G McLean
General Practice and Primary Care, Community Based Sciences, University of Glasgow, 1 Horselethill Road, Glasgow
Correspondence to: G McLean, Department of General Practice and Primary Care, Community Based Sciences, University of Glasgow, 1 Horselethill Road, Glasgow G12 9LX Email: gml17y@clinmed.gla.ac.uk
SMJ 2005 50(3): 109-113
Abstract
Background: The primary aim of the Platform Project is to maximise the use of routine data for primary care research in Scotland. Aims: To assess the extent to which routine data is available and has been used in studies on CHD in General Practice research in Scotland. To assess the advantages and limitations of using routine data in this setting Methods: Literature review using a variety of databases catalogues and websites, bibliographies of articles retrieved and searching through journals by hand not available electronically. Results: This review has found that the use of routine data in CHD studies in General Practice research in Scotland remains small. There has been little work undertaken which has combined the use of routine data with other research methods. Limitations with routine data exist particularly with regard to risk factors and ethnicity. However, despite such limitations there exists an increasingly extensive range of data, which exists to help explain tends in CHD, which so far has been largely underused.
Key words: Routine data, CHD, general practice
Introduction
The primary aim of the Platform Project is to maximise the use of routine data for primary care research in Scotland. 1 The specific focus of Phase II of the Platform Project is to look at a ‘systems of care’ approach with emphasis on CHD. It has been argued that the quality of UK data on CHD is remarkably patchy and poor despite the fact that CHD has been and remains the largest cause of death.2 The relevance of examining issues around heart disease can be seen by the fact that it is a national priority, is a substantial cause of avoidable inequalities in health and a condition for which there is a considerable potential contribution from primary care.2
Although routinely collected data provide large quantities of health information such sources have limitations and said to be underused.3 As part of this process it is important to assess the extent to which routine data has been used in studies on CHD. Routine data had been described as data collected irrespective of the procedure or outcome. Most data for research purposes is collected with a specific purpose in mind and requires a data collection proforma. The advantage of routine data is that it is low cost, large size of database, large population coverage; collection usually spans a significant time period, breadth and diversity to explore unexpected avenues.4 The aim of this paper is to assess the extent to which routine data is available and has been used in studies on CHD in general practice research in Scotland and to assess the advantages and limitations of using routine data in this setting.
Methods and criteria
A research paper was selected for inclusion here if it conformed with the following criteria: it examines issues concerning general practices and/or uses general practice population data; is based in Scotland; and dates from 1990 on and uses routine data sources. For the purposes of this study routine data is defined as data that are routinely collected without any specific question in mind.3 This definition encompasses a wide range of potential data available at national and local level such as: health outcome data, exposures and health determinant data, disease prevention data, demographic data and geographical data.
In the undertaking of this literature review a variety of databases catalogues and websites were searched. Electronic databases used included the following: EMBASE, MEDLINE and the Social Science Index (ISI). A number of journal websites were also searched including the BMJ website’s multiple journal search facility. Additional studies were collected from the bibliographies of articles retrieved and searching though journals by hand not available electronically. In exploring the literature on CHD in Scotland in a general practice setting, this review has only examined those studies conducted after 1990. This is due to the fact that data sources and information on risk factors, blood pressure and cholesterol levels can be said to have been very limited during the 1980’s and far more extensive thereafter.3
Routine data sources in Scotland
Compared to the rest of the UK, Scotland is seen as having an advantage with regard to the collection of CHD data. The Information and Statistics Division (ISD) collects good data on all patients treated for CHD and the procedures received. Scotland’s routine NHS data are of high quality and data linkage allows for the investigation of the epidemiology and treatment of heart disease across the population and at general practice level, with comprehensive analysis then being possible on different forms of the disease.2,3 The main areas covered by routine data, which have been available, involve the following: mortality, prescribing, inpatients and procedures, outpatients, continuous morbidity recording (CMR), and the Scottish Health Survey.
The HEARTS5 project aims to develop a myocardial infraction register in each interested practice, and to collect comprehensive data on these patients to feedback for clinical use. It is an ongoing study using January 1977 as a start date. Information is gathered on patients who have had an acute myocardial infarction or undergone coronary angioplasty or artery bypass grafting in the Tayside region. Patients are traced through a range of consultations with their GPs. The information will be fed back to practices in various ways, in an effort to support them in improving care. It is believed that such a process will better able GPs to determine whether patients’ risk factors and medication are optimised.
The HYPER trial6 is a pragmatic randomised controlled trial designed to evaluate different levels of feedback developed from data extracted from the computerised record systems of general practices in Scotland. Fifty-two practices representing urban and rural Scotland with a range of practice size and deprivation were recruited and randomised to three groups: control, feedback of audit data and strategic feedback prioritising patients by absolute risk of death from stroke. Electronic data on demography, morbidity and prescribing were extracted from the practice computer systems and the relevant feedback developed.
The Scottish Health Survey appeared first in 1995, and aims to provide a comprehensive picture of the whole of the population, its biological characteristics, or healthrelated behaviour. The survey is modelled closely on the Health Survey for England with the most recent survey published in 1998 and available from the UK Data Archive based at the University of Essex.7 However, unlike the English survey, which is published, annually there has been a five-year gap in Scotland. The 2003 survey involves interviews with around 10,000 adults and 4,000 children. The two main aims of the survey are to monitor change over time in people’s health status and health-related behaviour and to explore differences between sub-groups in the population. The main focus of the 2003 study is cardiovascular disease.
The CMR project was co-ordinated by ISD and involved prospective data collection from face to face contacts between doctors and patients from selected general practices. This involved 53 practices with a total registered practice of 307,741 patients (around 6% of the total population) and said to be representative of the Scottish population for age, sex, socio-economic status, and mix of rural and urban locations, although it has been said to have a relative deficit of severely deprived individuals.8
Comprehensive information was collected about the index, including whether it is a first occurrence, recurrent, or persistent), up to nine concomitant medical problems, and prescriptions issued and renewed. The CMR had developed into the Practice Team Information (PTI) available from ISD. The PTI is a system for the collection of primary care data from the general practice team, including general practitioners, practice and community nurses. At the time of writing there are 63 practices taking part in PTI.
Results
Pears et al9 attempted to determine whether deprivation, in addition to gender and age differences, exist in the primary care management of hypertension. The methods used in the study involved a cross sectional analysis of computerized general practice data and was carried out in 43 practices in Scotland contributing to CMR. McAlister et al10 used the CMR to examine whether there are socio-economic gradients in the incidence, prevalence, treatment and follow up of patients with heart failure in primary care. This was a population-based study set in 53 practices (307,741 patients) participating in the CMR project between 1 April 1999 and 31 March 2000.
Murchie et al11 undertook a study whose objectives were to evaluate the efforts of nurse led clinics in primary care on secondary prevention, total mortality, and coronary event rates after four years. This work emerged from a systematic review of randomised trials, which concluded that programmes for disease management improved processes of care, reduced admissions to hospital and enhanced quality of life.12 The authors conducted one of the randomised trials included in this review and their findings were typical of the pooled results. These findings included nurse led secondary prevention clinics in primary care improved medical and lifestyle components of secondary prevention (except smoking) and health related quality of life.13, 14 ,15
The setting involved a stratified random sample of 19 general practices in northeast Scotland. Participants were 1343 patients (673 intervention and 670 control) under 80 years with a working diagnosis of coronary heart disease but without terminal illness or dementia and not housebound. Data on the date and cause of death were obtained from ISD. Coronary events were defined as coronary deaths or non-fatal myocardial infarctions. Data was collected on nonmyocardial infractions during a review of general practice case notes and from hospital records held by ISD. Mitchell et al16 used data from the HYPER trial to evaluate the impact of different levels of feedback on identification, treatment, and control of older people with hypertension. Practices were randomly allocated to either control (n=19), audit only feedback (n=16), or audit plus risk feedback, prioritising patients by absolute risk (n=17). Electronic data were extracted from practice computer systems annually from 1999 to 2001 and used to develop feedback. Data were collected for 30,345 patients aged 65-79 years from 52 practices.
Routine data has also been used to examine the costs involved with regard to CHD in Scotland.17,18 Two studies undertaken by Stewart et al18,19 attempted to predict the future burden of heart failure in Scotland from contemporary epidemiological data. Scotland has an ageing although numerically stable population. Current estimates of prevalence, GP consultation rates, and hospital admission rates related to heart failure were applied to the whole Scottish population. Age and specific rates of GP consultations per 1000 population were obtained from ISD. Further work has been concerned with attempts to calculate the current cost of angina pectoris to the NHS in the UK in 2000.19 The methods used involved calculation of the cost of hospital admissions, revascularisation procedures, hospital outpatient procedures, general practice consultations, and prescribed drug treatment.
Donnan, Dougall and Sullivan20 used evidence from the Heart Disease Evidence-based Audit and Research in Tayside, Scotland (HEARTS) to determine the optimal strategy to identify subjects with a myocardial infarction from general practice criteria. This was a cross-sectional study of 10 general practices in Tayside Scotland. A random sample of all subjects aged over 35(n=5061) and registered with the general practices was obtained. The main outcome measures were sensitivity, specificity, positive predictive value (PPV) and yield (the number of records that need to be examined to detect a true case). HEARTS has also be used to investigate whether practices will be ready for the data reporting requirements for the new General Medical Services (GMS) contract.21 The electronic GP records of all CHD patients in five Scottish practices were searched and data was validated by manual searches in 50 randomly selected patients in each practice. The main outcome measures used were recording of family history, smoking status, blood pressure, diabetes testing, and aspirin therapy and cholesterol measurement.
Combining routine data with other sources
Some of studies included in this review have combined routine with other methods as part of their study design. Murchie11 used the follow up of a randomised controlled trial by postal questionnaires and review of case notes and national datasets. Stewart et al17 used two comparable cross sectional studies undertaken by McDonagh and colleagues22 (Scottish data involving 1640 individuals aged 45-75 years) and Morgan et al23 to derive age and sex specific estimates of the population prevalence of individuals with symptomatic left ventricular systolic dysfunction (one form of chronic heart failure.
Findings from use of routine data
Results obtained from the use of routine data include those by Pears et al9 who found that that compared with males, female hypertensive patients were more likely to receive a thiazide and less likely to be given an ACE inhibitor, calcium channel blocker or secondary preventive treatment. Elderly hypertensive patients were less likely than the youngest patients to be under active GP review, more likely to be on a thiazide, calcium channel blocker or anti-platelet treatment, and less likely to be on statin. More deprived hypertensive patients were less likely under GP review, or to be on a thiazide or a statin, but were more likely to be on a calcium channel broker or an antiplatelet drug than the most affluent group. The age and sex standardised incidence of heart failure increased with greater socio-economic deprivation, from 1.8 per 1000 population in the most affluent stratum to 2.6 per 1000 in the most deprived stratum (odds ratio 1.44, p=0.0003). On average, patients were seen 2.4 times yearly, but follow up rates were less frequent with increasing socio-economic deprivation (from 2.6 years yearly in the most affluent sub-group to 2.0 yearly in the most deprived sub-group, p=0.0009). Prescribing patterns did not vary on deprivation categories on univariate or multivariate analysis.
The work of Murchie et al11 indicated that significant improvements were shown in the intervention groups in all components of secondary prevention except for smoking at one year, and these were sustained after four years except for exercise. The control group, most of whom attended clinics after the initial year, caught up before final follow up and differences between groups were no longer significant. The authors argued that the results showed that nurse-led secondary prevention improved medical and lifestyle components of secondary prevention and this seemed to lead to significantly fewer total deaths and ‘probably’ fewer coronary events.
Stewart et al17 found that by 2020 the annual number of male and female hospital admissions associated with a principal diagnosis of heart failure is expected to increase by 34% and 12 % respectively. It was argued by the authors that unless rapid and major changes occur in the incidence of heart failure, the burden of this disorder will continue to increase in both primary and secondary care over the next two decades.
Evidence from the study conducted by Mitchell et al16 showed that there were improvements in the ascertainment and treatment aspects of the rule of halves. Most 65-79- year olds have blood pressure recorded on computer, as did almost all diagnosed hypertensive patients. Although there were still significant numbers of treated hypertensives with high blood pressure, the authors argued that sending patients-specific feedback can impact on detection and management of hypertension in the elderly and produce a significant difference in control.
Results from work that has used the HEARTS study20 concluded that patients who experienced an MI can be easily identified from the combination of a Read code for MI and a record of hospitalisation for an MI giving 100% sensitivity and specificity with a yield of 1.11. In situations where the Read coding is of poor quality, the alternative search strategy of a hospital record of MI or receiving aspirin or nitrates was optimum.
Further work using HEARTS21 found that in Scotland there is a high level of recording and testing for all the important information regarding patients with recorded CHD, irrespective of whether practice have fully electronic records, paper-based records, or a mixture of the two. If practices have fully electronic patient records, the information can be extracted easily, but unless there is a standard template, the information can only be viewed in isolation and is of little use for comparative purposes.
Limitations with routine data
The use of routine data has many advantages but there are also limitations that need to be addressed. The majority of studies reviewed in this paper recognise that results found in their studies often need to be addressed with caution. The first lesson in the use of routine data must be to be aware of the limitation of the work being undertaken. Missing data on potential confounding factors has been recognised as a problem.
In the study undertaken by Pears9 this has included data on smoking habits, alcohol consumption, dietary and exercise habits, and body mass index. Further unavailability of data included the level of blood pressure at the time of diagnosis; level of blood pressure at different time points whilst on treatment, total history of anti-hypertensive treatments; compliance with therapy or side effects experienced; all of which are factors that will influence choice of treatment. This meant that residual confounding could account for at least part of the findings.
McAllister et al10 note that the limitations to their data include the fact that they do not have any independent confirmation of heart failure diagnoses nor data on disease severity. They also did not have access to data on potential confounders related to cardiovascular risk factors, and these may vary substantially across socio-economic gradients. The problem with data on potential confounders may relate to the irregular publishing of the Scottish Health Survey in comparison to its English counterpart. Indeed the fact that the most recent survey in circulation dates back to 1998 may account for absence of the use of the survey in any of the studies examined in this review. The study undertaken by Murchie11 faced limitations with regard to due to the limited scope of routine data sources to explain differences in outcomes based on processes of care and risk factors.
Work undertaken by Stewart has also faced a number of difficulties.17,19 Using routine data to compare findings across different parts of the UK faces problems with the extrapolation of Scottish rates of GP consultation, hospital admission, and revacascularation to the UK as a whole. The prevalence of heart disease in Scotland is greater than in some parts of England, hence it is possible that data used in this way may marginally inflate the overall UK cost of angina. Conversely it is not possible to calculate the cost of accident and emergency department visits not leading to hospital admission, as reliable data are not available. It has previously been argued that data for women and the elderly population are scarce, however on the context of CHD in Scotland this would not appear to be the case. However, there remains limited data on ethnic minorities both with regard to patients and GPs. Moreover, research has shown that routine data sources with partial coverage are more likely to be located in affluent areas.8
Conclusion
The purpose of this paper has been to identify and review work using routine data that has focused on CHD in general practice research in Scotland on behalf of the Platform Project. Perhaps the most important finding of this review, has been how surprisingly few studies have used routine data. This maybe considered something of a surprise given the fact that, despite a number of limitations recognised with the use of routine data, Scotland is seen as having an advantage compared to the rest of the UK with regard to routine data on CHD.3 There has been little research undertaken which has combined the use of routine data with other methods. Work using routine data has been undertaken on myocardial infarction, heart failure, and hypertension.
There has been little work that has focused on risk factors such as lifestyle and diet. There remain a number of limitations in the use of routine data. The failure to produce an annual Health survey for Scotland in line with its English counterpart is perhaps the biggest disadvantage. This limits the scope of routine data sources to help explain the impact of potential confounding factors on CHD prevalence. There is limited data available on practice and community nurses and their role in treatment of CHD, although the development of the PTI may begin to help to fill this gap. There is also little data available concerning ethnicity for both patients and clinicians. However, there remains a wide range of data available with both partial and full coverage of Scotland, which exists to help explain tends in CHD, which have largely been underused thus far.
REFERENCES
1 The Platform Project. Glasgow University (http://www.gla.ac.uk/projects/ platform/)
2 Unal B, Critchley JA and Capewell S. Missing, Mediocre or merely obsolete. An evaluation of UK data sources for coronary heart disease. Journal of Epidemiology and Community Health 2003; 57: 530-535
3 Bain M, Charmers J and Brewster D. Routinely collected data in national and regional databases-an under used resource. Journal of Public Health Medicine 1997; 19: 413-418
4 Powell A, Davies H, Thomson R. Using routine comparative data to assess the quality of health care: understanding and avoiding common pitfalls. Quality and Safety in Health Care 2003; 12:122-128
5 http:/www.hearts.org.uk
6 Sullivan F, Watt G, Mitchell E, Grimshaw J and Walker A. Turning general practice data into useful knowledge-a randomised trail of two strategies of feedback to improve management of elderly hypertensives using computerised data. (see http://www.sign.ac.uk/guidelines/published/index.html for further details)
7 http://www.data-archive.ac.uk
8 Milne RM, Taylor MW and Taylor RJ. Audit of populations in general practice: the creation of a national resource for the study of morbidity in Scottish general practice. Journal of Epidemiology Community Health 1998; 52 (suppl 1): 20S-24S.
9 Pears E, Hannaford PC and Taylor MW. Gender, age and deprivation differences in the primary care management of hypertension in Scotland: a cross sectional database study. Family Practice 2003; 20(1): 22-31
10 McAlister F, Murphy N, Simpson C, Stewart S, Macintyre K, Kirkpatrick M, Chalmers J et al. Influence on Socio-economic deprivation on the primary care burden and treatment of patients with a diagnosis of heart failure in general practice: population based study. Br Med J 2004;328:1110
11 Murchie P, Campbell NC, Ritchie LD, Simpson JA, and Thain J. Secondary Prevention clinics for heart disease: four year follow up of a randomised controlled trial in Primary Care. Br Med J 2003; 326: 84
12 McAlister FA, Lawson FME, Teo KK, and Armstrong PW. Randomised trials of secondary prevention programmes in coronary heart disease: systematic review. Br Med J 2001; 323: 957-962
13 Campbell NC, Thain J, Deans HG, Ritchie LD, Rawles JM, and Squair JL. Secondary prevention clinics for coronary heart disease: a randomised trial of effect on health. BMJ 1998; 316: 1434-1437
14 Campbell NC, Thain J, Deans HG, Ritchie LD, Rawles JM and Squair JL. Secondary prevention in coronary heart disease: a randomised of nurse led clinics in primary care. Heart 1998; 80: 447-452
15 Campbell NC, Thain J, Deans HG, Ritchie LD, Rawles JM and Aquair JL. Secondary prevention in coronary heart disease: baseline survey of provision in general practice. Br Med J 1998; 316: 1430-1434
16 Mitchell E, Sullivan F, Grimshaw J, Donnan P, and Watt G. Improving management of hypertension in general practice: a randomised controlled trial of feedback derived from electronic patient data. British Journal of General Practice 2005; 55:94-101
17 Stewart S, Macintyre K, Capewell S, and McMurray JJ. Heart Failure and the aging population: an increasing burden in the 21st Century. Heart 2003 Jan; 89(1): 49-53
18 Stewart S, Jenkins A, Buchan S, McGuire A. Capewell S and McMurray JJ. The current cost of Heart Failure to the National Health Service in the UK. European Journal of Heart Failure 2002 June; 4(3): 361-371
19 Stewart S, Murphy N, Walker A, McGuire A, and McMurray JJV. The current cost of angina pectoris to the National Health Service in the UK. Heart; 2003; 848-853
20 Doonan P, Dougall H, Sullivan F. Optimal strategies for identifying patients with myocardial infarction in general practice. Family Practice 2003; 20(6): 706-710
21 Morris L, Taylor M, Campbell M and Sullivan F. How will practices cope with information for the new GMS contract? Coronary Heart Disease recording in five Scottish practices. Informatics in Primary Care 2003; 11(3): 121-7
22 McDonagh T, Morrison C, Lawrence A, Ford I, Tunstall-Pedoe H, McMurray J and Dargie J. Symptomatic and asymptomatic left-ventricular systolic dysfunction in an urban population. Lancet; 1997; Sept: 350: 829-833
23 Morgan S, Smith H, Simpson I, Liddiard G, Raphael H, Pickering R and Mant D. Prevalence and clinical characteristics of left ventrical dysfunction among elderly patients in general practice setting: cross sectional survey. Br Med J; 1999; 318: 368-372