Objective To assess the methodological and reporting quality of systematic reviews that comprise the American Psychiatric Association (APA) Practice Guideline for the Treatment of Patients with Schizophrenia and to determine the extent to which results from Cochrane systematic reviews published after guideline development would alter or confirm current recommendations.
Participants Systematic reviews that underpinned recommendations in the APA guidelines and Cochrane systematic reviews.
Main outcome Three independent reviewers scored all systematic reviews referenced in the guideline for quality and reporting using AMSTAR and PRISMA checklist, respectively. Items in both tools were individually graded and compared to identify consistently low-performing areas within the systematic reviews. Post hoc analysis of the Cochrane systematic reviews since the latest revision of APA’s guidelines were performed to determine whether their findings were congruent with recent recommendations.
Results The mean score of the 57 reviews on the PRISMA checklist was 70%. The mean AMSTAR score was 6.8, correlating with a moderate quality score. Post hoc analysis revealed that 171 Cochrane reviews had been published since the APA guideline release. Only half of the reviews of pharmacological interventions confirmed current recommendations.
Conclusions and relevance The methodological quality of the systematic reviews included in the APA guideline was deficient in key areas. Our study brings to light the importance of using high-quality evidence in the development of clinical practice guidelines. An updated APA guideline (last updated in 2009) is necessary to provide the highest quality treatment recommendations for clinicians in the management of schizophrenia.
Trial registration number UMIN-CTR, UMIN000023099.
- systematic reviews
- american psychiatry association
Statistics from Altmetric.com
What is already known about this subject?
Clinical practice guidelines for schizophrenia have been developed to provide evidence-based treatment recommendations for clinicians; however, they have been a source of continued controversy in recent years.
The American Psychiatric Association (APA) Practice Guideline for the Treatment of Patients with Schizophrenia, first developed in 2004 and updated in September of 2009, do not include evidence for at least five new antipsychotic medications and at least five long-acting injectable formulations, all of which have been approved by the Food and Drug Administration since the APA’s most recent revision.
The APA recognises the importance of systematic reviews in guideline formulation.
Ensuring that studies included in systematic reviews are unbiased is important in providing the highest quality data when producing treatment-based guidelines.
What are the new findings?
Review of the methodological and reporting quality of the evidence comprising the most recent APA Guidelines for the Treatment of Schizophrenia highlights key areas of concern, specifically within the systematic reviews within the guideline recommendations.
Improvements in systematic reviews methodology would strengthen the guideline recommendations made by the APA.
The APA Treatment Guidelines are outdated. Only 50% of Cochrane Database systematic reviews published since the last updated version of the APA guidelines confirmed current pharmacological treatment recommendations by the APA.
How might it impact clinical practice in the foreseeable future?
Clinical practice guidelines play a vital role in assisting physicians in clinical settings; however, the use of clinical practice guidelines in psychiatry is low. One way to overcome this barrier is by producing high-quality systematic reviews and using them as supporting evidence in guideline updates. Researchers can improve the methodological rigour of systematic reviews by following the PRISMA guidelines and consulting AMSTAR.
To reduce the time and conflicts health providers face when implementing clinical practice guideline strategies, there needs to be a pre-established framework. Fischer et al provide an eight-step framework that healthcare facilities can use to establish and monitor the use of clinical practice guidelines in their establishment. By implementing these strategies, there will be a strong improvement in systematic reviews, clinical practice guidelines and their use in clinical settings.
Clinical practice guidelines for schizophrenia have been developed to provide evidence-based treatment recommendations for clinicians; however, they have been a source of continued controversy in recent years. A study comparing the methodological quality of 24 international schizophrenia guidelines noted that most were moderate at best and reported large variations in the type and frequency of psychosocial interventions.1 The American Psychiatric Association (APA) Practice Guideline for the Treatment of Patients with Schizophrenia—first developed in 2004 and updated in September of 20092—do not include evidence for at least five new antipsychotic medications and at least five long-acting injectable formulations, all of which have been approved by the Food and Drug Administration since the APA’s most recent revision.3
To further complicate matters, a number of psychosocial therapies for schizophrenia have been developed or refined since the 2009 guideline update, and this body of evidence is not reflected in the current recommendations.4 5 The more recent UK National Institute for Health and Care Excellence guidelines have been criticised for being biased towards psychosocial interventions since only 24% of its recommendations relate to pharmaceutical use.6 By comparison, 60% of recommendations in the Scottish Intercollegiate Guidelines Network guidelines are solely based on pharmacological therapies. These variations call into question the methodological quality of evidence used to inform the recommendations, whether they are based on the latest available evidence, and which types of interventions are most effective.
Recognising limitations in existing APA guidelines, the APA has committed to improving the development of guidelines by establishing an updated policy based on the Institutes of Medicine guideline standards. The APA recognises the importance of systematic reviews in guideline formulation.7 This policy addresses improved transparency, conflicts of interest and advances the use of systematic reviews. The policy established the Systematic Review Group to perform reviews on specific clinical questions, and to rate the quality of evidence when no Cochrane or Agency for Healthcare Research and Quality reviews exists for consideration. While these efforts will likely improve future psychiatric clinical practice guidelines, little is known about the quality of the evidence of current guidelines, such as those developed for schizophrenia. Here, we evaluate the methodological and reporting quality of systematic reviews informing the current APA Practice Guideline for the Treatment of Patients with Schizophrenia. We also compared the quality of systematic reviews of psychosocial interventions with those of pharmacological interventions to determine which are based on higher-quality evidence. Finally, we examine Cochrane systematic reviews not included in the APA guidelines to identify omitted evidence that might affect or alter current recommendations.
Protocol development and registration
All search strategy, eligibility criteria and data abstractions for this study were based on a protocol developed and piloted a priori. This study did not meet the regulatory definition of human subject research as defined in 45 CFR 46.102(d) and (f) of the Department of Health and Human Services’ Code of Federal Regulations,8 and thus was not subject to Institutional Review Board oversight. Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA)9 and Statistical Analyses and Methods in the Published Literature guidelines10 for descriptive statistics were applied when relevant. This study is registered on the University hospital Medical Information Network Clinical Trial Registry.
Data sources and eligibility
We defined the term “clinical practice guideline” as “statements that include recommendations intended to optimize patient care that are informed by a systematic review of evidence and an assessment of the benefits and harms of alternative care options”, using the definition developed by the Institute of Medicine.7 We sought systematic reviews used as references specifically in the Practice Guideline for the Treatment of Patients with Schizophrenia, second edition,2 published in 2004 by the APA, and the focused update published in 2009,3 which represent the most current guidelines for schizophrenia treatment published within the USA. This guideline and its focused update was the only guideline to be included in the study, and was the only guideline to meet a priori inclusion criteria. These inclusion criteria required that clinical practice guidelines have been recognised by a national or governmental body and/or a professional organisation, be published in English and have an accessible reference list.
Following retrieval of the full text of the guideline, a team of three reviewers (GA, CH, AD) individually and in blinded fashion searched the reference section with keyword searches to identify any systematic reviews or meta-analyses. These reviewers also hand-searched the reference section of the guideline to identify any further articles likely to represent systematic reviews or meta-analyses and retrieved all eligible articles in full text. Disagreements were solved by consensus. When disagreements could not be resolved by consensus, an additional reviewer was consulted. The term “systematic review” was defined a priori as “A scientific investigation that focuses on a specific question and uses explicit, prespecified scientific methods to identify, select, assess, and summarize the findings of similar but separate studies. It may include a quantitative synthesis (meta-analysis), depending on the available data”. This definition was based on that developed by the Institute of Medicine.8 It was developed with the intent of being as inclusive as possible to avoid bias. In addition, we avoided definitions setting standards for numbers of databases searched or performing meta-analyses, to avoid conflict with our assessment tools. To be considered for inclusion, a systematic review must have been referenced in the guideline. Only English-language systematic reviews were used. Animal studies were likewise excluded. We attempted to find the specific version of systematic reviews used in the guideline. If the version of the systematic review used in the guideline had since been updated, or was otherwise unavailable, that systematic review was excluded. Figure 1 shows the process of study selection during the search process.
Data abstraction and scoring
Prior to coding, reviewers were trained via detailed tutorials outlining the abstraction process. Reviewers then completed a piloted practice abstraction form to familiarise themselves with the abstraction procedures. The same team of reviewers (GA, CH, AD) independently abstracted data from a subset of all eligible studies using piloted abstraction forms. Following abstraction, each form was validated by a second reviewer in the team. Disagreements were solved by consensus, as above. This method was followed throughout the abstraction process. For each eligible systematic review, reviewers abstracted study characteristics for year of publication, participant population, intervention, number of included studies, sample size across all included studies and study design of included studies (ie, randomised control trials, cohort etc). The same reviewers then independently scored each study according to the PRISMA checklist and A Measurement Tool to Assess Systematic Reviews (AMSTAR)11 to assess clarity of reporting and methodological quality, respectively. Scoring for AMSTAR was based on the method described by Sharif et al,12 and the recommended changes to AMSTAR suggested by Burda et al.13 Scoring for the PRISMA checklist was based on guidelines outlined in Liberati et al.14
To assess the methodological quality of eligible studies, we used AMSTAR,15 which consists of 11 items designed to assess various aspects representative of methodological quality in a systematic review. AMSTAR has been shown to be a reliable and valid tool on several occasions, and has been shown to be more reliable than other methods of assessing methodological quality, such as the Overview of Quality Assessment Questionnaire and revised AMSTAR (R-AMSTAR).15–17 We used AMSTAR, rather than R-AMSTAR, as AMSTAR has been shown to have more feasibility over R-AMSTAR, and is more easily applied consistently, while subjectivity and repetition inherent in R-AMSTAR limits the usefulness of that tool.18 It is worthwhile to note that AMSTAR 2 released in November 2017 but was not used in this study as it came out after data analysis in this study took place.
We applied recommended revisions made by Burda et al 13 to our AMSTAR. These changes focus on improving validity, reliability and usability in assessing methodological quality, and include changes in order of items, wording of item titles and instructions, and modifications to the focus of original items 7, 8 and 11. In addition, these recommendations address aspects noted to be problematic in numerous studies, and improve specificity to methodological quality over quality of reporting or risk of bias.18 However, the recommended additional item described by Burda et al was not included, as subgroup analyses are not applicable to all systematic reviews and meta-analyses, and the addition of an additional item complicates scoring of the tool. Where modified instructions were further unclear, additional instructions were provided to reviewers.
Scoring of AMSTAR was performed according to the method described by Sharif et al.12 Answer choices include ‘yes’ if a systematic review met all criteria outlined in item instructions, ‘no’ if the review met none of the criteria or ‘cannot answer’ if the review met some, but not all, of the criteria outlined. A ‘not applicable’ response was retained for item 10 of the tool (publication bias), to account for of diagnostic accuracy or those with fewer than 10 included studies. A systematic review was awarded one point for every ‘yes’ answer, and zero points for every ‘no’ or ‘cannot answer’, with a maximum score of 11. ‘Low’, ‘moderate’ and ‘high’ levels of methodological quality were assigned based on total AMSTAR scores of 0–3, 4–7 and 8–11, respectively.16 Total scores for each systematic review were averaged across the clinical practice guideline, giving a global mean AMSTAR score.
To evaluate clarity of reporting, we scored each eligible systematic review according to the PRISMA checklist. The checklist comprises 27 items evaluating the clarity and completeness of reporting of systematic reviews and is considered best practice for systematic review reporting.19 20 Although traditionally used as a tool for systematic reviewers to improve their own practices, this checklist is acknowledged for its use in critically appraising systematic reviews and meta-analyses.19 21 However, this tool is not applicable in the assessment of the methodological quality of systematic reviews and meta-analyses, necessitating the use of additional tools for this purpose. Scoring for this tool was based on guidelines presented in Liberati et al.19 Reviewers assessed whether each systematic review met full criteria, no criteria or partial criteria. Reviews meeting full criteria were awarded one point, with half a point being given for partial criteria, and no points for those that met no criteria. Reviewers were asked to perform grading via the PRISMA checklist before AMSTAR to reduce bias introduced through endorsement of PRISMA items in AMSTAR. Total scores were calculated as a proportion of criteria met for each systematic review and were averaged to provide a global mean with standard deviations Clinical Practice Guideline.
In addition to the primary study conducted, we performed a post hoc investigation evaluating the extent to which Cochrane systematic reviews authored by the Cochrane Schizophrenia Group have produced evidence that could be used for Practice Guideline for the Treatment of Patients with Schizophrenia updates. We retrieved all Cochrane reviews produced by this group from September 2009 until July 2016. September 2009 was chosen because it was the publication date of the latest guideline addendum. Each Cochrane review was categorised into one of three treatment groups: psychosocial, pharmaceutical and other. Of the pharmaceutical studies, we compared the results of the Cochrane studies to the APA guidelines. We determined whether the conclusion of the drug studied in the Cochrane Schizophrenia Group was altered, not mentioned, or comparable with the APA guidelines. We repeated this process using the studies categorised as psychosocial and other.
We searched all references within the most current APA guideline for treatment of schizophrenia and the focused update to the guidelines. Of 1491 total references retrieved from these guidelines, 682 articles were excluded immediately as primary studies. The remaining 709 were screened using our inclusion criteria: 573 were excluded as non-systematic reviews or other publications; 136 studies proceeded to full-text screening. After review, a total of 57 studies were included for scoring. The characteristics of each study included are outlined in table 1. Publication dates of the systematic reviews assessed range from 1990 to 2009, with an average date of publication being 2002. The average population per study was 34, with randomised controlled trials being the most common study design used. Unsurprisingly, the most common patient population studied was patients diagnosed with schizophrenia.
Detailed PRISMA and AMSTAR scores for each systematic review are available in online supplementary tables 1 and 2. The total PRISMA quality rating of the APA guidelines was a 70%. Total PRISMA scores for each systematic review ranged from 31% to 94% and from 1 to 11 on AMSTAR. The five systematic reviews with the lowest PRISMA values also had an average AMSTAR quality assessment rating of 2.6. Conversely, the top five reviews with the highest scores on PRISMA averaged a score of 9.4 of 11 on the AMSTAR checklist. Nearly one quarter of the systematic reviews (n=13) had a total PRISMA score below 60%, with the lowest score being 31%. The highest individual PRISMA rating was 94% and was recorded in two of the systematic reviews.
In addition to calculating scores for each systematic review, the value of each individual item used within AMSTAR and PRISMA was analysed. An individual item’s average score was calculated across all the included systematic reviews found in the APA guidelines. The scores for these individual items ranged from 25% to 100% on the PRISMA survey (27 items in total). Tables 2 and 3 depict the mean individual score for each individual item in AMSTAR and PRISMA, respectively, indicating on average how much each item contributed to the score of all included reviews. This value illustrates the cumulative strengths and weaknesses of systematic reviews in the APA guidelines. Evident weaknesses in PRISMA scores across the study sample include scores produced such as in item 5 (a score of 25%), which assesses the use of a review protocol and a registration number by the systematic reviewer. Item 23, which assesses the presence of additional analytical results in studies, had an average individual score of 42%. Item 3 of PRISMA, which asks whether a review provided information in the context of what is already known, had an average individual score of 100%, indicating that it contributed to the individual scores of all 57 reviews. Item 26 was another strength across reviews, addressing the use of general interpretations of the study results in the context of other evidence and implications for future research, and resulted in an average individual score of 98%.
Further analysis of the 57 reviews revealed an average AMSTAR score of 6.8 of 11. Twelve per cent (n=6) of the systematic reviews were found to be of low methodological quality, scoring at or below 3 of 11 on AMSTAR. The same six reviews scored an average of 44% on PRISMA. Forty-two per cent (n=24) of the systematic reviews were found to be of moderate methodological quality, with scores between 4 and 7, while 47% (n=27) received a high-quality assessment, having scores between 8 and 11. Results from AMSTAR also indicated common weak points in the global scores of items across all included reviews.
The inclusion of grey literature (assessed by item 3 of AMSTAR) in the reviews was lacking, as indicated by its individually itemised score of 46%. Analysis of the use of duplicate study selection and extraction among the reviewers (item 4) revealed an average individual score of 33%. Data analysis of AMSTAR items also revealed item 5, which assesses the use of listed studies used in the systematic reviews, to have a value of 44%.
Data represented in figure 2 show the AMSTAR scores for each of the tool’s 11 items as plotted on a radar graph, with the lower scores more centrally located compared with higher scores found towards the periphery. Tables 2 and 3 describe the average values of each item in AMSTAR and PRISMA, respectively, across all the systematic reviews(figure 3). Online supplementary tables 1 and 2 detail each systematic review or meta-analyses scores across rows, with averaged scores of each individual item found at the bottom of each column.
Post hoc investigation results
In addition to our primary analysis, we conducted a post hoc investigation to determine whether the results of Cochrane reviews produced by the Cochrane Schizophrenia Group since publication of the APA treatment guidelines, if used as supporting evidence, would alter or confirm the current APA recommendations. Our search yielded 171 systematic reviews published between October 2009 and July 2016. Of the reviews not already included in the APA guideline, 19% (n=33) involved treatments using psychosocial interventions, whereas 69% (n=118) investigated pharmaceutical-based therapies. The remaining reviews (15%, n=25) involved other interventions (eg, acupuncture).
Of the systematic reviews of pharmaceutical treatments, 17 were completed studies that investigated medications also mentioned in the APA guidelines. These studies were fully accessible for analysis on the Cochrane Database and are outlined in table 4. Of these, more than half (n=10) stated that there was insufficient evidence to draw conclusions based on the available evidence. Approximately one-third (n=5) reported at least slight improvements to schizophrenia symptoms following treatment, and one found an increase in adverse events following treatment. The remaining 58 systematic reviews categorised as psychosocial were narrowed to eight studies relevant to our topic. Of the eight studies, only 50% (n=4) produced similar conclusions in the Cochrane study and the APA schizophrenia guideline. Seven systematic reviews not categorised as pharmaceutical or psychotherapeutic treatment options were classified as other and are shown in online supplementary table 3. The treatments under investigation in these studies were not mentioned in the APA guideline. However, the results of the studies did not reveal any evidence to warrant a change in the current treatment guidelines.
In summary, we were able to correspond results from 17 systematic reviews from the Cochrane Database regarding pharmaceutical treatment results to treatment recommendations from the APA guidelines. Of these pharmaceutical treatment recommendations, 18% (n=3) did not agree, specifically the recommendations regarding ACE inhibitors, the use of aripiprazole and the role of cognitive-behavioural therapy in the treatment of schizophrenia. Six of the 17 studies could not be compared, owing to inconsistencies between the Cochrane reviews and the APA recommendations (eg, Cochrane review evaluated the use of ACE inhibitors as a primary treatment, whereas the APA guideline referenced ACE inhibitors as an adjunctive therapy). Eight systematic reviews presented findings that were consistent with the treatment recommendations found in the APA guidelines.
The quality of reporting of systematic reviews that comprise the APA Practice Guidelines for the Treatment of Patients with Schizophrenia is deficient in key areas. Two particular areas of weakness were found: systematic reviewers did not assess the risk of bias within their studies and the results of additional analyses were not reported. Our analysis also demonstrated the a priori use of a protocol as an area for needed improvement.
Results of this study also suggest that the latest available evidence is not being used. Our results found 171 Cochrane systematic reviews produced by the Cochrane Schizophrenia Group were not referenced in the practice guideline. There is strong evidence that Cochrane reviews apply a more stringent methodology than non-Cochrane reviews,21–24 and thus, these reviews should be carefully considered for use in clinical practice guidelines. Previous studies have indicated the underuse of Cochrane reviews in clinical practice guideline recommendations. For instance, Brok et al 25 reported that neonatal guidelines rarely used evidence from Cochrane reviews in their development. Silagy et al 26 found that evidence from Cochrane systematic reviews could have been used in 39%–73% of the recommendations for smoking cessation guidelines, but were only used in 0%–36% of the actual recommendations. Of the 171 Cochrane reviews, we found 17 new medications evaluated in these reviews that were not mentioned in APA guideline recommendations. In addition to the pharmaceutical studies, we found eight psychosocial studies or other studies that were not referenced in the guideline.
Implications of results
Without adequate evaluation of biases, such as publication bias, the risk of skewed evidence increases. Outcomes such as summary effect sizes could potentially be exaggerated if based only on published studies, making grey literature searches necessary. Improvements can be made, however, such as requiring journal editors and peer reviewers to use the PRISMA checklist, which contains an item related to publication bias, during data selection. Ensuring that studies included in systematic reviews are unbiased is important in providing the highest quality data when producing treatment-based guidelines such as the APA schizophrenia guidelines.2 27
In 2011 (after publication of the APA schizophrenia guidelines), the APA adopted a new process for clinical practice guideline development7 with the intent of meeting the Institute of Medicine’s standards of guideline development. This new process involves mechanisms to improve transparency and methodological rigour, including ongoing monitoring of the latest available evidence. According to the APA, “After guideline publication, new evidence is identified by continuous monitoring of the literature, and guideline modules are updated in a targeted fashion if there are important changes in the supporting evidence”.28 Our results suggest that the clinical practice guidelines produced by the APA do not reflect current clinical practice and bring to light areas of potential improvement in future revisions. Of the 1491 references listed in the APA guidelines, 57 were identified as systematic reviews. The relatively small number of systematic reviews cited, and the high omission of relevant Cochrane reviews, may add scepticism by clinicians who intend to use the guideline in treating patients with schizophrenia.
Strengths and limitations
The strength of this study resides in the tools used to evaluate systematic reviews. The use of PRISMA and AMSTAR provided a framework for assessing the methodological quality of systematic reviews. Another strength of this study is the manner in which data were abstracted and scored. Three researchers independently abstracted data, which were then validated by an additional reviewer. If there was a dispute during validation, a third reviewer was consulted for resolution. This methodology provided a stable framework to reduce any error that may have been introduced throughout the process. A final strength of this study is the post hoc analysis. Evaluating 171 systematic reviews from the Cochrane Collaboration emphasises that there are high-quality studies available to update the current APA schizophrenia guidelines.
There are multiple limitations to our study. Due to the nature of the search performed, only systematic reviews included in the clinical practice guidelines were scored. If articles did not identify themselves clearly as systematic reviews or meta-analyses, they may have been missed despite measures taken to limit this omission. In addition to limitations in systematic review identification, we found that a small number of systematic reviews had been updated since their inclusion in the APA guideline. Furthermore, it is important to acknowledge the limitations of systematic reviews. Reviews may be prone to bias due to the host of factors that have been detailed in this study. Nonetheless, systematic reviews have also contributed greatly to scientific knowledge, and as research studies continue to be conducted at increasingly higher rates, there will be a need to synthesise findings from these studies in an informed manner. Finally, it is possible that if different tools were used to evaluate reporting and methodological quality from AMSTAR and PRISMA, there would be a change in the results.
Implications for research practice
Clinical practice guidelines play a vital role in assisting physicians in clinical settings. They allow for a unified approach to various clinical situations. However, as Fischer et al 29 state, the use of clinical practice guidelines in psychiatry is low. One way to overcome this barrier is by producing high-quality systematic reviews and using them as supporting evidence in guideline updates. Researchers can improve the methodological rigour of systematic reviews by following the PRISMA guidelines and consulting AMSTAR. An alternative would be to use studies produced by the Cochrane Collaboration. These studies are proven to be high quality and provide updates to the current guidelines. Gaebel et al 1 showed that, in Germany, after the schizophrenia guidelines were updated, there was an improvement in quality. The hope is that by improving the quality of the guidelines, there will be better adherence to their recommendations. To reduce the time and conflicts health providers face when implementing clinical practice guideline strategies, there needs to be a pre-established framework. Fischer et al 29 provide an eight-step framework that healthcare facilities can use to establish and monitor the use of clinical practice guidelines in their establishment. By implementing these strategies, there will be a strong improvement in systematic reviews, clinical practice guidelines and their use in clinical settings.
As clinical practice guidelines become more influential in healthcare, the methodological quality of the studies that comprise them must continually be assessed. It is the quality of the reviews in the guidelines that ultimately determines the value of the recommendations they establish. We suggest that resources such as the PRISMA checklist and AMSTAR should be standard components in evaluating methodological quality.
Contributors GA: data curation, formal analysis, investigation, methodology, writing—original draft, writing—review and editing. CH: data curation, formal analysis, investigation, writing—review and editing. AD—data curation, formal analysis, investigation, writing—review and editing. ALJ: investigation, methodology, writing—review and editing, corresponding author. JB: conceptualisation, supervision, writing—review and editing. YB: data curation, formal analysis, investigation, writing—review and editing. MH: supervision, writing—review and editing. AL: supervision, writing—review and editing. PS—conceptualisation, investigation, writing—review and editing. MV: conceptualisation, project administration, supervision, writing—review and editing.
Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.
Competing interests None declared.
Patient consent for publication Not required.
Provenance and peer review Not commissioned; externally peer reviewed.
Data availability statement Data are publicly available on figshare.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.