevant studies and the heterogeneity of their interventions and outcomes, as demonstrated by this sample. Tellingly, there are also no distinct meta-analyses of employee involvement in the HRM literature of the non-health sector.

Performance management

All of these studies consider interventions or practices that evaluate and feed back on the performance of health professionals. A narrative synthesis is preferred here despite the homogeneity of some clusters of studies in this sample. No systematic reviews of performance management in the health sector were found, but there are several meta-analyses of performance management in the non-health sector HRM literature. These cover feedback,400,401 goal setting401 and supervisor ratings. 402,403