PMCCPMCCPMCC

Search tips
Search criteria 

Advanced

 
Logo of heartHeartVisit this articleSubmit a manuscriptReceive email alertsContact usBMJ
 
Heart. 2007 October; 93(10): 1256–1262.
Published online 2007 February 19. doi:  10.1136/hrt.2006.110171
PMCID: PMC2000947

Inequalities in the primary care of patients with coronary heart disease and serious mental health problems: a cross‐sectional study

Abstract

Objective

To determine whether UK patients with coronary heart disease (CHD) who also have schizophrenia or bipolar disorder are less likely to receive primary care in accordance with the agreed national standards of the UK than patients without these mental health problems.

Design

Cross‐sectional study.

Setting

485 UK general practices contributing anonymised medical records of over 3.26 million patients to the QRESEARCH database.

Participants

127 932 patients with CHD of whom 701 had a diagnosis of schizophrenia or bipolar disorder.

Main outcome measures

The relative risks of receiving statin medication and each of the CHD care indicators defined in the UK General Medical Services contract, for patients with schizophrenia or bipolar disorder compared with patients with neither condition. The results were adjusted for age, sex, deprivation, diabetes, stroke and smoking status, and allowed for clustering by practice.

Results

Patients with schizophrenia were 15% less likely to have a recent prescription for a statin (95% CI 8% to 20%) and 7% less likely to have a recent record of cholesterol level (95% CI 3% to 11%). There were no significant differences in the adjusted analyses between mental health groups on recording smoking status, advising on smoking cessation, recording blood pressure, achieving target blood pressure or cholesterol values, or prescribing aspirin, antiplatelets, anticoagulants or β blockers.

Conclusions

Although the majority of CHD care indicators are achieved equally for patients who also have a serious mental health problem, there is a shortfall in identifying and treating raised cholesterol among patients with schizophrenia, despite their higher level of risk factors.

Coronary heart disease (CHD) accounts for more deaths worldwide than all causes of cancer combined,1 and it is the main medical factor associated with disability in Europe.2 Statins have a proven role in reducing further cardiovascular events and improving survival among people with existing CHD.3,4,5 The new General Medical Services(GMS) contract for UK primary care physicians has set national standards for quality care for CHD based on a number of indicators including cholesterol and blood pressure monitoring, advice and referral on smoking, and prescribing appropriate medications.6

CHD and coronary risk factors such as smoking, obesity and diabetes are more common among people with schizophrenia and bipolar disorder than in the remaining population,7 and contribute to their shorter life expectancy.8,9,10 Although the overall consultation rates with primary care physicians in the UK are three to four times higher in patients with serious mental health problems,11 there is some evidence that they are less likely to be offered health‐promotion interventions such as blood pressure checks or prescriptions for leisure.12,13 It is not clear from previous research whether patients with serious mental health problems experience inequalities in care for CHD, and differences between healthcare systems may contribute to the variation in findings. Two US studies found deficits in care for patients with schizophrenia after acute myocardial infarction, contributing to an excess in mortality,14,15 but other studies carried out within the US Veterans Health Administration (VA) system found no such shortfall.16,17

We undertook a study within a very large representative population in UK primary care to determine whether patients with CHD who also have schizophrenia or bipolar disorder were less likely than patients without these mental health problems to receive good quality care in accordance with the agreed national standards of the UK.

Methods

Data

The source of data was the QRESEARCH database which contains the anonymised electronic healthcare records of over 9 million patients ever registered with 499 general practices throughout England, Wales, Scotland and Northern Ireland. QRESEARCH is an aggregated patient‐level database derived from a representative sample of 6% of all UK general practices. Consent to provide data is sought from practices using the EMIS medical records system (Egton Medical Information Systems Limited, Leeds, UK), and detailed analyses have shown these practices to be somewhat larger than non‐participating practices but in all other respects very similar.18 The information recorded on the database includes patient demographics (year of birth, sex, socioeconomic data derived from the UK 2001 census), characteristics (height, weight and smoking status), symptoms, clinical diagnoses, consultations, referrals, prescribed medication and the results of investigations. QRESEARCH V.8.0 database was used for this analysis. The database has been validated by comparing birth rates, death rates, consultation rates, prevalence and mortality rates with other data sources including the General Household Survey and the General Practice Research Database.19 The age–sex structure of the population has been compared with that reported in the 2001 census. There was good correspondence for all these measures, although our population is slightly older and our prevalence figures are marginally higher than less recent data.20 We have also compared practices taking part in regional research networks on these and other measures, and have found a good correspondence.21 The prevalence of schizophrenia by age and sex matches closely with the figures published by the Office for National Statistics,19 and the sex‐specific prevalence of bipolar disorder is close to that recently reported for UK general practices.22

For this study, we included general practices if they had been using their current EMIS computer system before 1 April 2003 and had the data uploaded after 1 April 2005. These criteria were chosen to ensure that the data were as complete as possible for the study period. Patients were included if they were registered with an eligible practice by 1 April 2005, had been registered with this practice for the previous 12 months and were not registered as a temporary resident. Patients were included in the analysis if they had a computer‐recorded diagnosis of CHD before 1 April 2005 and were aged [gt-or-equal, slanted]25 years.

The following information was extracted for each patient:

  • personal and area demographics: age, sex and area Townsend deprivation score (grouped by quintile, with quintile 5 being the most deprived);
  • serious mental health problems: a recorded diagnosis of schizophrenia or bipolar disorder before 1 April 2005 (see appendix A for code list);
  • comorbid physical conditions: diagnosis of stroke or diabetes before 1 April 2005, and current smoking status;
  • CHD management: any prescription of statins before 1 April 2005; the CHD indicators from the GMS contract (which was implemented in UK general practice on 1 April 2004, 12 months before the analysis date for this study).6 These indicators were calculated according to the rule set for the quality and outcomes framework, and comprise CHD2, patients with newly diagnosed angina referred for exercise testing and/or specialist assessment; CHD3, smoking status recorded in the previous 15 months, or at least once if never smoked; CHD4, smoking cessation advice in the previous 15 months; CHD5: blood pressure recorded in the previous15 months; CHD6, last blood pressure value [less-than-or-eq, slant]150/90 mm Hg; CHD 7, cholesterol blood test recorded in the previous 15 months; CHD 8, last cholesterol value [less-than-or-eq, slant]5 mmol/l; CHD9, treatment with aspirin, antiplatelet or anticoagulant in the previous 15 months; and CHD10, current treatment with β blocker. For each indicator, there is a variable showing eligibility (eg, only current smokers are eligible for smoking cessation advice), and a variable showing achievement.

Analysis

We compared the outcomes between three groups of patients: those with schizophrenia, those with bipolar disorder and those with neither mental health condition. Patients with diagnoses of both schizophrenia and bipolar disorder were included in the schizophrenia group, and we carried out a sensitivity analysis to determine whether this classification had a significant effect on the results. In view of the relatively high proportion of patients achieving these indicators, we compared the groups using relative risks (RRs) rather than odds ratios, and estimated these using Poisson regression with robust error variance.23 RRs were calculated first unadjusted and then adjusted for age group, sex, deprivation (quintiles), comorbid stroke and diabetes, and allowing for clustering by practice. Adjustment was also made for smoking except in the models for CHD3 and CHD4, where only current smokers are eligible. We selected a p value of <0.01 (two tailed) as statistically significant.

We estimated the numbers of patients in the UK with CHD and the numbers with CHD who also have either schizophrenia or bipolar disorder by determining the prevalence in each 5 year age–sex band in our data and standardising it to the UK population (2001 census). These values were used to estimate the numbers of patients with CHD in the UK, with and without serious mental health problems, who are currently not treated with statins. CIs for these estimates were derived by bootstrapping, to take account of the clustered nature of the dataset and avoid underestimating SEs.24 All the analyses were conducted using STATA V.9.2.

Results

In all, 485 UK general practices met the inclusion criteria, with 3 262 979 patients registered as on 1 April 2005. Table 11 describes the patients by sex, age and deprivation level, and shows that 0.24% of patients had a recorded diagnosis of schizophrenia and 0.16% a diagnosis of bipolar disorder. Overall, 3.2% of females and 4.6% of males had a diagnosis of CHD. The prevalence of this condition in the schizophrenia group was, 4.6% females, 4.0% males; and in the bipolar group, 7.4% females, 8.8% males (not tabulated). Standardising to the UK population gives an estimated 2 020 070 patients with CHD, of whom 5360 also have schizophrenia and 6578 also have bipolar disorder. Table 11 also shows the socioeconomic profile of each group: there was a slight over‐representation of the least‐deprived quintile on the overall database, and also among patients with CHD. The schizophrenia group was strongly skewed towards the most‐deprived quintile, and there was a similar but less‐marked effect in the bipolar group.

Table thumbnail
Table 1 Total patients on database and those with schizophrenia, bipolar disorder or coronary heart disease

There were 127 932 patients aged [gt-or-equal, slanted]25 years with a recorded diagnosis of CHD, and these formed the sample for the study: their characteristics are shown in table 22.. A total of 332 had a recorded diagnosis of schizophrenia, 408 had a diagnosis of bipolar disorder, and 39 had both these diagnoses, giving 701 patients with either schizophrenia or bipolar disorder. Patients with schizophrenia were younger, more likely to be female and more likely to live in a deprived area than those with neither mental health condition. Patients with bipolar disorder were of similar age as those with neither condition but were more likely to be female and living in a deprived area. The prevalence of diabetes and smoking were highest among patients with schizophrenia and the prevalence of stroke was highest among patients with bipolar disorder.

Table thumbnail
Table 2 Characteristics of patients with coronary heart disease, with and without serious mental health problems

Table 33 shows the number of patients in each mental health group who were eligible for each indicator, and the number of eligible patients who achieved each outcome. Almost 75% of patients, overall, had been prescribed a statin in the previous 12 months, but this was only 65% for patients with schizophrenia. There was little difference between groups in the proportion who had taken statins in the past but were no longer doing so (around 6–7%). The largest differential between groups on eligibility for the indicators was for CHD4, smoking cessation advice, reflecting the higher proportion of smokers in the schizophrenia group. Many of the indicators had over 90% achievement among eligible patients, with little difference between mental health groups.

Table thumbnail
Table 3 Eligibility and achievement of indicators and use of statins in patients with and without serious mental health problems

Table 44 shows the RR of each outcome for eligible patients with schizophrenia or bipolar disorder, compared with patients having neither condition, first unadjusted and then adjusted for age, sex, deprivation, diabetes, stroke and smoking, and allowing for clustering by practice. Adjustment made little difference to the estimated RRs, and only two of the measures showed a significant difference between groups. Eligible patients with schizophrenia were 15% less likely than those with neither condition to have had statins prescribed in the last 12 months (95% CI 8% to 20%). This shortfall persisted if the 39 patients with both schizophrenia and bipolar disorder were excluded (14% less likely, 95% CI 7% to 20%). Patients with schizophrenia were also 7% less likely to have had their cholesterol recorded in the past 15 months (CHD7, 95% CI 3% to 11%). There was a marginally significant shortfall in exercise testing or referral (CHD2) among patients with schizophrenia (p = 0.04), but only 10% of the group were eligible for this indicator and the comparison was therefore based on small numbers.

Table thumbnail
Table 4 RRs for use of statins and achievement of indicators in eligible patients with and without serious mental health problems

As reported above, we estimated that there are just over 2 million patients with CHD in the UK, of whom 74.3% (95% CI 73.6% to 75.0%) are currently prescribed statins. This would indicate that over 500 000 patients with CHD in the UK are not receiving statin treatment. In the subgroup of approximately 5360 patients with CHD who also have schizophrenia, the lower rate of statin prescription at 65.1% (95% CI 60.4% to 69.7%) indicates that an estimated 1870 (95% CI 1620 to 2120) would be untreated.

Discussion

This study of the primary care of over 127 000 UK patients with CHD found that for the majority of nationally specified care indicators, there were no inequalities for patients who also had a serious mental health problem (schizophrenia or bipolar disorder). There were two areas of care where patients with schizophrenia showed a significant deficit: the recording of cholesterol values and the prescription of statins. There were no significant differences in the recording of smoking status or provision of advice on smoking cessation, the recording of blood pressure, the achievement of target blood pressure or cholesterol values, or the prescription of aspirin, antiplatelets, anticoagulants or β blockers.

Strengths and weaknesses

This study was based on the routine primary care records of a large and representative population‐based sample. Recording of clinical diagnoses and prescribed medications has been shown to have good levels of accuracy and completeness in UK general practices,25 with the quality of the electronic record now surpassing the conventional paper‐based system.26 The indicators of care included in this study were agreed as a national standard as part of the UK GMS contract for general practitioners, and the quality of recording would be expected to be high.

Our study was limited to patients who access primary care, which would have excluded any individuals, such as those in prisons or other institutions, who were not registered with a general practitioner. There is a marginal under‐representation of the most‐deprived socioeconomic group on the overall database. There was a similar deprivation gradient among the patients with CHD, when the opposite might have been expected: possible explanations for this include underdiagnosis of the condition in the most‐deprived groups, or their shorter survival with the condition.

The group with schizophrenia had only a marginally higher prevalence of CHD than the general population, whereas in the bipolar group, it was substantially higher; however, the schizophrenia group was much younger. We based our study on prevalent cases of CHD, which will have excluded anyone who had died before the study date, and in view of the higher mortality of people with serious mental health problems, it is possible that the survivors are untypical of the whole population.

Previous studies

Previous studies of care for CHD among people with mental health problems have focused on patients hospitalised for acute myocardial infarction.

One study found that patients with a comorbid mental disorder were substantially less likely to undergo coronary revascularisation procedures14; another found that differences in five indicators of post‐acute care (reperfusion, aspirin, β blockers, ACE inhibitors and smoking cessation advice) accounted for much of the excess 1 year mortality among patients with mental health conditions.15 However, two studies conducted on samples from the US VA system found no differences in the likelihood of revascularisation, receipt of medications or presence of a cholesterol‐management plan.16,17 The authors of these latter studies suggest that barriers to effective medical care for people with mental health problems may exist to a lesser extent in the US VA system because of its integrated nature and high volume of mental healthcare.

Implications

Although we found no inequalities in care on most of the measures, it is a cause for concern that patients with CHD and schizophrenia are less likely to receive effective monitoring and treatment for raised cholesterol. People with schizophrenia seem to be at an inherently higher risk of some physical illnesses,27 and may also have increased risks associated with poor diet, physical inactivity or smoking.12 Antipsychotic medications are known to have cardiac effects,28 and the newer atypical agents, particularly clozapine and olanzapine, are associated with significant weight gain and higher risk of diabetes.29,30 These factors emphasise the need for careful monitoring and treatment of CHD in people with schizophrenia.

Based on the present analysis we cannot confirm whether patients with schizophrenia are less likely to be offered cholesterol checks or statin treatment, or less likely to accept them. The proportion of patients eligible for cholesterol testing was somewhat lower in the group with schizophrenia (94.9% compared with patients with neither condition, 97.4%), reflecting a higher rate of exception codes recorded by their general practitioner, which may be an indication of patient's choice. However, a previous study found that people with severe mental illness were as likely as other patients to respond to an invitation for cardiac risk assessment.31 Despite the deficits in cholesterol monitoring and provision of statins, there was apparently no inequality between mental health groups in the proportion with recent cholesterol value in the target range (indicator CHD8). However, the proportion eligible for this indicator was lower among patients with schizophrenia (85.5% compared with patients with neither condition, 90.0%), raising the possibility that the differential use of exception codes removes patients with schizophrenia from consideration and masks any inequality that exists. It should also be noted that the indicator refers to total serum cholesterol, and would not necessarily capture any differences in the component measures: a recent study found that high‐density lipoprotein levels, but not total cholesterol levels, differed significantly between patients with and without severe mental illness.32

Equitable access to healthcare is about patients receiving the level of care appropriate to their level of need, rather than a “one size fits all” approach. Patients with higher levels of need and greater risk of adverse outcomes, such as patients with schizophrenia who have higher rates of smoking and obesity, are likely to need more energetic screening and management, and this may need to be done at a younger age to get the maximum health gain possible. Global incentives such as those contained in the UK GMS contract may do little to improve the situation for individuals who have failed to benefit from previous initiatives such as screening and health promotion. Further research is needed to determine why these inequalities exist and also to determine the best strategies to address them.

Conclusion

While the majority of CHD care indicators are achieved equally for patients who also have a serious mental health problem, there is a shortfall in identifying and treating raised cholesterol levels among patients with schizophrenia, despite their higher level of risk factors.

Acknowledgements

We thank EMIS practices contributing to QRESEARCH, David Stables (EMIS Computing) for his help and expertise in establishing QRESEARCH, and Gavin Langford and Mike Heaps (University of Nottingham) for carrying out the data extraction.

Abbreviations

CHD - coronary heart disease

GMS - General Medical Services

VA - Veterans Health Administration

Appendix A

READ CODES FOR MENTAL HEALTH DIAGNOSES

Schizophrenia

E10, E100, E100‐1, E1000, E1001, E1002, E1003, E1004, E1005, E100z,

E101, E1010, E1011, E1012, E1013, E1014, E1015, E101z,

E102, E1020, E1021, E1022, E1023, E1024, E1025, E102z,

E103, E1030, E1031, E1032, E1033, E1034, E1035, E103z,

E104, E104‐1,

E105, E1050, E1051, E1052, E1053, E1054, E1055, E105z,

E106, E106‐1,

E107, E107‐1, E1070, E1071, E1072, E1073, E1074, E1075, E107z,

E10y, E10y‐1, E10y0, E10y1, E10yz, E10z

Bipolar disorder

146D, E11, E11‐1, E11‐3,

E111, E1110, E1111, E1112, E1113, E1114, E1115, E1116, E111z,

E114, E114‐1, E1140, E1141, E1142, E1143, E1144, E1145, E1146, E114z,

E115, E115‐1, E1150, E1151, E1152, E1153, E1154, E1155, E1156, E115z,

E116, E1160, E1161, E1162, E1163, E1164, E1165, E1166, E116z,

E117, E1170, E1171, E1172, E1173, E1174, E1175, E1176

E11y1, E11y3, E11yz, E11z

Eu250, Eu250‐1, Eu250‐2, Eu30, Eu30‐1, Eu302‐3, Eu30y, Eu30z, Eu311, Eu312, Eu31y‐2

Footnotes

Funding: This study was supported by the Disability Rights Commission. The sponsor was involved in defining the initial research question but had no involvement in the study design, the extraction, analysis of data, the writing of the report, or the decision to submit the paper for publication. JH‐C, CC and YV are core‐funded university staff, CP received NHS R&D support funding.

Competing interests: None.

Ethics approval was granted by Trent MREC (reference 05/MRE04/47).

All authors contributed to the study design, analysis plan, and manuscript. In addition, JH‐C initiated the study, obtained ethical approval, supervised the data extraction and is the study guarantor.

References

1. World Health Organisation World health report. Geneva: WHO, 2004
2. National Institute for Clinical Excellence Statins for the prevention of coronary events: draft guidance. London: NICE, 2005
3. Scandinavian Simvastatin Survival Group Randomised trial of cholesterol lowering in 4444 patients with coronary heart disease: the Scandinavian Simvastatin Survival Study (4S). Lancet 1994. 3441383–1389.1389 [PubMed]
4. Long‐term Intervention with Pravastatin in Ischaemic Heart Disease (LIPID) Study Group Prevention of cardiovascular events and death with pravastatin in patients with coronary heart disease and a broad range of initial cholesterol levels. N Engl J Med 1998. 3391349–1357.1357 [PubMed]
5. Hippisley‐Cox J, Coupland C. The effect of statins on mortality in patients with ischaemic heart disease: population‐based cohort study with case control analysis. Heart 2005. 61523
6. The NHS Confederation The new GMS contract. London: The NHS Confederation, 2003. 108
7. Hippisley‐Cox J, Pringle M. Health inequalities experienced by people with schizophrenia and manic depression: analysis of general practice data in England and Wales. Report to the Disability Rights Commission. University of Nottingham, Nottingham, UK: Disability Rights Commission, 2005
8. Harris S C, Barraclough B. Excess mortality of mental disorder. Br J Psychiatry 1998. 17311–53.53 [PubMed]
9. Dembling B P, Chen D T, Vachon L. Life expectancy and causes of death in a population treated for serious mental illness. Psychiatr Serv 1999. 501036–1042.1042 [PubMed]
10. Tsuang M T, Woolson R F, Fleming J A. Premature deaths in schizophrenia and affective disorders. Arch Gen Psychiatry 1980. 37979–983.983 [PubMed]
11. Seymour L. Not all in the mind. The physical health of mental health service users. Radical mentalities—briefing paper 2. London: Mentality, 2003
12. Brown S, Birtwistle J, Roe L. et al The unhealthy lifestyle of people with schizophrenia. Psychol Med 1999. 29697–701.701 [PubMed]
13. Burns T, Cohen A. Item of service payments for general practitioner care of severely mentally ill patients: does the money matter? Br J Gen Pract 1998. 481415–1416.1416 [PMC free article] [PubMed]
14. Druss B G, Bradford D W, Rosenheck R A. et al Mental disorders and use of cardiovascular procedures after myocardial infarction[see comment]. JAMA 2000. 283506–511.511 [PubMed]
15. Druss B G, Bradford W D, Rosenheck R A. et al Quality of medical care and excess mortality in older patients with mental disorders. Arch Gen Psychiatry 2001. 58565–572.572 [PubMed]
16. Petersen L A, Normand S ‐ L T, Druss B G. et al Process of care and outcome after acute myocardial infarction for patients with mental illness in the VA health care system: are there disparities? Health Serv Res 2003. 38(Pt 1)41–63.63 [PMC free article] [PubMed]
17. Desai M M, Rosenheck R A, Druss B G. et al Mental disorders and quality of care among postacute myocardial infarction outpatients. J Nerv Ment Dis 2002. 19051–53.53 [PubMed]
18. Hippisley‐Cox J, Vinogradova Y, Coupland C. et alComparison of key practice characteristics between general practices in England and Wales and general practices in the QRESEARCH data. Report to the Health and Social Care Information Centre. University of Nottingham, Nottingham, UK: Health and Social Care Information Centre, 2005
19. Office for National Statistics Key health statistics in general practice. London: Office of National Statistics, 1998
20. Hippisley‐Cox J, Pringle M. Prevalence, care and outcomes for patients with diet controlled diabetes in general practice: cross sectional survey. Lancet 2004. 364423–428.428 [PubMed]
21. Hammersley V, Hippisley‐Cox J, Wilson A. et al A comparison of research General Practices and their patients with other practices ‐ cross sectional survey in Trent. Br J Gen Pract 2002. 52463–468.468 [PMC free article] [PubMed]
22. PRIMIS+ Severe mental illness statistics. University of Nottingham, Nottingham, UK, 2006 http://www.primis.nhs.uk (accessed 3 Mar 2007)
23. McNutt L ‐ A, Wu C, Xue X. et al Estimating the relative risk in cohort studies and clinical trials of common outcomes. Am J Epidemiol 2003. 157940–943.943 [PubMed]
24. Efron B, Tibshirani R. Bootstrap methods for standard errors, confidence intervals, and other measures of statistical accuracy. Stat Sci 1986. 154–77.77
25. Jick H, Jick S S, Derby L E. Validation of information recorded on general practitioner based computerised data resource in the United Kingdom. BMJ 1991. 302766–768.768 [PMC free article] [PubMed]
26. Hippisley‐Cox J, Pringle M, Cater R. et al The electronic record in primary care—regression or progression? Cross‐sectional survey. BMJ 2003. 3261439–1443.1443 [PMC free article] [PubMed]
27. Ryan M C, Thakore J H. Physical consequences of schizophrenia and its treatment: the metabolic syndrome. Life Sci 2002. 71239–257.257 [PubMed]
28. Witchel H J, Hancox J C, Nutt D J. Psychotropic drugs, cardiac arrhythmia, and sudden death. J Clin Psychopharmacol 2003. 2358–77.77 [PubMed]
29. Henderson D C, Ettinger E P. Glucose intolerance and diabetes in schizophrenia,. In: Meyer JM, Nasrallah HA, eds. Medical Illness and Schizophrenia. Washington DC: American Psychiatric Publishing, 2003. 99–114.114
30. Sernyak M J, Leslie D L, Alarcon R D. et al Association of diabetes mellitus with use of atypical neuroleptics in the treatment of schizophrenia. Am J Psychiatry 2002. 159561–566.566 [PubMed]
31. Osborn D P J, King M B, Nazareth I. Participation in screening for cardiovascular risk by people with schizophrenia or similar mental illnesses: cross sectional study in general practice. BMJ 2003. 3261122–1123.1123 [PMC free article] [PubMed]
32. Osborn D P J, Nazareth I, King M B. Risk for coronary heart disease in people with severe mental illness: cross‐sectional comparative study in primary care. Br J Psychiatry 2006. 188271–277.277 [PubMed]

Articles from Heart are provided here courtesy of BMJ Publishing Group