Abstract

Background Statins are recommended for the secondary prevention of cardiovascular disease, although they are often used in suboptimal doses and some patients may not receive lipid-lowering therapy. The Primary Care Data Quality (PCDQ) programme is an audit-based educational intervention.

Objective To report the PCDQ programme’s effect on the cholesterol management in cardiovascular disease.

Subjects and methods Anonymized general practice data from 99 practices; 5% (n = 29 915) had cardiovascular diagnoses.

Results Mean cholesterol fell from 4.75 to 4.64 mmol l−1; patients achieving cholesterol target (< 5 mmol l−1) rose from 45.3 to 53.2%. Coronary heart disease patients achieved better control (mean 4.57 mmol l−1) than those with stroke (4.87 mmol l−1) or peripheral vascular disease (4.93 mmol l−1). Statin prescribing increased from 57.5 to 62.7%. Patients with diabetes [odds ratio (OR) 2.06, 95% confidence interval (95% CI) 1.91–2.21], prior myocardial infarction (MI) (OR 1.93, 95% CI 1.80–2.07), revascularization (OR 1.52, 95% CI 1.33–1.73) and smokers (OR 1.31, 95% CI 1.23–1.39) were more likely to receive statins, whereas people aged 75+ (OR 0.48, 95% CI 0.45–0.50), females (OR 0.90, 95% CI 0.86–0.94) and non-CHD-diagnosed (OR 0.36, 95% CI 0.34–0.38) were less likely.

Conclusions Diagnostic coding and number of patients who had their cholesterol measured and treated increased. There was no significant change in dosage used or inequity between the different groups prescribed statins.

Introduction

Lowering cholesterol using statin therapy reduces cardiovascular risk, particularly in people with pre-existing ischaemic heart disease. Consequently, guidelines for secondary prevention in coronary heart disease, stroke and peripheral vascular disease recommend cholesterol lowering in patients with cardiovascular disease. In the United Kingdom, the target is to reduce total cholesterol to <5mmoll−1. This guidance was first issued in the National Service Framework for Coronary Heart Disease (NSF CHD) in 20001 and subsequently reinforced in financially incentivized quality targets in the new 2005 contract for general practice which was included as Quality and Outcomes Framework (QOF).2 The value of lowering cholesterol with statins in people with ischaemic heart disease is well established in a range of large, well-designed, randomized controlled trials.3–6 The lipid-lowering effect seen with statin therapy is dose related,7 and the majority of the statin efficacy studies3–6 have used medium or high doses of statin.8,9 Statins also have a role in stroke prevention in that they reduce stroke in those with ischaemic heart disease.10 The effect does not appear to be related to total cholesterol levels, because those with a cholesterol level <5mmoll−1 also benefit from statin therapy.11,12 Peripheral vascular disease is an arterial disease whose management benefits from tight cholesterol control.1,13 Some groups notably the elderly (aged over 75 years),14 females and smokers are less likely to be offered treatment, whereas others with acute myocardial infarction (MI) and revascularization are more likely to receive therapy.15,16

The Primary Care Data Quality (PCDQ) programme is an audit-based educational intervention.17 PCDQ uses the feedback of routinely collected computer data to improve data quality and the quality of chronic disease management.18,19 The programme is deployed in clinical areas where there is a strong evidence base for an intervention, ideally, supported by National guidance, and where there is scope for change which can be implemented in primary care. Wherever possible, the programme is locally led, with PCDQ providing the technical expertise and an educational framework.20

This study set out to determine the impact of a quality improvement programme on cholesterol management in people with cardiovascular disease including identifying which patients receive which statin at what dose.

Method

From the participating practice perspective, the PCDQ programme consists of three steps: (i) baseline data collection, with feedback of data in an educational context; (ii) practices encouraged to develop local plans to address any quality issues identified; and (iii) a second data collection, to complete the audit cycle. The educational element consists of locality meetings at which comparative data are presented to representatives of participating practices—provision of written data summaries comparing the quality of care between practices, as represented by a practice’s computer records; we leave in each practice a list of patients who may require further review or intervention. With the permission of the practices, summary data are provided to the clinical lead in the local primary care organization.

Underpinning the programme is a sophisticated, but standardized methodology that enables us to consistently process data from different brands of GP computer system. We developed audit criteria and defined a data set to provide feedback on cholesterol management to practices as well as enabling us to report the effectiveness of the intervention. We take into account how these clinical concepts, especially diagnostic data,21 are likely to be coded by GPs, practice nurses and other practice staff. Morbidity Information Query and Export Syntax (MIQUEST), a Department of Health sponsored data extraction tool, was then used to extract data. Our first data extraction was from two pilot practices. This pilot output was then presented to the study group, and modifications were made to the MIQUEST data extraction queries; the final data set is summarized in Table 1.

Table 1

Variables included in the Primary Care Data Quality (PCDQ) cardiovascular programme

Patient identifiers + multiple sclerosis as data quality indictor 
1. MIQUEST unique ID 
2. YoB (year of birth) 
3. Sex 
4. First part Post Code (to link to socioeconomic data) 
5. Ethnicity (very poor levels of recording) 
6. MS code (multiple sclerosis is used as data quality index) 
7. Usual doctor code, to identify patterns between professionals 
Read coded variables 
1. Systolic BP (latest, value, + date recorded) 
2. Diastolic BP (latest, value, + date recorded) 
3. Height and weight (latest) 
4. Smoking (smoking habit data Y/N + date given of last code) 
    a. Smoking code that implies current smoker (+ date given) 
    b. Smoking code that implies non-smoker/ex-smoker 
    c. Smoking code NEVER smoked (+ date recorded) 
    d. Advice to stop smoking given to members of groups A + B above (Y/N + date given) 
5. Cholesterol and lipids 
    a. Total cholesterol (most recent, no mg, date recorded) 
    b. LDL (most recent, mmoll−1, date recorded) 
    c. HDL (most recent, mmoll−1, date recorded) 
    d. TGs (most recent, mmoll−1, date recorded) 
    e. TC/HDL ratio (most recent, value, date recorded) 
    f. We may need earliest available cholesterol and LDL and date 
    g. Date when first diagnosed ‘Hyperlipidaemia’ 
    h. Weight at time of diagnosis 
    i. Date first prescribed a lipid-lowering agent and first agent 
    j. Advice to adjust diet given to members of groups A + B above (Y/N + date given) 
6. Diabetes mellitus 
    a. Diagnosis codes 
    b. Blood glucose (most recent, mmoll−1, date recorded) 
    c. HbA1c (glycosylated haemoglobin most recent, date recorded) 
    d. Insulin dependence and date first prescribed if available 
    e. Date diagnosis made of diabetes mellitus 
    f. Weight at the time of diagnosis 
    g. Advice to adjust diet given to members of groups A + B above (Y/N + date given) 
7. Hypertension and date diagnosis made 
8. Last two lipid-lowering drug prescriptions issued (exact preparation including dose, number of tablets and date issued) 
9. Statin monitoring and side effects proxy 
    a. Transaminase (most recent, value, date) 
    b. CPK (most recent, value, date) 
Patient identifiers + multiple sclerosis as data quality indictor 
1. MIQUEST unique ID 
2. YoB (year of birth) 
3. Sex 
4. First part Post Code (to link to socioeconomic data) 
5. Ethnicity (very poor levels of recording) 
6. MS code (multiple sclerosis is used as data quality index) 
7. Usual doctor code, to identify patterns between professionals 
Read coded variables 
1. Systolic BP (latest, value, + date recorded) 
2. Diastolic BP (latest, value, + date recorded) 
3. Height and weight (latest) 
4. Smoking (smoking habit data Y/N + date given of last code) 
    a. Smoking code that implies current smoker (+ date given) 
    b. Smoking code that implies non-smoker/ex-smoker 
    c. Smoking code NEVER smoked (+ date recorded) 
    d. Advice to stop smoking given to members of groups A + B above (Y/N + date given) 
5. Cholesterol and lipids 
    a. Total cholesterol (most recent, no mg, date recorded) 
    b. LDL (most recent, mmoll−1, date recorded) 
    c. HDL (most recent, mmoll−1, date recorded) 
    d. TGs (most recent, mmoll−1, date recorded) 
    e. TC/HDL ratio (most recent, value, date recorded) 
    f. We may need earliest available cholesterol and LDL and date 
    g. Date when first diagnosed ‘Hyperlipidaemia’ 
    h. Weight at time of diagnosis 
    i. Date first prescribed a lipid-lowering agent and first agent 
    j. Advice to adjust diet given to members of groups A + B above (Y/N + date given) 
6. Diabetes mellitus 
    a. Diagnosis codes 
    b. Blood glucose (most recent, mmoll−1, date recorded) 
    c. HbA1c (glycosylated haemoglobin most recent, date recorded) 
    d. Insulin dependence and date first prescribed if available 
    e. Date diagnosis made of diabetes mellitus 
    f. Weight at the time of diagnosis 
    g. Advice to adjust diet given to members of groups A + B above (Y/N + date given) 
7. Hypertension and date diagnosis made 
8. Last two lipid-lowering drug prescriptions issued (exact preparation including dose, number of tablets and date issued) 
9. Statin monitoring and side effects proxy 
    a. Transaminase (most recent, value, date) 
    b. CPK (most recent, value, date) 

Codes were identified which are used in primary care to categorize people with cardiovascular disease. A broad definition of cardiovascular disease was used, as defined in the NSF CHD audit criteria, including cerebrovascular and peripheral arterial disease. Data were also collected for the individual component diseases: coronary heart disease, stroke (which for the purpose of this analysis includes people with transient ischaemic attacks) and peripheral vascular disease. Only cholesterol levels recorded within the previous 15 months (the standard set by the QOF2) were included in the analysis. Approximately 6 months later, following feedback and initiation of practice plans, a second data collection was made. Data were processed using a five-step procedure22 derived from an error reduction approach proposed by Berndt etal.23 These stages are (i) migration of the data into a data repository, in this case an My-SQL relational database; (ii) integration of the data with data from other practices; (iii) data cleaning; (iv) data processing; and (v) transfer of data into an appropriate statistical package for analysis. Data were analysed using SPSS (Statistical Package for Social Sciences) Version 12.

For the purposes of this analysis, the following data items were compared across the two data collections: (i) rate of recording of cardiovascular disease diagnoses including diabetes, standardized24 using the 2001 Census population for England and Wales;25 (ii) total cholesterol, including proportion of patients reaching the UK National target (<5mmoll−1); (iii) statin use (preparation and dose); and (iv) any change in demographics or subgroup of patients more or less likely to receive statins.

We record the ‘flux’ within the population denominator and index conditions to make any differences between the before and after populations transparent. We note the percentage change in the denominator and in the index conditions—in this study, ischaemic heart disease, cerebrovascular and peripheral vascular disease.

Statistical methods

Mean, standard deviation and standard error were used to describe normally distributed variables; non-parametric variables were described using both median and interquartile range and log-transformed geometric means and standard deviation. Independent sample t-tests were used to compare normally distributed continuous variables; Pearson’s chi-square test was used to test whether the proportions achieving treatment targets on or off therapy were significantly different. Logistic regression was used to characterize patients on and off statin therapy, according to a range of categorical variables. The Wald test was used to test significance.26

Results

The study population was taken from a registered practice population of 594059 people registered for both data collections at the 99 participating practices. The interval between data collections was ∼8 months, longer than the intended 6 months. The mean number of days between collections was 243 days (standard deviation 69.7 days). The demographic profile of this study group differed slightly from that of the UK population, in that adults aged 30–49, especially men, were over-represented in the sample, whereas children <5 years old and people aged 60–79 years were under-represented.

Before the intervention, 29094 individuals were identified with one or more cardiovascular diagnoses. After the intervention, this had risen by 3%, to 29915—55.5% were male (16588) and 44.5% were female (13327). Mean age for men was 68.3 years (standard deviation 11.9, standard error 0.092) and 72.4 years for women (standard deviation 12.9, standard error 0.078). The excess of younger people will reduce the apparent prevalence of coronary heart disease; hence, these results have been adjusted using the 2001 Census population for England age profile (Table 2). The age-adjusted prevalence for ischaemic heart disease is 4.03% (male 4.64%, female 3.40%), 1.86% for stroke/transient ischaemic attack (male 1.82%, female 1.90%), and 0.70% for peripheral vascular disease (male 0.83%, female 0.56%). Over the study period, the population registered with the study practices declined by 5.7% (34270); the proportions of leavers who had cardiovascular disease were similar to those in the study population as a whole: ischaemic heart disease 3.9%, stroke/transient ischaemic attack 2.1% and peripheral vascular disease 0.75%.

Table 2

Prevalence of coronary heart disease, stroke and transient ischaemic attack and peripheral vascular disease in the study sample

 Number of cases
 
  Age-adjusted prevalence
 
  
 Total Male Female Total (%) Male (%) Female (%) 
Coronary heart disease       
0–24 27 19 0.02 0.02 0.01 
25–44 359 247 112 0.19 0.24 0.13 
45–64 5806 3995 1811 4.39 5.81 2.85 
65–84 12801 7244 5557 19.16 24.54 14.81 
85+ 2029 703 1326 24.12 28.47 21.86 
All 21022 12208 8814 4.03 4.64 3.40 
Stroke/transient ischaemic attack       
0–24 73 35 38 0.05 0.05 0.05 
25–44 398 126 272 0.21 0.12 0.31 
45–64 2128 1182 946 1.61 1.72 1.49 
65–84 5761 2978 2783 8.63 10.09 7.42 
85+ 1348 431 917 16.02 17.46 15.12 
All 9708 4752 4956 1.86 1.82 1.90 
Peripheral vascular disease       
0–24 10 0.01 0.00 0.01 
25–44 98 45 53 0.05 0.04 0.06 
45–64 799 515 284 0.60 0.75 0.45 
65–84 2350 1431 919 3.52 4.85 2.45 
85+ 360 157 203 4.28 6.36 3.35 
All 3617 2151 1466 0.70 0.83 0.56 
 Number of cases
 
  Age-adjusted prevalence
 
  
 Total Male Female Total (%) Male (%) Female (%) 
Coronary heart disease       
0–24 27 19 0.02 0.02 0.01 
25–44 359 247 112 0.19 0.24 0.13 
45–64 5806 3995 1811 4.39 5.81 2.85 
65–84 12801 7244 5557 19.16 24.54 14.81 
85+ 2029 703 1326 24.12 28.47 21.86 
All 21022 12208 8814 4.03 4.64 3.40 
Stroke/transient ischaemic attack       
0–24 73 35 38 0.05 0.05 0.05 
25–44 398 126 272 0.21 0.12 0.31 
45–64 2128 1182 946 1.61 1.72 1.49 
65–84 5761 2978 2783 8.63 10.09 7.42 
85+ 1348 431 917 16.02 17.46 15.12 
All 9708 4752 4956 1.86 1.82 1.90 
Peripheral vascular disease       
0–24 10 0.01 0.00 0.01 
25–44 98 45 53 0.05 0.04 0.06 
45–64 799 515 284 0.60 0.75 0.45 
65–84 2350 1431 919 3.52 4.85 2.45 
85+ 360 157 203 4.28 6.36 3.35 
All 3617 2151 1466 0.70 0.83 0.56 

Cardiovascular disease comorbidity was observed in 14.2% of patients: 3863 (12.9%) had two diagnoses recorded and 396 (1.3%) had three. Comorbid patients were more likely to be older (men: mean age 72.2 years, median 73.3 and women: mean age 76.2 years, median 78.1): 28.2% of men and 15.8% of women in the sample had a history of acute MI; 18.4% of men and 14.9% of women had diabetes; 7.2% of men and 2.9% of women had undergone a coronary revascularization procedure; 25.4% of men and 21.7% of women were recorded as current or ex-smokers.

Pre-intervention, a recent cholesterol record (defined as within 15 months in the UK general practice contract) existed for 76.3% of patients. The arithmetic mean cholesterol was 4.86mmoll−1 [95% confidence interval (95% CI) 4.85–4.88]; geometric mean was 4.75mmoll−1 (95% CI 4.74–4.76). Given that cholesterol levels have a non-normal distribution (data not shown), a geometric mean is preferred to give a more meaningful result. Within the cardiovascular disease population, 45.3% achieved the target cholesterol of <5mmoll−1. There was a significant variation within this group, more of those on a statin (61.4%) achieved target, compared with 23.5% of those not taking a statin (chi-square, P < 0.001). There were also significant differences between patients with the three index diseases (Table3). People with CHD tended to have better control of cholesterol than those with stroke or peripheral vascular disease. Low-density lipoprotein cholesterol (LDL-C) was only recorded for 59% of people with cardiovascular disease overall, and there were quite marked interpractice variation in recording: median recording level 47%, interquartile range 19.2–68.1%.

Table 3

Pre- and post-intervention cholesterol levels in cardiovascular disease

 Pre-intervention   Post-intervention   
Any cardiovascular diagnosis On statins 16 733 (57.5%) Not on statins 12 361 (42.5%) All patients 29 094 (100%) On statins 18 761 (62.7%) Not on statins 11 154 (37.3%) All patients 29 915 (100%) 
    Cholesterol recorded (% of total) 92.3 54.6 76.3 95.2 62.9 83.1 
    Total cholesterol (geometric mean) 4.60 5.11 4.75 4.49 5.06 4.64 
    95% CI 4.59–4.62 5.09–5.17 4.74–4.76 4.47–4.51 5.03–5.08 4.63–4.65 
    Cholesterol < 5 mmol l−1 (% of total) 61.4 23.5 45.3 67.8 28.6 53.2 
Coronary heart disease On statins 13 678 (66.4%) Not on statins 6922 (33.6%) All patients 20 600 (100%) On statins 15 036 (71.5) Not on statins 5987 (28.5%) All patients 21 023 (100%) 
    Cholesterol recorded (% of total) 92.8 63.7 83.0 95.7 72.0 88.9 
    Total cholesterol (geometric mean) 4.56 5.03 4.67 4.45 4.97 4.57 
    95% CI 4.55–4.56 5.00–5.06 4.66–4.69 4.44–4.47 4.94–5.00 4.55–4.58 
    Cholesterol <5 mmol l−1 (% of total) 63.5 29.6 52.1 69.8 35.8 60.1 
Stroke (no coronary heart disease) On statins 2398 (35.5%) Not on statins 4352 (64.5%) All patients 6750 (100%) On statins 4100 (58.3%) Not on statins 2929 (41.7%) All patients 7029 (100%) 
    Cholesterol recorded (% of total) 89.0 42.3 58.9 94.1 55.2 69.2 
    Total cholesterol (geometric mean) 4.80 5.27 5.01 4.63 5.18 4.87 
    95% CI 4.75–4.84 5.23–5.32 4.98–5.04 4.59–4.67 5.14–5.23 4.85–4.88 
    Cholesterol <5 mmol l−1 (% of total) 52.0 15.6 28.5 43.0 28.7 37.1 
    Geometric mean total cholesterol: cardiovascular versus coronary heart disease   P < 0.00001   P < 0.0001 
Peripheral vascular disease (no coronary heart disease or stroke) On statins 656 (37.6%) Not on statins 1088 (62.4%) All patients 1744 (100%) On statins 816 (43.3%) Not on statins 1057 (56.7%) All patients 1863 (100%) 
    Cholesterol recorded (% of total) 91.9 45.9 63.2 92.3 58.9 70.4 
    Total cholesterol (geometric mean) 4.85 5.27 5.04 4.73 5.22 4.93 
    95% CI 4.76–4.94 5.17–5.37 4.97–5.10 4.66–4.81 5.13–5.31 4.89–4.98 
    Cholesterol <5 mmol l−1 (% of total) 51.5 16.3 29.5 56.5 19.5 36.0 
    Geometric mean total cholesterol: peripheral vascular disease versus coronary heart disease   P < 0.00001   P < 0.00001 
 Pre-intervention   Post-intervention   
Any cardiovascular diagnosis On statins 16 733 (57.5%) Not on statins 12 361 (42.5%) All patients 29 094 (100%) On statins 18 761 (62.7%) Not on statins 11 154 (37.3%) All patients 29 915 (100%) 
    Cholesterol recorded (% of total) 92.3 54.6 76.3 95.2 62.9 83.1 
    Total cholesterol (geometric mean) 4.60 5.11 4.75 4.49 5.06 4.64 
    95% CI 4.59–4.62 5.09–5.17 4.74–4.76 4.47–4.51 5.03–5.08 4.63–4.65 
    Cholesterol < 5 mmol l−1 (% of total) 61.4 23.5 45.3 67.8 28.6 53.2 
Coronary heart disease On statins 13 678 (66.4%) Not on statins 6922 (33.6%) All patients 20 600 (100%) On statins 15 036 (71.5) Not on statins 5987 (28.5%) All patients 21 023 (100%) 
    Cholesterol recorded (% of total) 92.8 63.7 83.0 95.7 72.0 88.9 
    Total cholesterol (geometric mean) 4.56 5.03 4.67 4.45 4.97 4.57 
    95% CI 4.55–4.56 5.00–5.06 4.66–4.69 4.44–4.47 4.94–5.00 4.55–4.58 
    Cholesterol <5 mmol l−1 (% of total) 63.5 29.6 52.1 69.8 35.8 60.1 
Stroke (no coronary heart disease) On statins 2398 (35.5%) Not on statins 4352 (64.5%) All patients 6750 (100%) On statins 4100 (58.3%) Not on statins 2929 (41.7%) All patients 7029 (100%) 
    Cholesterol recorded (% of total) 89.0 42.3 58.9 94.1 55.2 69.2 
    Total cholesterol (geometric mean) 4.80 5.27 5.01 4.63 5.18 4.87 
    95% CI 4.75–4.84 5.23–5.32 4.98–5.04 4.59–4.67 5.14–5.23 4.85–4.88 
    Cholesterol <5 mmol l−1 (% of total) 52.0 15.6 28.5 43.0 28.7 37.1 
    Geometric mean total cholesterol: cardiovascular versus coronary heart disease   P < 0.00001   P < 0.0001 
Peripheral vascular disease (no coronary heart disease or stroke) On statins 656 (37.6%) Not on statins 1088 (62.4%) All patients 1744 (100%) On statins 816 (43.3%) Not on statins 1057 (56.7%) All patients 1863 (100%) 
    Cholesterol recorded (% of total) 91.9 45.9 63.2 92.3 58.9 70.4 
    Total cholesterol (geometric mean) 4.85 5.27 5.04 4.73 5.22 4.93 
    95% CI 4.76–4.94 5.17–5.37 4.97–5.10 4.66–4.81 5.13–5.31 4.89–4.98 
    Cholesterol <5 mmol l−1 (% of total) 51.5 16.3 29.5 56.5 19.5 36.0 
    Geometric mean total cholesterol: peripheral vascular disease versus coronary heart disease   P < 0.00001   P < 0.00001 

Following the initial data collection and feedback process, a second data collection was performed. A greater proportion had their cholesterol measured (83.1%) (Table3) and reached target (53.2%): 10% more patients with ischaemic heart disease and stroke and 7% of patients with peripheral vascular disease reached targets. The mean cholesterol was lowered to 4.64mmoll−1 (arithmetic mean 4.75 mmoll−1).

As with the pre-intervention results, people with ischaemic heart disease appeared to be managed much more aggressively than those with stroke or peripheral vascular disease: 88.9% of those with ischaemic heart disease had had a recent cholesterol reading; mean total cholesterol was 4.67mmoll−1 and 60.1% achieved treatment goal. In contrast, only 69.2% of stroke patients without co-existing ischaemic heart disease had a cholesterol reading recorded. The mean cholesterol of those with a recording in stroke was 4.98mmoll−1 and 37.1% of the population reached target. The proportions were similar in peripheral vascular disease.

Before the PCDQ intervention, 57.5% of the patients with cardiovascular disease were taking a statin, which rose to 62.7% in the second data collection. Despite this increase in overall prescription of statins, the pattern of prescribing remained little changed, in terms of both agents used and the mean dose of each agent. Post-intervention, the mean dose of simvastatin was 24.3mg with more than 20% of patients receiving 10mg. Almost 60% of pravastatin patients were prescribed 40mg, whereas the remainder received 10 or 20mg (Table 4). Although 95.3% (n = 17858) of prescriptions for lipid-lowering drugs were for statins, other agents were used. Ezetimibe accounted for 2.4% (n = 443); fibrates for 1.5% (n = 342); fish oils, anion exchange resins and nicotinic acid were used rarely (n = 48, n = 31 and n = 19, respectively).

Table 4

Patterns of statin use before and after Primary Care Data Quality (PCDQ) intervention

 Patients taking a statin both pre- and post-intervention
 
     Patients taking a statin as a result of the intervention
 
  
 Pre-PCDQ
 
  Post-PCDQ
 
  n % Mean dose (mg) 
 n % Mean dose (mg) n % Mean dose (mg)    
Atorvastatin 5909 35.7 19.6 6081 36.8 20.8 657 33.8 17.9 
Fluvastatin 327 2.0 37.5 298 1.8 38.6 20 1.0 39.0 
Pravastatin 2090 12.6 29.6 1956 11.8 30.1 110 5.7 33.5 
Rosuvastatin 787 4.8 13.7 861 5.2 13.6 99 5.1 11.1 
Simvastatin 7427 44.9 24.6 7344 44.4 25.6 1055 54.4 27.0 
Any statin 16540   16540   1941   
 Patients taking a statin both pre- and post-intervention
 
     Patients taking a statin as a result of the intervention
 
  
 Pre-PCDQ
 
  Post-PCDQ
 
  n % Mean dose (mg) 
 n % Mean dose (mg) n % Mean dose (mg)    
Atorvastatin 5909 35.7 19.6 6081 36.8 20.8 657 33.8 17.9 
Fluvastatin 327 2.0 37.5 298 1.8 38.6 20 1.0 39.0 
Pravastatin 2090 12.6 29.6 1956 11.8 30.1 110 5.7 33.5 
Rosuvastatin 787 4.8 13.7 861 5.2 13.6 99 5.1 11.1 
Simvastatin 7427 44.9 24.6 7344 44.4 25.6 1055 54.4 27.0 
Any statin 16540   16540   1941   

Patients with total cholesterol >5mmoll−1 not treated with statins are unlikely to reach target. In the ischaemic heart disease population, only 53.1% (n = 15871) had a total cholesterol <5mmoll−1, either as a result of statin treatment (78.7%) or because this is their natural untreated level (21.3%). In 42.8 of the 46.9% who failed to reach target, this reflected inadequate dosing of lipid-lowering therapy, whereas the remainder (57.2%) were on no treatment at all. Although the absolute proportion of those reaching target varies according to the precise diagnosis, the underlying trend remains the same (data not shown).

Logistic regression was used to explore the factors that predict a failure to prescribe statins to people with a total cholesterol level of ≥5mmoll−1 (Table 5). Those with non-ischaemic heart disease, cardiovascular disease, stroke or peripheral vascular disease were less likely to be treated with a statin compared with people with ischaemic heart disease. Those aged 75 or over and women were significantly less likely to be treated [odds ratios (OR) 0.48 and 0.90, respectively]. People with uncontrolled hypertension (defined as blood pressure of >160/95mmHg, n = 685, 2.3% of CHD population) were significantly less likely to be on statins. However, patients with diabetes were significantly more likely to be treated (OR 2.06), as were smokers and ex-smokers and those who had undergone revascularization procedures. Post-intervention data show that there was no significant difference between pre- and post-intervention.

Table 5

Factors influencing likelihood of treatment with statins: logistic regression analysis

Risk factor Odds ratio 95% CI Coefficient SE Z P-value 
Diabetes 2.06 1.92–2.21 0.7214 0.0362 19.91 <0.00001 
Acute myocardial infarction 1.93 1.80–2.07 0.6585 0.0362 18.17 <0.00001 
Revascularization 1.52 1.33–1.73 0.4165 0.0665 6.26 <0.00001 
Smoker or ex-smoker 1.31 1.23–1.39 0.2711 0.0315 8.61 <0.00001 
Female sex 0.90 0.86–0.95 −0.1043 0.0264 −3.95 <0.0001 
Raised blood pressure (>160/95) 0.79 0.67–0.94 −0.2300 0.0837 −2.75 <0.01 
Aged 75+ 0.48 0.45–0.50 −0.7426 0.0265 −28.03 <0.00001 
Non-ischaemic heart disease index event 0.36 0.34–0.38 −1.0188 0.0290 −35.13 <0.00001 
Constant   0.8809 0.0273   
Risk factor Odds ratio 95% CI Coefficient SE Z P-value 
Diabetes 2.06 1.92–2.21 0.7214 0.0362 19.91 <0.00001 
Acute myocardial infarction 1.93 1.80–2.07 0.6585 0.0362 18.17 <0.00001 
Revascularization 1.52 1.33–1.73 0.4165 0.0665 6.26 <0.00001 
Smoker or ex-smoker 1.31 1.23–1.39 0.2711 0.0315 8.61 <0.00001 
Female sex 0.90 0.86–0.95 −0.1043 0.0264 −3.95 <0.0001 
Raised blood pressure (>160/95) 0.79 0.67–0.94 −0.2300 0.0837 −2.75 <0.01 
Aged 75+ 0.48 0.45–0.50 −0.7426 0.0265 −28.03 <0.00001 
Non-ischaemic heart disease index event 0.36 0.34–0.38 −1.0188 0.0290 −35.13 <0.00001 
Constant   0.8809 0.0273   

Chi-square statistic = 4062.3224, P < 0.00001.

Likelihood ratio = 4272.7503, P < 0.00001.

Discussion

Main findings of this study

More patients with cardiovascular disease have been identified in primary care. Good progress has been made towards treatment targets in ischaemic heart disease, with 60% of the population with a total cholesterol of <5mmoll−1. A greater proportion have their cholesterol measured and are being treated, largely with statins. However, about one-third of patients treated with a statin fail to reach the <5mmoll−1 cholesterol target. More than one-half of patients with cerebrovascular or peripheral vascular disease continue to have a total cholesterol of >5mmoll−1and one-half are not taking a statin. Where statins are being used, they tend to be prescribed at low doses and only a small proportion of patients have their dose increased, changed to a more potent statin or have other lipid-lowering agents added. There are a number of identifiable factors that determine the likelihood of an individual receiving a statin. Patients with diabetes, a past history of acute MI or coronary revascularization and smokers or ex-smokers were more likely to be treated. Conversely, treatment was less likely in women, those aged over 75 and patients with stroke, transient ischaemic attack or peripheral vascular disease as their index event. In addition, patients with a most recently recorded blood pressure >160/95mmHg were less likely to receive statin treatment. However, this is a small proportion of patients with cardiovascular disease (2.3%) and may represent those in the process of treatment adjustment.

What is already known on this topic

The prevalence of diseases included in the QOF was published in June 2005. The prevalence of ischaemic heart disease was 3.46–4.27%.27 The age-adjusted figure of 4.03% found in the present study fits within this range. For stroke, our prevalence of 1.86% falls well above the UK range of 1.33–1.76%. Although this difference appears small, on a population of the size of our sample (almost 600000), it represents a difference of between 550 and 1760 patients. One likely explanation for this discrepancy is the relatively limited range of Read codes used to define cerebrovascular disease included in QOF. In the present study, all possible codes were included. This phenomenon may help explain why there is a cohort of under-treated patients with non-ischaemic heart disease diagnoses.28 With computerized decision support only identifying QOF-qualifying patients, it is not surprising that the broader secondary prevention falls short in this subset.

The data fit with other studies, which have shown suboptimal control of abnormal lipid profiles post-MI29 and in people with stroke.30 They also correspond with previous findings that older people and women are less likely to be prescribed statins, whereas those with MI and revascularization are more likely.14–16 Our finding that smokers were more likely to be prescribed statin treatment, however, is at variance with other published data. This may reflect that we pooled both current and ex-smokers in our analysis, whereas others have examined current smoking status only. Our data regarding the suboptimal dosing of statins are consistent with the pattern seen elsewhere.31,32 It may be that data which cast doubt on the value of intensive statin therapy reinforce the use of low-dose therapy33—a similar phenomenon is seen in the management of heart failure.34,35

What this study adds

To achieve all the relevant quality points and receive maximum quality-related pay, practices have to record cholesterol within the last 15 months in 90% of patients and 60% of patients with cardiovascular disease must have total cholesterol of <5mmoll−1.2 These data show that participating practices reached the highest quality standard half-way through the first year of the new GP contract. However, behind the quality points, there is considerable unmet need.

Certain groups appear less likely to receive statins. The association between female sex and the likelihood of statin treatment is small (OR 0.90) but highly significant (95% CI 0.86–0.95; P < 0.0001). It is possible that this reflects a confounding factor that has not been taken into account in our model. The evidence for discrimination against the elderly (aged over 75 years) is less equivocal (OR 0.48, 95% CI 0.45–0.50, P < 0.00001) and reflects a genuine dilemma. On the one hand, no randomized control trials of statin therapy have studied patients beyond their early 80s, and in one study, there was a suggestion of tailing-off of effect in the older age group.36 On the other hand, meta-analysis of all statin studies shows a remarkable degree of homogeneity between treatment groups,37 with some evidence for enhanced treatment benefit amongst the very elderly.38 It is difficult to justify this apparent bias against secondary prevention patients aged >75 years.

Limitations of the study

There is no evidence that the PCDQ intervention resulted in the changes reported. Financial incentives within the QOF may have been much more important. However, a cynical approach to purely increase QOF points would be directed towards achieving monitoring and control in people with an existing diagnosis, rather than finding new cases. The intervention failed to encourage the use of higher doses of statins or greater equity in the use of statins.

The study may under-report the quality of care; computer-generated searches of data can also miss patients.39–41 Some patients may have been found to have an elevated cholesterol level and initiated on statin therapy; their improved cholesterol will not as yet have been recorded. We only reported structured or Read coded data. It is not possible to extract narrative or free text data in a format that can be analysed readily. Therefore, if cholesterol and other data were contained in letters or written in text, it would not be included in the study.

We did not report LDL-C because, although we extracted the data, recordings were variable and some laboratories routinely report HDL-C and HDL-C/total cholesterol ratio rather than LDL-C. We were unable to identify individuals who had satisfied NSF criteria by achieving a reduction of 30% or more, while still retaining post-intervention cholesterol of >5mmoll−1, as there was no reliable way of ascertaining pre-treatment cholesterol levels from the PCDQ database. However, it is likely that the numbers in this category were small, as to remain above the 5mmoll−1 target would have required a pre-treatment level of >7.15mmoll−1. Data derived from a previously published study would suggest that fewer than 5% of untreated CHD patients in the United Kingdom would potentially fulfil this criterion.17

Call for further research

We need to understand the rationale for prescribing low-dose statins to patients not achieving current cholesterol targets. The failure to increase dose, switch statin or add another lipid-lowering agent for these patients is difficult to understand.

Conclusions

Practitioners have improved cholesterol management across the population with cardiovascular disease; however, much remains to be done to improve the management of cholesterol in people with stroke/transient ischaemic attack and peripheral vascular disease. Practitioners who achieve maximum quality points may feel that they have achieved the highest possible standard of cholesterol management and be unaware of the unmet need. There is scope to tighten quality targets, either by lowering the cholesterol target or by increasing the proportion of patients that should achieve it. More needs to be done to improve the basics of cholesterol management in coronary heart disease; we need to fill the gaps in statin prescribing and reduce the use of suboptimal doses.

References

1
Department of Health. National Service Framework: Coronary Heart Disease.
London 2000.
  http://www.dh.gov.uk/assetRoot/04/05/75/19/04057519.pdf (
11
July
2006
, date last accessed).
2
Department of Health. NHS Confederation and British Medical Association.
Quality and Outcomes Framework (QOF) Guidance and Evidence Base
 . http://www.dh.gov.uk/assetRoot/04/08/86/93/04088693.pdf (
11
July
2006
, date last accessed).
3
Scandinavian Simvastatin Survival Study Group. Randomised trial of cholesterol lowering in 4444 patients with coronary heart disease: the Scandinavian Simvastatin Survival Study (4S).
Lancet
 
1994
;
344
:
1383
–9.
4
Sacks
FM
, Pfeffer MA, Moye LA et al. The effect of pravastatin on coronary events after myocardial infarction in patients with average cholesterol levels.
N Engl J Med
 
1996
;
335
:
1001
–9.
5
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
 
1995
;
339
:
1349
–57.
6
Heart Protection Study Collaborative Group. MRC/BHF Heart Protection Study of cholesterol lowering with simvastatin in 20,536 high-risk individuals: a randomised placebo-controlled trial.
Lancet
 
2002
;
360
:
7
–22.
7
LaRosa
JC
, Grundy SM, Waters DD et al. Intensive lipid lowering with atorvastatin in patients with stable coronary disease.
N Engl J Med
 
2005
;
352
:
1425
–35.
8
Law
MR
, Wald NJ, Rudnicka AR. Quantifying effect of statins on low density lipoprotein cholesterol, ischaemic heart disease, and stroke: systematic review and meta-analysis.
BMJ
 
2003
;
326
:
1423
–7.
9
Edwards
JE
, Moore RA. Statins in hypercholesterolaemia: a dose-specific meta-analysis of lipid changes in randomised, double blind trials.
BMC Fam Pract
 
2003
;
4
:
18
–37.
10
Bedi
A
, Flaker GC. How do HMG-CoA reductase inhibitors prevent stroke?
Am J Cardiovasc Drugs
 
2002
;
2
:
7
–14.
11
Amarenco
P
, Lavallee P, Touboul PJ. Statins and stroke prevention.
Cerebrovasc Dis
 
2004
;
17
:
81
–8.
12
Briel
M
, Studer M, Glass TR et al. Effects of statins on stroke prevention in patients with and without coronary heart disease: a meta-analysis of randomized controlled trials.
Am J Med
 
2004
;
117
:
596
–606.
13
Kim
CK
, Schmalfuss CM, Schofield RS et al. Pharmacological treatment of patients with peripheral arterial disease.
Drugs
 
2003
;
63
:
637
–47.
14
Hippisley-Cox
J
, Pringle M, Cater R et al. Coronary heart disease prevention and age inequalities: the first year of the National Service Framework for CHD.
Br J Gen Pract
 
2005
;
55
:
369
–75.
15
Reid
FD
, Cook DG, Whincup PH. Use of statins in the secondary prevention of coronary heart disease: is treatment equitable?
Heart
 
2002
;
88
:
15
–9.
16
DeWilde
S
, Carey IM, Bremner SA et al. Evolution of statin prescribing 1994–2001: a case of agism but not of sexism?
Heart
 
2003
;
89
:
417
–21.
17
de Lusignan
S
, Dzregah B, Hague N et al. Cholesterol management in patients with ischaemic heart disease: an audit-based appraisal of progress towards clinical targets in primary care.
Br J Cardiol
 
2003
;
10
:
223
–8.
18
de Lusignan
S
, Hague N, Brown A et al. An educational intervention to improve data recording in the management of ischaemic heart disease in primary care.
J Public Health
 
2004
;
28
:
34
–7.
19
St
.
George
’s – University of London.
PCDQ (Primary Care Data Quality) Programme. What is MIQUEST?
  http://www.pcdq.org/miquest.htm (
11
July
2006
, date last accessed).
20
de Lusignan
S.
Using the feedback of routinely collected computer data to improve the quality of chronic disease management in UK primary care.
Acta Informatica Medica
 
2005
;
13
:
59
–63.
21
Faulconer
E
, de Lusignan S. An eight-step method for assessing diagnostic data quality in practice: chronic obstructive pulmonary disease as an exemplar.
Inform Prim Care
 
2004
;
12
:
243
–54.
22
van Vlymen
J
, de Lusignan S, Hague N et al. Ensuring the quality of aggregated general practice data: lessons from the Primary Care Data Quality Programme (PCDQ).
Stud Health Technol Inform
 
2005
;
116
:
1010
–15.
23
Berndt
DJ
, Fisher JW, Hevner AR et al. Healthcare data warehousing and quality assurance.
IEEE Computer
 
2001
;
34
:
56
–65.
24
NHS Public Health Network.
Directly Age-Standardised Rates.
  http://www.avon.nhs.uk/phnet/methods/directly_age_standardised_rates. htm (
17
June
2005
, date last accessed).
25
National Statistics.
Population Trends
 .
2003
;113 (Autumn):45 (Table .4). http://www.statistics.gov.uk/downloads/theme_population/PT113.pdf (
11
July
2006, date last accessed).
26
North Carolina State University.
Logistic Regression
 . http://www2.chass. ncsu.edu/garson/pa765/logistic.htm (
11
July
2006
, date last accessed).
27
Department of Health.
Prevalence Recorded for Diseases in the New GMS Contract for 2004/5.
  http://www.gponline.com/attachments/pdfs/UKprevalence.pdf (
17
June
2005
, date last accessed).
28
Williams
P
, de Lusignan S. Does a higher ‘quality points’ score mean better care in stroke? An audit of general practice medical records.
Inform Prim Care
 
2006
;
14
:
29
–40.
29
Baessler
A
, Fischer M, Huf V et al. Failure to achieve recommended LDL cholesterol levels by suboptimal statin therapy relates to elevated cardiac event rates.
Int J Cardiol
 
2005
;
101
:
293
–8.
30
Rudd
AG
, Lowe D, Hoffman A et al. Secondary prevention for stroke in the United Kingdom: results from the National Sentinel Audit of Stroke.
Age Ageing
 
2004
;
33
:
280
–6.
31
Mantel-Teeuwisse
AK
, Klungel OH, Schalekamp T et al.
Suboptimal choices and dosing of statins at start of therapy. Br J Clin Pharmacol
 
2005
;
60
:
83
–9.
32
Bazalo
GR.
Managed care trends in statin usage.
Manag Care
 
2001
;
10
:
48
–59.
33
Pitt
B.
Low-density lipoprotein cholesterol in patients with stable coronary heart disease – is it time to shift our goals?
N Engl J Med
 
2005
;
352
:
1483
–4.
34
Toal
M
, Walker R. Angiotensin converting enzyme (ACE) inhibitors in the treatment of heart failure in general practice in north Cumbria.
Eur J Heart Fail
 
2000
;
2
:
201
–7.
35
Porcu
M
, Opasich C, Scherillo M et al. Improving practice patterns in heart failure through a national cardiological network: the case of ACE-inhibitors.
Ital Heart J
 
2002
;
3
:
730
–7.
36
Shepherd
J
, Blauw GJ, Murphy MB et al. Pravastatin in elderly individuals at risk of vascular disease (PROSPER): a randomised controlled trial.
Lancet
 
2002
;
360
:
1623
–30.
37
Vrecer
M
, Turk S, Drinovec J et al. Use of statins in primary and secondary prevention of coronary heart disease and ischemic stroke. Meta-analysis of randomized trials.
Int J Clin Pharmacol Ther
 
2003
;
41
:
567
–77.
38
Allen
Maycock CA
, Muhlestein JB, Horne BD et al. Statin therapy is associated with reduced mortality across all age groups of individuals with significant coronary disease, including very elderly patients.
J Am Coll Cardiol
 
2002
;
40
:
1777
–85.
39
Thiru
K
, De Lusignan S, Sullivan F et al. Three steps to data quality.
Inform Prim Care
 
2003
;
11
:
95
–102.
40
Thiru
K
, Hassey A, Sullivan F. Systematic review of scope and quality of electronic patient record data in primary care.
BMJ
 
2003
;
326
:
1070
–5.
41
Gray
J
, Majeed A, Kerry S et al. Identifying patients with ischaemic heart disease in general practice: cross sectional study of paper and computerised medical records.
BMJ
 
2000
;
321
:
548
–50.