Abstract
Objectives
To describe incidence and prevalence of cardiovascular disease (CVD), its risk factors, medication prescribed to treat CVD and predictors of CVD within a nationally representative dataset.
Design
Cross-sectional study of adults with and without CVD.
Setting
The Royal College of General Practitioners (RCGP) Research and Surveillance Centre (RSC) is an English primary care sentinel network. RCGP RSC is over 50 years old and one of the oldest in Europe. Practices receive feedback about data quality. This database is primarily used to conduct surveillance and research into influenza, infections and vaccine effectiveness but is also a rich resource for the study of non-communicable disease (NCD). The RCGP RSC network comprised 164 practices at the time of study.
Results
Data were extracted from the records of 1 275 174 adults. Approximately a fifth (21.3%; 95% CI 21.2% to 21.4%) had CVD (myocardial infarction (MI), angina, atrial fibrillation (AF), peripheral arterial disease, stroke/transient ischaemic attack (TIA), congestive cardiac failure) or hypertension. Smoking, unsafe alcohol consumption and obesity were more common among people with CVD. Angiotensin system modulating drugs, 3-hydroxy-3-methylglutaryl-coenzyme (HMG-CoA) reductase inhibitors (statins) and calcium channel blockers were the most commonly prescribed CVD medications. Age-adjusted and gender-adjusted annual incidence for AF was 28.2/10 000 (95% CI 27.8 to 28.7); stroke/TIA 17.1/10 000 (95% CI 16.8 to 17.5) and MI 9.8/10 000 (95% CI 9.5 to 10.0). Logistic regression analyses confirmed established CVD risk factors were associated with CVD in the RCGP RSC network dataset.
Conclusions
The RCGP RSC database provides comprehensive information on risk factors, medical diagnosis, physiological measurements and prescription history that could be used in CVD research or pharmacoepidemiology. With the exception of MI, the prevalence of CVDs was higher than in other national data, possibly reflecting data quality. RCGP RSC is an underused resource for research into NCDs and their management and welcomes collaborative opportunities.
Keywords: cardiac epidemiology, primary care, epidemiology, health informatics
Strengths and limitations of this study.
High data quality for demographic, clinical and social variables.
Ability to report prevalence and incidence.
From our rich dataset, we are able to report diagnosis and pharmacological and physiological (eg, blood pressure) data.
Data collected in primary care represent only a brief abstraction from a consultation.
Prevalence of myocardial infarction was lower than reported in other UK primary care datasets linked to hospital data. This might indicate that acute conditions are less well coded in primary care records.
Introduction
Cardiovascular disease (CVD) is a major cause of mortality and morbidity. Improved preventive strategies could reduce the burden of disease. There were over 17 million CVD-related deaths in 2012; CVD accounts for almost one-third of all deaths.1 Ischaemic heart disease has topped the list of causes of years of life lost for more than a decade,2 highlighting the shift in the global burden of disease from communicable to chronic disease.3 Risk factors for CVD, including raised blood pressure, hypercholesterolaemia and high body mass index (BMI), are among the most important contributors to disability-adjusted life years.4 Primary prevention of CVD is achievable through early identification and modification of ‘lifestyle risk factors’ and secondary prevention, through appropriate risk reduction, which slows disease progression. Effective prevention and treatment is reliant on the identification of people at risk of or with current CVD, and systems which facilitate monitoring of management.
Large datasets provide real-world insights into the epidemiology of CVD and better use of analytics may lead to improvements in care.5 Notwithstanding, the widespread use of computerised medical record (CMR) published research on the use of large datasets in CVD research remains limited.6 The majority of CVD management takes place in primary care, providing opportunities for research.7 8 In the UK, a pay-for-performance scheme (P4P), the Quality and Outcomes Framework (QOF) was introduced to incentivise general practitioner (GP) practices to achieve indicator thresholds for the management of chronic diseases,9 which has enhanced data quality in primary care. The ubiquitous use of CMR systems in UK primary care allows ready analysis enhanced by the UK’s registration-based systems, which provide an accurate denominator.10
The English Royal College of General Practitioners (RCGP) Research and Surveillance Centre (RSC) is one of the longest established primary care sentinel networks.11 It comprises a nationally representative sample,12 monitoring infections and respiratory disease, particularly influenza and assessing vaccine effectiveness.13 14 15 The RCGP RSC extended its remit into diabetes, with a focus on disparities and adherence/persistence, and has been shown to have good data quality.16 One recent analysis has used the cardiovascular outcome data from the RCGP RSC17 but there has not previously been a systematic assessment of CVD data quality and completeness. We describe the incidence and prevalence of CVD in the RCGP RSC cohort; including across different patient characteristics, medications and risk factors for CVD.
Objectives
To identify the proportion of adults currently registered within the RCGP RSC who have a diagnosis of CVD.
To describe the prevalence of risk factors for CVD and medications prescribed within the RSCP RSC network.
To compare the incidence and prevalence with other nationally reported datasets.
Methods
Study design and study population
This is a cross-sectional study using routinely collected data extracted from general practices registered within the RCGP RSC. All adults registered at the end of the study period, 31 December 2016, were included. Data were contributed by 164 practices across England.
Data source
The Read classification is currently used to code key clinical data.18 Diagnoses and key symptoms, examination findings, therapies, investigations and test results, and processes of care are increasingly coded. Interactions with secondary care are generally coded into a patient’s primary care record.
Data quality is very important to RCGP RSC.19 20 A sentinel system has to differentiate first, from new to follow-up of cases. So there is a lot of emphasis on making sure all consultations have a ‘problem’ coded, and it is so-called ‘episode type’ (ie, first, new or ongoing). Practices receive extensive feedback, practice visits, a monthly newsletter and more recently a dashboard refreshed weekly.21 We are also looking at ways of automated classification of unclassified episodes.22
The individuals in the cohort have their identity pseudonymised. We pseudonymise National Health Service (NHS) number to protect patients’ privacy—but can apply the same pseudonymisation algorithm to other datasets (eg, hospital data or cancer registries), so we can privately link patient-level data. We can, in addition, use a combination of probabilistic and deterministic methods to link data where there is no NHS number.23
Data collection takes place twice per week with near real-time processing
We collect data from all patients twice per week, making us one of the freshest data sources. We have the capability to characterise and monitor disease and use of therapies, and compare use according to guidance.24 The RCGP RSC has produced a ‘weekly return’ of infections and respiratory disease since 1967, though over time it has expanded in terms of size, scope, sample collection and its capability with linking with other datasets. We can process large numbers of patients’ data rapidly. Our weekly report involves the processing of around 1.75 million patients data to produce our weekly report of over 30 conditions in around 4 hours.
Identifying CVD
Demographic, social and clinical characteristics of the study population were identified using Read codes. Ethnicity was characterised using an established ontological approach: a combination of recorded ethnicity and information which infers ethnicity (such as language).25 Socioeconomic status (SES) was determined using the Index of Multiple Deprivation (IMD—the official UK SES measure) from each patient’s postal code,26 which we also have the capability to map to geocoordinate.
We used the latest code recorded for each patient to report smoking status. The categories we created were: never, current and ex-smoker. For alcohol use, we included alcohol-related disease or complications. We stratified alcohol use into: not recorded, within limits (<14 units per week) or alcohol consumption with no amount specified or excess (≥14 units per week or other codes consistent with heavy drinking). We categorised BMI using WHO categories: underweight (<18.5 kg/m2), normal (18.5–24.9), overweight (25.0–29.9), obese class 1 (30.0–34.9), obese class 2 (35.0–39.9) and obese class 3 (≥40.0).27 For BMI and blood pressure, we used the latest value recorded. We also report missing data.
CVD cases include a diagnosis of one or more of the following conditions: coronary artery disease (including myocardial infarction and angina), atrial fibrillation (AF), peripheral arterial disease, stroke, transient ischaemic attack and congestive cardiac failure (online supplementary appendix 1, Read code list28). We included hypertension in our analysis of CVD.
bmjopen-2017-020282supp001.pdf (452.4KB, pdf)
Risk factors for CVD and disease prevalence
In addition to those described above, we measured prevalence of established CVD risk factors demonstrated to predict CVD.29 Our analysis included latest systolic blood pressure (SBP) in excess of 140 mm Hg, chronic kidney disease (CKD, including stages 3–5) and diabetes mellitus.
Prevalence and incidence of CVD
Prevalence of CVD was described by the number of different conditions in people by age groups (<50 years, 50–59 years, 60–69 years, 70–79 years and ≥80 years). Incidence of CVD was derived for the last 5 years (2012–2016). We reported age-adjusted and gender-adjusted annual incidence of CVD using the 2011 Census for England and Wales. We also compared prevalence of CVD conditions and risk factors/comorbidities in the RCGP RSC dataset with those reported for the QOF/P4P national financially incentivised chronic disease management scheme.
CVD prescription use
We report the following cardiovascular medication prescriptions: ACE inhibitors, alpha blockers, angiotensin II receptor blockers, antiplatelet therapy (including aspirin), glycosides, calcium channel blockers, centrally acting antihypertensives, statins, fibrates, other lipid-lowering drugs, loop diuretics, nicorandil, nitrates, potassium sparing diuretics, super ventricular antiarrhythmic, thiazide and thiazide-like diuretics, vasodilator antihypertensives, and beta-blockers. We defined medication use as the recording of one or more prescriptions in the GP record.
Predictors of CVD
Logistic regression was used to identify variables that predict CVD. Variable selection was based on established risk factors for CVD. Since CVD is a composite of conditions, which present at different stages, we also explored associations between variables and the most common CVD conditions identified in this study.
Inclusion criteria
All adults, aged ≥18 years, who were registered at an RCGP RSC practice on 31 December 2016 were included in the analysis.
Exclusion criteria
Any patients who have codes suggesting they declined any form of data sharing are not analysed by RCGP RSC (approximately 2.2% of the registered population).
Statistical analysis
We used descriptive statistics calculated using the open source software, R-studio. We reported counts and percentages of crude data. Proportions were compared using χ2 tests, while the independent samples t-test and Mann-Whitney U test were used to compare measures of central tendency (means and medians).
As mentioned above, we used logistic regression to identify variables that predict CVD as an outcome. Three models were run, which included the following outcome variables: (1) CVD as a composite of conditions (coronary artery disease, myocardial infarction, angina, AF, peripheral arterial disease, stroke, transient ischaemic attack and congestive cardiac failure); (2) coronary artery disease (including myocardial infarction and angina) and (3) AF.
Within each model, we adjusted for the following predictor variables: age, gender, ethnicity, SES (using IMD stratified into quintiles), smoking status, alcohol consumption, BMI category, presence of diabetes (no diabetes, type 1 diabetes, type 2 diabetes) and presence of other known risk factors (uncontrolled SBP: >140 mm Hg; hypertension and CKD (stages 3–5). All variables were retained in the regression analyses irrespective of associations in bivariate analyses.
We reported ORs with 95% CIs and p values for each model parameter. Results were deemed significant if they were associated with a significance level of p<0.05.
Ethics
The data used for the analysis were pseudonymised at the point of extraction and encrypted prior to uploading to the Clinical Informatics Research Group secure server. Personal data were not identifiable during the analysis. This study was approved by the RCGP RSC study approval committee and was classified as a study of ‘usual practice’.30 RCGP RSC data are available to researchers and applications should be made direct to RCGP.31
Patient and public involvement
No patients or public were involved in the design, recruitment or conduct of this study.
Results
Demographics about people with and without CVD
A total of 1 275 174 adults were registered with RCGP RSC practices. SES was identified within almost the entire adult population and three-quarters had their ethnicity recorded. The least deprived quintiles (IMD quintiles 4 and 5) were over-represented in the cohort (table 1).
Table 1.
The characteristics of the adult population with and without CVD
| Characteristic | All adults, n (%) (n=1 275 174) |
Any CVD* and hypertension, n (%) (n=271 684) |
No CVD, n (%) (n=1 003 490) |
P values |
| Age mean (SD) | 47.5 (18.9) | 67.9 (13.7) | 41.9 (16.1) | <0.001 |
| Age median (IQR) | 46.0 (31.0–62.0) | 69.0 (59.0–78.0) | 40.0 (29.0–53.0) | <0.001 |
| Female | 651 298 (51.1) | 133 765 (49.2) | 517 533 (51.6) | <0.001 |
| Ethnicity recorded | 975 924 (76.5) | 219 732 (80.9) | 756 192 (75.4) | |
| White | 828 900 (65.0) | 198 884 (73.2) | 630 016 (62.8) | <0.001 |
| Asian | 78 603 (6.2) | 9853 (3.6) | 68 750 (6.9) | <0.001 |
| Black | 39 805 (3.1) | 8165 (3.0) | 31 640 (3.2) | <0.001 |
| Mixed | 14 955 (1.2) | 1577 (0.6) | 13 378 (1.3) | <0.001 |
| Other | 13 661 (1.1) | 1253 (0.5) | 12 408 (1.2) | <0.001 |
| Not recorded | 299 250 (23.5) | 51 952 (19.1) | 247 298 (24.6) | <0.001 |
| IMD recorded | 1 271 764 (99.7) | 271 120 (99.8) | 1 000 644 (99.7) | |
| IMD quintile 5 (least) | 337 239 (26.4) | 75 278 (27.7) | 261 961 (26.1) | <0.001 |
| IMD quintile 4 | 277 357 (21.8) | 62 736 (23.1) | 214 621 (21.4) | <0.001 |
| IMD quintile 3 | 232 524 (18.2) | 50 534 (18.6) | 181 990 (18.1) | <0.001 |
| IMD quintile 2 | 216 466 (17.0) | 41 311 (15.2) | 175 155 (17.5) | <0.001 |
| IMD quintile 1 (most deprived) | 208 178 (16.3) | 41 261 (15.2) | 166 917 (16.6) | <0.001 |
| IMD quintile not available | 3410 (0.3) | 564 (0.2) | 2846 (0.3) | <0.001 |
| Smoking status recorded | 1 198 579 (94.0) | 263 970 (97.2) | 934 609 (93.1) | |
| Never | 514 481 (40.3) | 78 045 (28.7) | 436 436 (43.5) | <0.001 |
| Current | 242 599 (19.0) | 38 455 (14.2) | 204 144 (20.3) | <0.001 |
| Ex-smoker | 441 499 (34.6) | 147 470 (54.3) | 294 029 (29.3) | <0.001 |
| Not recorded | 76 595 (6.0) | 7714 (2.8) | 68 881 (6.9) | <0.001 |
| Alcohol use recorded | 1 025 722 (80.4) | 258 187 (95.0) | 767 535 (76.5) | |
| None | 205 708 (16.1) | 50 670 (18.7) | 155 038 (15.4) | <0.001 |
| Within limits | 466 063 (36.5) | 93 408 (34.4) | 372 655 (37.1) | <0.001 |
| Excess | 353 951 (27.8) | 114 109 (42.0) | 239 842 (23.9) | <0.001 |
| Not recorded | 249 452 (19.6) | 13 497 (5.0) | 235 955 (23.5) | <0.001 |
| BMI recorded | 1 100 374 (86.3) | 262 715 (96.7) | 837 659 (83.5) | |
| BMI mean (SD) | 26.5 (5.9) | 41.9 (6.0) | 25.8 (5.6) | <0.001 |
| Underweight | 42 258 (3.3) | 3291 (1.2) | 38 967 (3.9) | <0.001 |
| Normal | 450 616 (35.3) | 64 395 (23.7) | 386 221 (38.5) | <0.001 |
| Overweight | 360 340 (28.3) | 99 941 (36.8) | 260 399 (25.9) | <0.001 |
| Obese class 1 | 159 638 (12.5) | 59 506 (21.9) | 100 132 (10.0) | <0.001 |
| Obese class 2 | 57 097 (4.5) | 23 328 (8.6) | 33 769 (3.4) | <0.001 |
| Obese class 3 | 31 018 (2.4) | 12 525 (4.6) | 18 493 (1.8) | <0.001 |
| Not recorded | 174 207 (13.7) | 8698 (3.2) | 165 509 (16.5) | <0.001 |
*CVD comprised any of coronary artery disease (including myocardial infarction and angina), atrial fibrillation, peripheral arterial disease, stroke and TIA and congestive cardiac failure.
BMI, body mass index; CVD, cardiovascular disease; IMD, Index of Multiple Deprivation; TIA, transient ischaemic attack.
The prevalence of CVD and hypertension was 21.3% (271 684) and more common among males. Smoking, unsafe drinking and obesity were more prevalent among people with CVD than without, as was white ethnicity. There was a lower proportion of Asian, mixed or other ethnicity individuals among those diagnosed with CVD by comparison to those without CVD, and the proportion of people in the two least deprived quintiles was higher in those with CVD (table 1).
Risk factors for CVD and disease prevalence
The most prevalent risk factor was smoking, followed by obesity, uncontrolled SBP, type 2 diabetes and CKD (table 2). Crude prevalence of smoking (active and ex-smokers) was higher in those of white and mixed ethnicity, and obesity was most prevalent in people of white and of black ethnicity. Crude prevalence of type 2 diabetes was greater in Asian and black ethnicity, and CKD in white people. The highest overall crude prevalence for CVD was in white ethnicity, particularly for AF.
Table 2.
Cardiovascular disease profile and risk factors of the complete adult population and by ethnicity
| Conditions | All adults, n (%) (n=1 275 174) |
Ethnicity, n (%) | |||||
| White n=8 28 900 | Asian n=78 603 | Black n=39 805 | Mixed n=14 955 | Other n=13 661 | Not known (n=299 250) | ||
| Cardiovascular risk factors | |||||||
| Uncontrolled (SBP >140 mm Hg)* | 217 978 (17.1) | 150 779 (18.2) | 5899 (7.5) | 5733 (14.4) | 1291 (8.6) | 992 (7.3) | 53 284 (17.8) |
| CKD (stages 3–5)† | 70 965 (5.6) | 53 001 (6.4) | 2000 (2.5) | 1613 (4.1) | 350 (2.3) | 212 (1.6) | 13 789 (4.6) |
| Type 1 diabetes | 5440 (0.4) | 4019 (0.5) | 198 (0.3) | 107 (0.3) | 63 (0.4) | 32 (0.2) | 1021 (0.3) |
| Type 2 diabetes | 84 320 (6.6) | 58 586 (7.1) | 7443 (9.5) | 3605 (9.1) | 698 (4.7) | 685 (5.0) | 13 303 (4.4) |
| Obesity: BMI ≥30 kg/m2 | 247 753 (19.4) | 177 255 (21.4) | 9180 (11.7) | 10 624 (26.7) | 2411 (16.1) | 2175 (15.9) | 46 108 (15.4) |
| Active smoker | 242 599 (19.0) | 164 861 (19.9) | 10 366 (13.2) | 5884 (14.8) | 3100 (20.7) | 2768 (20.3) | 55 620 (18.6) |
| Ex-smoker | 441 499 (34.6) | 326 723 (39.4) | 16 671 (21.2) | 9172 (23.0) | 3773 (25.2) | 2946 (21.6) | 82 214 (27.5) |
| Cardiovascular diseases | |||||||
| Coronary artery disease | 44 141 (3.5) | 33 814 (4.1) | 1707 (2.2) | 360 (0.9) | 155 (1.0) | 166 (1.2) | 7939 (2.7) |
| Atrial fibrillation | 34 700 (2.7) | 26 684 (3.2) | 418 (0.5) | 226 (0.6) | 82 (0.5) | 92 (0.7) | 7198 (2.4) |
| Peripheral arterial disease | 10 027 (0.8) | 7812 (0.9) | 188 (0.2) | 111 (0.3) | 24 (0.2) | 22 (0.2) | 1870 (0.6) |
| Stroke and TIA | 27 222 (2.1) | 20 323 (2.5) | 606 (0.8) | 430 (1.1) | 96 (0.6) | 66 (0.5) | 5701 (1.9) |
| Congestive cardiac failure | 13 907 (1.1) | 10 452 (1.3) | 373 (0.5) | 226 (0.6) | 55 (0.4) | 60 (0.4) | 2741 (0.9) |
| Hypertension (diagnosed) | 234 838 (18.4) | 171 825 (20.7) | 8941 (11.4) | 7902 (19.9) | 1461 (9.8) | 1117 (8.2) | 43 592 (14.6) |
*Includes people with and without a diagnosis of hypertension.
†Number of people with condition (% of total column population), for example, CKD (stages 3–5) in in 53 001 (6.4) people in the white ethnic group.
BMI, body mass index; CKD, chronic kidney disease; SBP, systolic blood pressure; TIA, transient ischaemic attack.
Prevalence of CVD
The prevalence and number of conditions increased with age (online supplementary table S1). CVD was present in less than 1% of people under the age of 50 years, increasing to a quarter of people with at least one condition between 70 and 79 years old, and over 40% in people over the age of 80 years.
bmjopen-2017-020282supp002.pdf (485.2KB, pdf)
We identified a higher prevalence of CVDs from RCGP RSC data than identified by the P4P(QOF) definitions (figure 1A). We also detected a higher proportion of other CVD risk factors and comorbidities (figure 1B).
Figure 1.
Prevalence of (A) cardiovascular diseases and hypertension, and (B) risk factors/comorbidities in the RCGP RSC network, compared with all-English practices, using Quality Outcomes Framework data. CKD, chronic kidney disease; RCGP, Royal College of General Practitioner; RSC, Research and Surveillance Centre; TIA, transient ischaemic attack.
Incidence of CVD
Hypertension showed the highest incidence (2012–2016) and age-adjusted and gender-adjusted annual incidence rate followed by AF and coronary artery disease (table 3).
Table 3.
Incidence of cardiovascular disease and hypertension in the last 5 years, and age-adjusted and gender-adjusted annual incidence in the adult population
| Cardiovascular conditions | Incident cases in last 5 years, n (%) | Annual age-adjusted and gender-adjusted incidence rate per 10 000 (95% CI) |
| Coronary artery disease | 14 033 (1.1) | 22.1 (21.8 to 22.5) |
| Atrial fibrillation | 17 913 (1.4) | 28.2 (27.8 to 28.7) |
| Peripheral arterial disease | 3817 (0.3) | 6.0 (5.8 to 6.2) |
| Stroke/transient ischaemic attack | 10 838 (0.8) | 17.1 (16.8 to 17.5) |
| Congestive cardiac failure | 7792 (0.6) | 12.3 (12.0 to 12.6) |
| Hypertension | 60 635 (4.8) | 96.5 (95.7 to 97.3) |
CVD prescription use
ACE inhibitors, statins and calcium channel blockers were the most commonly prescribed, with over half of the CVD cohort prescribed at least one of these medications (online supplementary table S2).
Predictors of CVD
When considering predictors of all types of CVD, people were more likely to have the disease if they were older, male, current or ex-smokers, hazardous drinkers or categorised as an alcoholic, or had a well-established comorbidity/risk factor (chronic kidney disease, diabetes and hypertension) for CVD (table 4). In addition, likelihood of having CVD increased with each BMI category compared with people with a normal BMI, while people in the more deprived groups (IMD quintiles 1–4) were more likely to have CVD than those in the least deprived group (IMD quintile 5), and people of non-white ethnic groups were less likely to have CVD than people of white ethnicity. Interestingly, people with uncontrolled SBP were less likely to have CVD than those with controlled SBP. However, bivariate analysis showed that people were more likely to have CVD if their SBP was uncontrolled (OR 2.09, 95% CI 2.06 to 2.12), which suggests that adjusting for other variables (eg, age and hypertension diagnosis) affected this relationship.
Table 4.
Multivariable logistic regression for CVD
| Characteristic | OR | 95% CI | P values |
| Age | 1.093 | 1.093 to 1.094 | <0.001 |
| Gender (male) | 2.105 | 2.070 to 2.140 | <0.001 |
| Ethnicity | |||
| White | 1.000 (reference) | ||
| Asian | 0.849 | 0.810 to 0.890 | <0.001 |
| Black | 0.520 | 0.486 to 0.557 | <0.001 |
| Mixed | 0.731 | 0.648 to 0.825 | <0.001 |
| Missing | 0.697 | 0.615 to 0.791 | <0.001 |
| Other | 0.933 | 0.915 to 0.952 | <0.001 |
| IMD quintile | |||
| IMD quintile 5 (least deprived) | 1.000 (reference) | ||
| IMD quintile 4 | 1.047 | 1.025 to 1.070 | <0.001 |
| IMD quintile 3 | 1.082 | 1.057 to 1.107 | <0.001 |
| IMD quintile 2 | 1.155 | 1.127 to 1.185 | <0.001 |
| IMD quintile 1 (most deprived) | 1.230 | 1.199 to 1.263 | <0.001 |
| IMD quintile not available | 1.034 | 0.873 to 1.225 | 0.700 |
| Smoking status | |||
| Never | 1.000 (reference) | ||
| Active smoker | 1.565 | 1.526 to 1.605 | <0.001 |
| Ex-smoker | 1.381 | 1.355 to 1.407 | <0.001 |
| Not recorded | 1.056 | 1.005 to 1.111 | 0.032 |
| Alcohol consumption | |||
| Safe use | 1.000 (reference) | ||
| Non-drinker | 1.189 | 1.162 to 1.216 | <0.001 |
| Hazardous use | 1.217 | 1.195 to 1.241 | <0.001 |
| Alcoholism | 1.594 | 1.533 to 1.657 | <0.001 |
| Not recorded | 0.866 | 0.834 to 0.900 | <0.001 |
| BMI category | |||
| Normal | 1.000 (reference) | ||
| Underweight | 1.151 | 1.082 to 1.225 | <0.001 |
| Overweight | 1.136 | 1.114 to 1.159 | <0.001 |
| Obese class 1 | 1.352 | 1.320 to 1.384 | <0.001 |
| Obese class 2 | 1.597 | 1.544 to 1.651 | <0.001 |
| Obese class 3 | 1.858 | 1.776 to 1.944 | <0.001 |
| Not recorded | 0.803 | 0.769 to 0.839 | <0.001 |
| Diabetes | |||
| No diabetes | 1.000 (reference) | ||
| Type 1 diabetes | 2.410 | 2.165 to 2.683 | <0.001 |
| Type 2 diabetes | 1.609 | 1.576 to 1.642 | <0.001 |
| Unknown/other diabetes type | 2.038 | 1.407 to 2.951 | <0.001 |
| Other known risk factors | |||
| Uncontrolled SBP (>140 mm Hg) | 0.712 | 0.700 to 0.725 | <0.001 |
| Hypertension diagnosis | 1.734 | 1.704 to 1.764 | <0.001 |
| CKD (stages 3–5) | 1.653 | 1.620 to 1.688 | <0.001 |
BMI, body mass index; CKD, chronic kidney disease; CVD, cardiovascular disease; IMD, Index of Multiple Deprivation; SBP, systolic blood pressure.
Similar associations were found for coronary artery disease and AF (online supplementary tables S3 and S4); however, people of Asian ethnicity were more likely to have coronary artery disease than people of white ethnicity. Current and ex-smokers and people with type 1 diabetes were less likely to have AF than non-smokers and those without diabetes, respectively.
Discussion
Our results demonstrate a high prevalence of CVD, including hypertension, in a nationally representative population sample. A CVD diagnosis was recorded in more than one in five adults in our population. Regarding specific cardiovascular conditions, hypertension was the most commonly diagnosed cardiovascular condition. Coronary artery disease was diagnosed in 3.5% of individuals. This compares to previous UK studies which have demonstrated a fairly consistent prevalence of coronary artery disease at 3% in England, and 4% in Scotland, Northern Ireland and Wales.32 Additionally, our reported prevalence of AF, 2.7%, is similar to modelling studies undertaken by Public Health England to estimate the true burden of AF in England,33 which in itself is significantly higher than the national prevalence of AF reported by QOF/P4P. This highlights the benefit of our approach to case definition and the use of a wider range of clinical codes than used within P4P business rules. Classical risk factors for CVD were well represented within the RCGP RSC population. For example 234 838 individuals had a diagnosis of hypertension.
The RCGP RSC database is suitable for CVD research. We have defined a cohort of people with CVD (n=271 684) and identified a higher prevalence of patients than the P4P/QOF scheme: though we note QOF uses a limited list of codes. Incident cases of CVD and CVD risk factors can be identified in the cohort. We identified more obesity, our results are more closely aligned with self-reported data.34 Our use of ontologies to improve case finding may make our approach more reproducible than studies that rely on an expert’s code list.35
Data quality was high in the CVD cohort for demographic, clinical and social variables. These are similar to those reported in other UK primary care databases, such as Clinical Practice Research Datalink (CPRD).36 We were able to identify uncontrolled hypertension, a risk factor for major cardiovascular events,37 38 and highlighted an area where management could be improved through a combination of patient education and self-management and pharmacological interventions as appropriate.39 We could identify appropriate medicines used to manage CVD and found statins prescribed at anticipated rates.40
The associations with ethnicity were generally those expected. We found that a higher prevalence of people of white ethnicity had CKD, while the highest prevalence for type 2 diabetes was seen in those of black or Asian ethnicity.41 42 Similarly a higher prevalence of hypertension was seen in people of black ethnicity.43 44
The age-adjusted and gender-adjusted annual incidence of CVD was highest in AF and similar to other findings that have reported rates in Europe,45 while the incidence of stroke was higher than previously reported in other UK primary care databases (CPRD).46 This may reflect an increase in disease in the years since these findings were reported, or enhanced case-finding strategies.
Logistic regression analyses demonstrated that established risk factors for CVD were associated with CVD in the RCGP RSC dataset. Unusually, people with uncontrolled SBP were shown to be less likely to have CVD. However, bivariate analysis confirmed that this finding was only after adjusting for other variables, and uncontrolled SBP is in fact associated with higher risk of CVD. This variable was for the latest recording only for people with or without a diagnosis of hypertension, and therefore, many people who had uncontrolled SBP are healthy.
Strengths and weaknesses
Strengths
We have presented a comprehensive assessment of the burden of CVD in a representative population sample. Our use of coding ontologies improves the accuracy of case identification. RCGP RSC has promoted high-quality data recording for over half a century.
Although RCGP data have traditionally been used for monitoring infectious diseases, the network data have recently started to be used more for non-communicable disease (NCD) research. Examples include the epidemiology of type 3c diabetes,47 liver disease48 and risks associated with anticoagulant use in people with AF and CKD.17
Weaknesses
The weakness of our data is that of all routinely collected data.18 Data are collected for routine patient care in the context of the 10 min GP consultation. Our coded data are therefore a brief abstraction of this process.49 We are still in the early stages of looking to ensure the robustness of real-world evidence studies.
The incidence of myocardial infarction was lower than reported UK primary care data from linked primary care, hospital and national audit datasets (CPRD).50 This suggests there are limitations to using primary care data alone, though we have the capability to link to these same datasets.
Conclusions
We have reported the prevalence and incidence of common CVDs in the RCGP RSC dataset. Our reported prevalence of coronary artery disease and AF is comparable with other UK studies. Therefore, the RCGP RSC is ready to be more active in research and quality improvement studies including CVD. The rich clinical data available within the RCGP RSC have substantial potential utility for epidemiological research into a variety of communicable and NCD. This has recently been demonstrated using the RCGP RSC dataset to explore CVD outcomes in renal disease.17 The combination of clinical and prescription data can be used to carry out real-world evidence studies and explore effectiveness beyond the traditional randomised control trial setting.
Supplementary Material
Acknowledgments
Rachel Byford and Julian Sherlock, SQL developers, database management and data extraction, and the participating practices and patients for providing the data for this cohort. Collaboration with the GP computer system suppliers—EMIS, InPractice Systems and TPP; Apollo systems; Public Health England and our other funders and collaborators.
Footnotes
Contributors: WH led the analysis and drafting of the paper. AM and AC supported the statistical analysis of the paper. SdL led the development of the cohort, conceived the idea and contributed to the drafting of the paper. RC, TSH, PS and FF also contributed to the preparation and reviewing of the manuscript.
Funding: This study was funded internally by the Department of Clinical and Experimental Medicine at the University of Surrey.
Competing interests: None declared.
Patient consent: Not required.
Provenance and peer review: Not commissioned; externally peer reviewed.
Data sharing statement: The RCGP RSC data set can be accessed by researchers, approval is on a project-by-project basis (www.rcgp.org.uk/rsc). Ethical approval by an NHS Research Ethics Committee is needed before any data release/other appropriate approval. Researchers wishing to directly analyse the patient-level pseudonymised data will be required to complete information governance training and work on the data from the secure servers at the University of Surrey. Patient-level data cannot be taken out of the secure network. We encourage interested researchers to attend the short courses on how to analyse primary care data/RCGP RSC data offered twice a year.
References
- 1. World Health Organization. Cardiovascular diseases (CVDs). http://www.who.int/mediacentre/factsheets/fs317/en/ (accessed 17 May 2017).
- 2. GBD 2015 Mortality and Causes of Death Collaborators. Global, regional, and national life expectancy, all-cause mortality, and cause-specific mortality for 249 causes of death, 1980-2015: a systematic analysis for the Global Burden of Disease Study 2015. Lancet 2016;388:1459–544. 10.1016/S0140-6736(16)31012-1 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Lozano R, Naghavi M, Foreman K, et al. Global and regional mortality from 235 causes of death for 20 age groups in 1990 and 2010: a systematic analysis for the Global Burden of Disease Study 2010. Lancet 2012;380:2095–128. 10.1016/S0140-6736(12)61728-0 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. GBD 2015 Risk Factors Collaborators. Global, regional, and national comparative risk assessment of 79 behavioural, environmental and occupational, and metabolic risks or clusters of risks, 1990-2015: a systematic analysis for the Global Burden of Disease Study 2015. Lancet 2016;388:1659–724. 10.1016/S0140-6736(16)31679-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Rumsfeld JS, Joynt KE, Maddox TM. Big data analytics to improve cardiovascular care: promise and challenges. Nat Rev Cardiol 2016;13:350–9. 10.1038/nrcardio.2016.42 [DOI] [PubMed] [Google Scholar]
- 6. Kim J. Big data, health informatics, and the future of cardiovascular medicine. J Am Coll Cardiol 2017;69:899–902. 10.1016/j.jacc.2017.01.006 [DOI] [PubMed] [Google Scholar]
- 7. de Lusignan S, Metsemakers JF, Houwink P, et al. Routinely collected general practice data: goldmines for research? A report of the European Federation for Medical Informatics Primary Care Informatics Working Group (EFMI PCIWG) from MIE2006, Maastricht, The Netherlands. Inform Prim Care 2006;14:203–9. 10.14236/jhi.v14i3.632 [DOI] [PubMed] [Google Scholar]
- 8. Mavaddat N, Mant J. Primary care research and clinical practice: cardiovascular disease. Postgrad Med J 2010;86:696–703. 10.1136/pgmj.2009.094771 [DOI] [PubMed] [Google Scholar]
- 9. Langdown C, Peckham S. The use of financial incentives to help improve health outcomes: is the quality and outcomes framework fit for purpose? A systematic review. J Public Health 2014;36:251–8. 10.1093/pubmed/fdt077 [DOI] [PubMed] [Google Scholar]
- 10. de Lusignan S, van Weel C. The use of routinely collected computer data for research in primary care: opportunities and challenges. Fam Pract 2006;23:253–63. 10.1093/fampra/cmi106 [DOI] [PubMed] [Google Scholar]
- 11. Fleming DM. Weekly returns service of the royal college of general practitioners. Commun Dis Public Health 1999;2:96–100. [PubMed] [Google Scholar]
- 12. Correa A, Hinton W, McGovern A, et al. Royal College of General Practitioners Research and Surveillance Centre (RCGP RSC) sentinel network: a cohort profile. BMJ Open 2016;6:e011092 10.1136/bmjopen-2016-011092 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Queenan JA, Williamson T, Khan S, et al. Representativeness of patients and providers in the canadian primary care sentinel surveillance network: a cross-sectional study. CMAJ Open 2016;4:E28–32. 10.9778/cmajo.20140128 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Fleming DM, Schellevis FG, Paget WJ. Health monitoring in sentinel practice networks: the contribution of primary care. Eur J Public Health 2003;13(3 Suppl):80–4. 10.1093/eurpub/13.suppl_1.80 [DOI] [PubMed] [Google Scholar]
- 15. Fleming DM, Miles J. The representativeness of sentinel practice networks. J Public Health 2010;32:90–6. 10.1093/pubmed/fdp087 [DOI] [PubMed] [Google Scholar]
- 16. McGovern A, Hinton W, Correa A, et al. Real-world evidence studies into treatment adherence, thresholds for intervention and disparities in treatment in people with type 2 diabetes in the UK. BMJ Open 2016;6:e012801 10.1136/bmjopen-2016-012801 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Kumar S, de Lusignan S, McGovern A, et al. Ischaemic stroke, haemorrhage, and mortality in older patients with chronic kidney disease newly started on anticoagulation for atrial fibrillation: a population based study from UK primary care. BMJ 2018;360:k342 10.1136/bmj.k342 [DOI] [PubMed] [Google Scholar]
- 18. de Lusignan S. Codes, classifications, terminologies and nomenclatures: definition, development and application in practice. Inform Prim Care 2005;13:65–9. 10.14236/jhi.v13i1.580 [DOI] [PubMed] [Google Scholar]
- 19. Liaw ST, Rahimi A, Ray P, et al. Towards an ontology for data quality in integrated chronic disease management: a realist review of the literature. Int J Med Inform 2013;82:10–24. 10.1016/j.ijmedinf.2012.10.001 [DOI] [PubMed] [Google Scholar]
- 20. Liaw ST, Pearce C, Liyanage H, et al. An integrated organisation-wide data quality management and information governance framework: theoretical underpinnings. Inform Prim Care 2014;21:199–206. 10.14236/jhi.v21i4.87 [DOI] [PubMed] [Google Scholar]
- 21. Pathirannehelage S, Kumarapeli P, Byford R, et al. Uptake of a dashboard designed to give realtime feedback to a sentinel network about key data required for influenza vaccine effectiveness studies. Stud Health Technol Inform 2018;247:161–5. [PubMed] [Google Scholar]
- 22. Smith N, Livina V, Byford R, et al. Automated differentiation of incident and prevalent cases in primary care computerised Medical Records (CMR). Stud Health Technol Inform 2018;247:151–5. [PubMed] [Google Scholar]
- 23. de Lusignan S, Navarro R, Chan T, et al. Detecting referral and selection bias by the anonymous linkage of practice, hospital and clinic data using Secure and Private Record Linkage (SAPREL): case study from the evaluation of the Improved Access to Psychological Therapy (IAPT) service. BMC Med Inform Decis Mak 2011;11:61 10.1186/1472-6947-11-61 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24. Fleming DM, Durnall H. Ten lessons for the next influenza pandemic-an English perspective: a personal reflection based on community surveillance data. Hum Vaccin Immunother 2012;8:138–45. 10.4161/hv.8.1.18808 [DOI] [PubMed] [Google Scholar]
- 25. Tippu Z, Correa A, Liyanage H, et al. Ethnicity recording in primary care computerised medical record systems: an ontological approach. J Innov Health Inform 2017;23:799 10.14236/jhi.v23i4.920 [DOI] [PubMed] [Google Scholar]
- 26. Gov.UK. Department for communities and local government the english indices of deprivation. 2015. https://www.gov.uk/government/statistics/english-indices-of-deprivation-2015 (accessed 17 May 2017).
- 27. World Health Organization. Physical status: the use and interpretation of anthropometry. Report of a WHO Expert Committee, WHO Technical Report Series 85. 1995. http://apps.who.int/iris/bitstream/10665/37003/1/WHO_TRS_854.pdf (accessed 17 May 2017). [PubMed]
- 28. McGovern A, Feher M, Munro N, et al. Sodium-Glucose Co-Transporter-2 (SGLT2) inhibitors: comparing trial and real world use (Study Protocol). Diabetes Ther 2017;8:355–63. 10.1007/s13300-017-0229-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29. Hippisley-Cox J, Coupland C, Vinogradova Y, et al. Predicting cardiovascular risk in England and Wales: prospective derivation and validation of QRISK2. BMJ 2008;336:1475–82. 10.1136/bmj.39609.449676.25 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30. Health Research Authority. Defining research table - October 2017. 2017. http://www.hra-decisiontools.org.uk/research/docs/DefiningResearchTable_Oct2017-1.pdf (accessed 6th Feb 2018).
- 31. Royal College of General Practitioners (RCGP). Research and Surveillance Centre (RSC). www.rcgp.org.uk/rsc (accessed 16 Jun 2018).
- 32. Bhatnagar P, Wickramasinghe K, Wilkins E, et al. Trends in the epidemiology of cardiovascular disease in the UK. Heart 2016;102:1945–52. 10.1136/heartjnl-2016-309573 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33. Public Health England. Atrial Fibrillation prevalence estimates in England: application of recent estimates of AF in Sweden. 2015. https://www.gov.uk/government/uploads/system/uploads/attachment_data/file/644869/atrial_fibrillation_AF_briefing.pdf (accessed 22 Feb 2018).
- 34. NHS Digital. Health survey for England. 2015. http://www.content.digital.nhs.uk/catalogue/PUB22610 (accessed 3 Aug 2017).
- 35. Liyanage H, Liaw ST, Kuziemsky C, et al. Ontologies to improve chronic disease management research and quality improvement studies - a conceptual framework. Stud Health Technol Inform 2013;192:180–4. [PubMed] [Google Scholar]
- 36. Herrett E, Gallagher AM, Bhaskaran K, et al. Data resource profile: clinical practice research datalink (CPRD). Int J Epidemiol 2015;44:827–36. 10.1093/ije/dyv098 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37. Lewington S, Clarke R, Qizilbash N, et al. Age-specific relevance of usual blood pressure to vascular mortality: a meta-analysis of individual data for one million adults in 61 prospective studies. Lancet 2002;360:1903–13. [DOI] [PubMed] [Google Scholar]
- 38. Turnbull F, Neal B, Ninomiya T, et al. Effects of different regimens to lower blood pressure on major cardiovascular events in older and younger adults: meta-analysis of randomised trials. BMJ 2008;336:1121–3. 10.1136/bmj.39548.738368.BE [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39. NICE. Hypertension in adults: diagnosis and management. CG127. London, 2016. [Google Scholar]
- 40. Murphy C, Bennett K, Fahey T, et al. Statin use in adults at high risk of cardiovascular disease mortality: cross-sectional analysis of baseline data from The Irish Longitudinal Study on Ageing (TILDA). BMJ Open 2015;5:e008017 10.1136/bmjopen-2015-008017 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41. Dreyer G, Hull S, Aitken Z, et al. The effect of ethnicity on the prevalence of diabetes and associated chronic kidney disease. QJM 2009;102:261–9. 10.1093/qjmed/hcn177 [DOI] [PubMed] [Google Scholar]
- 42. Riste L, Khan F, Cruickshank K. High prevalence of type 2 diabetes in all ethnic groups, including Europeans, in a British inner city: relative poverty, history, inactivity, or 21st century Europe? Diabetes Care 2001;24:1377–83. 10.2337/diacare.24.8.1377 [DOI] [PubMed] [Google Scholar]
- 43. Cappuccio FP, Cook DG, Atkinson RW, et al. Prevalence, detection, and management of cardiovascular risk factors in different ethnic groups in south London. Heart 1997;78:555–63. 10.1136/hrt.78.6.555 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 44. Primatesta P, Bost L, Poulter NR. Blood pressure levels and hypertension status among ethnic groups in England. J Hum Hypertens 2000;14:143–8. 10.1038/sj.jhh.1000960 [DOI] [PubMed] [Google Scholar]
- 45. Zoni-Berisso M, Lercari F, Carazza T, et al. Epidemiology of atrial fibrillation: European perspective. Clin Epidemiol 2014;6:213–20. 10.2147/CLEP.S47385 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 46. Lee S, Shafe AC, Cowie MR. UK stroke incidence, mortality and cardiovascular risk management 1999-2008: time-trend analysis from the General Practice Research Database. BMJ Open 2011;1:e000269 10.1136/bmjopen-2011-000269 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 47. Woodmansey C, McGovern AP, McCullough KA, et al. Incidence, demographics, and clinical characteristics of diabetes of the exocrine pancreas (Type 3c): a retrospective cohort study. Diabetes Care 2017;40:1486–93. 10.2337/dc17-0542 [DOI] [PubMed] [Google Scholar]
- 48. Williams R, Alexander G, Armstrong I, et al. Disease burden and costs from excess alcohol consumption, obesity, and viral hepatitis: fourth report of the Lancet Standing Commission on Liver Disease in the UK. Lancet 2018;391:1097–107. 10.1016/S0140-6736(17)32866-0 [DOI] [PubMed] [Google Scholar]
- 49. Kumarapeli P, de Lusignan S. Using the computer in the clinical consultation; setting the stage, reviewing, recording, and taking actions: multi-channel video study. J Am Med Inform Assoc 2013;20:e67–e75. 10.1136/amiajnl-2012-001081 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 50. Herrett E, Shah AD, Boggon R, et al. Completeness and diagnostic validity of recording acute myocardial infarction events in primary care, hospital care, disease registry, and national mortality records: cohort study. BMJ 2013;346:f2350 10.1136/bmj.f2350 [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
bmjopen-2017-020282supp001.pdf (452.4KB, pdf)
bmjopen-2017-020282supp002.pdf (485.2KB, pdf)