Undertreatment and overtreatment with statins: the Oslo Health Study 2000–2001

Authors


Serena Tonstad MD, PhD, Department of Preventive Cardiology, Preventive Medicine Clinic, Ullevål University Hospital, N-0047 Oslo, Norway (fax: +47 22 11 99 75; e-mail: serena.tonstad@ulleval.no).

Abstract.

Objective.  We examined the prevalence and factors associated with use of cholesterol-lowering statins in the population.

Methods.  Demographic, medical, anthropometric and lifestyle data was obtained from 6233 men and 7521 women born in 1924/25, 1940/41, 1955 and 1960 that participated in the Oslo Health Study 2000–2001. A nonfasting blood sample was collected.

Results.  Of subjects with a heart attack, angina, stroke or diabetes 45% of men and 35% of women were taking a statin (P < 0.001). Of subjects with cardiovascular disease (CVD) or diabetes taking statins 61% of men and 40% of women achieved total serum cholesterol levels ≤5 mmol L−1. The odds ratio for taking a statin was increased amongst subjects who also took antihypertensive drug(s) or acetylsalicylic acid, subjects with a family history of coronary heart disease (CHD) and women who had visited the general practitioner within the last year. Amongst presumed healthy subjects use of statins increased from about 1% in women aged 40–45 years, to 7% at age 60 and to 12% at age 75 whilst the corresponding figures for men were 3%, 8% and 9%, respectively. About 22% of men but <2% of women aged 60 who were not taking statins had a 10-year Framingham CHD risk score >20%. Determinants of statin use were similar to those amongst subjects with CVD or diabetes.

Conclusion.  People with CVD or diabetes remain undertreated with statins, women more so than men. Use of other preventive drugs, the family history and recent contact with the general practitioner were the most important determinants of statin use in primary and secondary prevention. Amongst healthy subjects aged 60 or 75 years women received statins disproportionately to their low CHD risk compared with men.

Introduction

Norwegian/Swedish guidelines for the prevention and management of coronary heart disease (CHD) are in line with European recommendations in suggesting cholesterol-lowering therapy for most patients with established CHD, other cardiovascular disease (CVD) or diabetes and for people without known atherosclerotic disease who are at high coronary risk because of a number of risk factors [1, 2]. The current guidelines suggest that healthy people with a risk of coronary disease of >20% in the next 10 years based on the Framingham risk equation are at high risk and should receive intensive lifestyle and drug therapy [1]. The aim of cholesterol-lowering therapy is to achieve total and LDL cholesterol levels <5 mmol L−1 and <3 mmol L−1, respectively. Statins are the preferred cholesterol-lowering drug for the prevention of CHD because of clinical trial evidence showing that they substantially reduce CHD morbidity and reduce mortality in men with previous CHD [3–5]. Despite the guidelines not all eligible patients with established CVD receive cholesterol-lowering therapy in clinical practice [6, 7]. Both undertreatment and overtreatment with statins have been documented [7]. Amongst healthy individuals, therapy may not be directed towards the most appropriate drug candidates in terms of overall risk [8–10]. For example, women and nonsmokers may be more likely to receive statins than their counterparts [8, 11]. The elderly or people with angina may be less likely to be treated than their counterparts with other clinical manifestations of ischaemic heart disease [12]. As a consequence of these inequities the benefits of cholesterol-lowering drugs projected from clinical trials may not apply to clinical practice.

Previous studies have examined medical records at tertiary care centres or general practices or prescription databases to determine the prevalence of statin use [6–10, 12]. Data regarding determinants of statin use in a population-based sample including subjects with and without known atherosclerotic disease is scarce. The Oslo Health Study 2000–2001 was a health survey that aimed in part to assess cardiovascular risk factors in selected birth cohorts living in Oslo in 2000–2001. In the present report, we examined the use of cholesterol-lowering drugs, primarily statins, amongst the men and women that participated in the survey. Moreover, we investigated factors associated with statin use in subjects grouped according to whether they had a history of CHD, diabetes or stroke and in presumed healthy subjects.

Material and methods

The Oslo Health Study 2000–2001 was a collaboration between the National Health Screening Service of Norway (now the Norwegian Institute of Public Health), the University of Oslo and the municipality of Oslo. All individuals in the county of Oslo born in 1924/25, 1940/41, 1955, 1960 and 1970 were invited to take part in the study. Subjects born in 1924/25 and 1940/41 are referred to as 75- and 60-year-old individuals, respectively, in this report. Subjects born in 1970 are not included in this report because only 1.1% of men and 0.8% of women reported use of cholesterol-lowering drugs. The response rate was 52% and 39% for women and men, respectively, in the 40- and 45-year age group, 57% and 53%, respectively, in the 60-year age group and 50% and 58%, respectively, in the 75-year age group after up to two reminders were sent. The study protocol was evaluated by the Regional Ethics Committee and approved by the Norwegian Data Inspectorate.

Baseline measurements included height, weight, blood pressure and nonfasting analyses of serum total cholesterol, HDL cholesterol, triglycerides and glucose. One self-administered questionnaire was part of the letter of invitation. In addition, two questionnaires were handed out at the site of the survey and were to be sent back in a stamped addressed envelope [13]. The questionnaires provided information on health status, symptoms, diseases, health behaviours including whether the respondent had visited their general practitioner within the last year and use of medications. In regard to health status, participants were asked to check yes or no as to whether they had experienced a number of illnesses including diabetes, heart attack, angina pectoris and stroke. They also responded to two questions regarding their use of cholesterol-lowering drugs. One question asked whether the respondent took a cholesterol-lowering drug at the present time, had taken a drug previously but not at present or had never taken a drug. The second question asked the respondent to list the names of drugs used for cholesterol-lowering (and for selected other conditions). In addition, participants were asked whether one or more of their parents or siblings had experienced a myocardial infarction or angina pectoris.

Of a total of 6578 men and 8078 women participants in the survey in the defined age groups, 6415 (97.5%) of men and 7785 (96.4%) had responded to the question on health status and could be classified for the purposes of this study as either presumed healthy or having had at a self-report of least one of diabetes, heart attack, angina or stroke. Of these subjects 97.2% of men and 96.6% of women also could be classified as users or nonusers of cholesterol-lowering drugs. This left a total of 6233 men and 7521 women in the study, of whom 888 men (14.2%) and 683 women (9.1%) reported CVD (heart attack, angina or stroke) or diabetes (or a combination).

The estimated 10-year incidence of CHD was calculated for healthy subjects aged 40, 45 or 60 years by entering into the Framingham risk model the following variables: age, gender, systolic blood pressure, diabetes, total and HDL cholesterol levels and current smoking status [14]. The Framingham risk score allows the risk estimation of individuals aged 30–74 years over a 12-year period. In the current analysis, left ventricular hypertrophy was considered to be absent. Subjects with 10-year risk scores >20% are considered to be at high risk of CHD [1, 2].

Statistical methods

Confidence intervals for proportions were calculated using the continuity corrected version of the score [15]. Odds ratios with 95% confidence intervals were estimated separately for men and women through logistic regression with adjustment for age, family history, visit to the general practitioner in the last year, cigarette smoking, use of antihypertensive drugs or acetylsalicylic acid, body mass index (BMI), educational and physical activity levels. In a second model level of HDL cholesterol was added. The two models yielded very similar results, and the results of the second model are presented. All analyses were carried out using SPSS 10.0 for Windows. Bivariate associations were evaluated using chi-square tests. Level of significance was set to P < 0.05.

Results

Of men and women who reported use of a cholesterol-lowering drug, 71.6% and 76.6%, respectively, also responded to the question on the name of the drug. These were statins in all but <0.1% of cases, and thus we refer to statins further in this paper. The most commonly used statin was simvastatin (55.0%), followed by atorvastatin (30.6%), pravastatin (7.9%), lovastatin (3.4%), cerivastatin (2.0%) and fluvastatin (1.1%).

Of subjects with one of heart attack, angina, stroke or diabetes (N = 888 men and 683 women), 44.9% of men and 35.0% of women took a statin (P < 0.001). In this group, 4.1% of men and 3.4% of women had used a statin previously (data not shown). Use of statins was lower in men and women aged 40 and 45 years than in subjects aged 60 or 75 years (P < 0.001 for each gender). Subjects with a history of heart attack tended to be more likely to be taking statins than subjects with angina, who in turn were more likely to be taking statins than subjects with stroke or diabetes, but confidence intervals for the younger age groups were wide and overlapped (Table 1).

Table 1.  Use of cholesterol-lowering drugs according to age, sex and presence of cardiovascular disease (CVD) or diabetes
Age (years)CategoryMen N (%, 95% CI)Women N (%, 95% CI)
75Heart attack123 (60.0, 52.9–66.7)62 (54.9, 45.3–64.2)
Angina123 (57.2, 50.3–63.9)101 (47.6, 40.8–54.6)
Stroke44 (32.4, 24.7–40.9)38 (29.0, 21.6–37.7)
Diabetes37 (31.6, 23.5–41.0)32 (33.0, 24.0–43.4)
Any of the above212 (44.6, 40.1–49.2)167 (38.7, 34.1–43.4)
None of the above84 (8.9, 7.2–10.9)168 (11.9, 10.3–13.8)
60Heart attack90 (73.8, 64.9–81.1)16 (55.2, 36.0–73.1)
Angina85 (66.9, 58.0–74.9)33 (60.0, 45.9–72.7)
Stroke21 (33.9, 22.7–47.1)16 (33.3, 20.8–48.5)
Diabetes42 (38.2, 29.2–48.0)18 (31.0, 19.9–44.7)
Any of the above161 (51.4, 45.8–57.1)58 (37.9, 30.3–46.1)
None of the above132 (7.9, 6.7–9.3)145 (7.1, 6.1–8.4)
40 and 45Any disease26 (26.0, 18.0–35.9)14 (14.3, 8.3–23.2)
None of the above79 (2.9, 2.3–3.6)47 (1.4, 1.0–1.9)

Amongst presumed healthy subjects, 5.5% of men and 5.3% of women took a statin (P = 0.5). This proportion increased with age. By the age of 75 years, more women than men were taking a statin (P = 0.02). Amongst these subjects previous use was 0.7% and 1.3% for men and women (P = 0.2). Amongst nonusers of statins, no women and very few men aged 40 and 45 years had a Framingham risk score >20%; by age 60 about 22% of men and <2% of women not taking statins had a high risk score (Table 2).

Table 2.  Framingham risk scores [10-year coronary heart disease (CHD) risk] of presumed healthy subjects not taking statins in the Oslo Health Study 2000–2001
Age (years)MenWomen
Mean (SD)Percentage with score >20%Mean (SD)Percentage with score >20%
40 + 456.5 (4.0)0.62.1 (2.1)0.1
6016.0 (6.8)22.47.7 (4.4)1.7

Mean levels of total serum cholesterol were lower amongst subjects who took cholesterol-lowering drugs compared with nonusers, with the exception of healthy subjects aged 40 or 45 years and women with CVD or diabetes aged 40–45 years (Table 3). HDL cholesterol levels tended to be lower in users than nonusers of cholesterol-lowering drugs. This difference was statistically significant in the youngest age group of 40–45 years and in healthy men and women age 60 years (Table 3). Of men and women with CVD or diabetes taking statins, 61% and 40%, respectively, had a total serum cholesterol level ≤5 mmol L−1; this proportion was 30% amongst healthy men and 20% amongst healthy women.

Table 3.  Total serum cholesterol and HDL cholesterol levels amongst users and nonusers of cholesterol-lowering drugs (mainly statins) in the Oslo Health Study 2000–2001
 MenWomen
Use statinsNo statinsUse statinsNo statins
  1. Mean ± SD values are shown.

  2. *P-value < 0.05; **P-value < 0.01; ***P-value < 0.001.

Subjects with cardiovascular disease or diabetes, N (% with available data)N = 354 (88.7%)N = 430 (94.9%)N = 225 (94.1%)N = 395 (93.8%)
 Total cholesterol (mmol L−1)
  Age 60 + 754.84 ± 0.925.75 ± 0.98***5.37 ± 1.106.23 ± 1.04***
  Age 40 + 454.87 ± 1.135.49 ± 1.11*5.39 ± 1.095.45 ± 0.87
 HDL cholesterol (mmol L−1)
  Age 60 + 751.31 ± 0.371.32 ± 0.411.60 ± 0.471.59 ± 0.44
  Age 40 + 451.12 ± 0.251.33 ± 0.40*1.10 ± 0.251.43 ± 0.37**
Presumed healthy subjects, N (% with available data)N = 272 (92.2%)N = 4811 (96.3%)N = 347 (96.4%)N = 6290 (97.8%)
 Total cholesterol (mmol L−1)
  Age 60 + 755.58 ± 1.066.04 ± 1.32***5.88 ± 1.006.32 ± 1.06***
  Age 40 + 455.57 ± 1.045.68 ± 1.025.25 ± 1.115.27 ± 0.91
 HDL cholesterol (mmol L−1)
  Age 60 + 751.39 ± 0.361.44 ± 0.401.68 ± 0.421.75 ± 0.47*
  Age 40 + 451.17 ± 0.281.29 ± 0.34*1.45 ± 0.291.60 ± 0.39*

Determinants of use of cholesterol-lowering drugs

Amongst subjects who reported a history of CVD or diabetes, users of statins were more likely to have a family history of CHD compared with nonusers, had a higher BMI, and were more likely to use antihypertensive drugs and acetylsalicylic acid than nonusers (Table 4). In multivariate logistic regression analyses, use of statins was associated with a positive family history, use of antihypertensive drugs and acetylsalicylic acid in men and women and with a visit to the general practitioner during the past year in women (Table 5). Men aged 60 years were more likely to take statins than younger men.

Table 4.  Medical and lifestyle characteristics of users and nonusers of statins amongst subjects with cardiovascular disease or diabetes and presumed healthy subjects in the Oslo Health Study 2000–2001
 MenWomen
UsersNonusersUsersNonusers
  1. CVD, cardiovascular disease; BMI, body mass index.

  2. *P-value < 0.05; **P-value < 0.01; ***P value < 0.001.

Family history of heart attack or angina, N (%)
 CVD or diabetes231 (58.3)170 (38.2)***154 (64.7)206 (50.4)***
 Healthy134 (46.0)1758 (35.5)***204 (57.3)2644 (41.4)***
No visit to the general practitioner in the past year, N (%)
 CVD or diabetes47 (12.3)55 (12.7)6 (2.6)37 (9.2)**
 Healthy51 (17.8)1891 (39.2)***27 (7.8)1514 (24.3)***
Current cigarette smoker (%)
 CVD or diabetes69 (17.6)100 (22.3)39 (16.7)94 (22.5)
 Healthy80 (27.4)1342 (27.1)73 (20.4)1754 (27.6)**
Use of antihypertensive drugs, N (%)
 CVD or diabetes245 (65.7)159 (35.4)***154 (69.7)143 (34.1)***
 Healthy162 (59.3)394 (7.9)***175 (51.2)542 (8.4)***
Use of acetylsalicylic acid, N (%)
 CVD or diabetes247 (61.9)125 (27.6)***149 (62.3)116 (27.6)***
 Healthy48 (16.3)196 (3.9)***52 (14.4)221 (3.4)***
BMI ≥ 25 kg m−2, N (%)
 CVD or diabetes272 (76.2)290 (67.6)**158 (70.2)245 (62.0)*
 Healthy216 (78.8)3080 (64.1)***224 (64.7)2896 (46.2)***
Education ≥ 13 years, N (%)
 CVD or diabetes176 (46.2)192 (44.5)49 (21.7)109 (26.8)
 Healthy146 (50.7)2890 (59.1)**96 (27.4)3438 (54.4)***
Table 5.  Multivariable adjusted odds ratios (OR) and 95% confidence intervals (CI) for the use of statins in subjects with cardiovascular disease or diabetes (N = 618 men and 484 women with complete data for all of the variables)
 MenWomen
OR (95% CI)βP-valueOR (95% CI)βP-value
Age (years)
 40 and 451.0 (referent)  1.0 (referent)  
 602.1 (1.0–4.2)0.74<0.052.4 (0.9–6.3)0.87ns
 751.3 (0.7–2.7)0.29ns2.4 (1.0–5.8)0.88ns
Family history
 Absent1.0 (referent)  1.0 (referent)  
 Present1.7 (1.2–2.5)0.53<0.011.6 (1.0–2.6)0.47<0.05
Visit to general practitioner in last year
 None1.0 (referent)  1.0 (referent)  
 At least one0.9 (0.5–1.5)−0.15ns4.5 (1.4–14.6)1.50<0.05
Cigarette smoking
 Past or never1.0 (referent)  1.0 (referent)  
 Current0.8 (0.5–1.3)−0.22ns0.6 (0.3–1.0)−0.56ns
Use of antihypertensive drugs
 No1.0 (referent)  1.0 (referent)  
 Yes3.0 (2.1–4.4)1.01<0.0013.1 (2.0–4.8)1.12<0.001
Use of acetylsalicylic acid
 No1.0 (referent)  1.0 (referent)  
 Yes4.5 (3.1–6.5)1.50<0.0014.7 (3.0–7.4)1.56<0.001
Body mass index (BMI, kg m−2)
 <251.0 (referent)  1.0 (referent)  
 ≥251.4 (0.9–2.1)0.31ns1.3 (0.8–2.2)0.29ns
Educational level (years)
 <131.0 (referent)  1.0 (referent)  
 ≥131.1 (0.8–1.6)0.11ns1.1 (0.6–1.8)0.08ns
HDL cholesterol (mmol L−1)
 High1.0 (referent)  1.0 (referent)  
 Low1.3 (0.8–2.0)0.23ns1.3 (0.8–2.2)0.28ns

Amongst healthy subjects, use of statins was associated with the family history, a recent visit to the general practitioner, a higher BMI, a lower educational level, use of antihypertensive drugs or acetylsalicylic acid in men and women and with nonsmoking in women (Table 4). In multivariate analyses, the family history, a recent visit to the general practitioner and use of antihypertensive drugs or acetylsalicylic acid were independent correlates of use of statins in men and women, whilst increased age and a low educational level increased the likelihood of using statins in women only. Amongst men, use of statins was again most likely at the age of 60 years compared with younger men (Table 6).

Table 6.  Multivariable adjusted odds ratios (OR) and 95% confidence intervals (CI) for the use of cholesterol-lowering drugs in presumed healthy subjects (N = 4346 men and 5783 women with data for all of the variables)
 MenWomen
OR (95% CI)βP-valueOR (95% CI)βP-value
Age (years)
 40 and 451.0 (referent)  1.0 (referent)  
 601.6 (1.1–2.3)0.46<0.053.5 (2.3–5.2)1.25<0.001
 750.7 (0.4–1.1)−0.35ns4.2 (2.8–6.4)1.44<0.001
Family history
 Absent1.0 (referent)  1.0 (referent)  
 Present1.4 (1.0–1.9)0.34<0.051.7 (1.3–2.2)0.50<0.001
Visit to the general practitioner in the last year
 None1.0 (referent)  1.0 (referent)  
 At least one1.5 (1.0–2.2)0.41<0.052.9 (1.7–4.8)1.06<0.001
Cigarette smoking
 Past or never1.0 (referent)  1.0 (referent)  
 Current1.1 (0.8–1.6)0.10ns0.9 (0.7–1.3)−0.05ns
Use of antihypertensive drugs
 No1.0 (referent)  1.0 (referent)  
 Yes14.3 (10.2–20.1)2.66<0.0014.9 (3.7–6.5)1.60<0.001
Use of acetylsalicylic acid
 No1.0 (referent)  1.0 (referent)  
 Yes3.3 (2.1–5.3)1.21<0.0012.3 (1.5–3.4)0.83<0.001
Body mass index (BMI, kg m−2
 <251.0 (referent)  1.0 (referent)  
 ≥251.5 (1.0–2.1)0.38<0.051.3 (1.0–1.7)0.28ns
Educational level (years)
 <131.0 (referent)  1.0 (referent)  
 ≥130.9 (0.7–1.2)−0.10ns0.6 (0.5–0.8)−0.52<0.001
HDL cholesterol (mmol/l)
 High1.0 (referent)  1.0 (referent)  
 Low1.2 (0.8–1.8)0.17ns0.9 (0.7–1.3)−0.08ns

Discussion

Our findings indicate that a substantial proportion of Oslo residents who report CVD or diabetes are not receiving statins. Moreover, serum total cholesterol levels were on the average ≥5.5 mmol L−1 in subjects with these diseases that were not receiving drugs. Women were less likely than men to be taking drugs. However, in primary prevention, women were almost as likely as men to be taking drugs and the proportion of women taking drugs tended to exceed that of men in the oldest age group of 75 years, despite women's lower CHD risk. Use of statins was associated with use of antihypertensive drugs and acetylsalicylic acid, a family history of CHD, and with a visit within 1 year to the general practitioner. Only a small minority of subjects (<3% overall) had taken statins previously but had stopped. Notably, over one-fifth of presumed healthy men that were not receiving statins had a 10-year Framingham risk score of >20%.

These data were obtained by self-administered questionnaire and included a population-based sample, in contrast to a number of earlier assessments that have been based on data from selected physician's practices [16], tertiary care centres or specialist providers [7, 17] or prescription databases [8, 9, 12, 18]. Close to 100% of those who attended our survey answered the questions on cholesterol-lowering drugs and 72% of these reported the trade name of a statin [19]. These findings are in accordance with results from other studies that show an accurate recall of medical and drug usage history in well-defined chronic conditions [20, 21]. If subjects who used statins forgot to report their use, we may have underestimated the proportions using these medications. Although just over one-half of the invited population responded, analysis of nonattendees found that although they tended to be less educated than attendees, prevalence estimates of BMI, smoking and other health indicators were robust (A. J. Søgaard, R. Selmer, E. Bjertness, D. Thelle, personal communication). Moreover, educational level only tended to be related to a lower use of statins in primary prevention in the present study, and not at all in secondary prevention. Thus, this may mean that statin use in healthy individuals was underestimated, however, this should not affect the observed differences between men and women.

The proportion taking statins in our study was lower than that reported in a neighbouring county to Oslo (Hedmark). In this county nearly one-fifth of 60 and 75-year old men and women were taking statins; however, these data were not analysed according to the presence of CVD or diabetes [22].

A number of studies have demonstrated that treatment with statins amongst patients with CHD may not be equitable. Whilst some studies have shown that the elderly are less likely to receive statins [11, 12, 18], this was not the case in the current study with the exception of healthy men aged 75 years. However, women with CVD or diabetes were less likely to be treated than men, possibly because practitioners took into consideration the more favourable HDL cholesterol levels of women. In data from general practices across England and Wales, women were not less likely to receive statins when the severity of their disease was considered [12], however, we found that even women with a history of heart attack seemed less likely to receive statins than men. The landmark Heart Protection Study demonstrated benefits of simvastatin therapy for high-risk individuals with total cholesterol levels ≥3.5 mmol L−1 but was published after the end of our survey [5]. However, Norwegian/Swedish guidelines issued in 1999 have emphasized treatment of all subjects with CHD, other atherosclerotic disease or diabetes and whose total cholesterol levels are >5.0 mmol L−1[1]. A substantial number of untreated people with these diseases in the current survey had higher levels than this cut-off point.

In primary prevention a number of studies have shown that physicians may overuse statins for groups with high-cholesterol levels, for example, older women [8, 9], whilst neglecting people with high total coronary risk and multiple CHD risk factors, including cigarette smoking and hypertension [10, 11]. This also seems to be the case in Oslo. Women aged 75 years had the highest prevalence of statin use amongst healthy subjects, despite the paucity of clinical trial data to support treatment of women aged 75 years with no history of CVD or diabetes [5, 23, 24]. Similarly, at age 60 nearly equal proportions of women and men were taking statins, despite the lower absolute CHD risk of women [1, 2].

Because we did not have data regarding the lipid profiles of our subjects prior to the start of statin therapy, we were not able to estimate the 10-year Framingham risk score of healthy people who were given statins, thus, we were not able to enter the score in the logistic regression model. Statins influence levels of total and HDL cholesterol, both of which are a part of the risk score. However, the association we found between use of antihypertensive drugs and statins indicates that subjects with hypertension were appropriately considered to be high risk; moreover, hypertension may be associated with hypercholesterolaemia and with components of the metabolic syndrome, including a low HDL cholesterol level [25]. Notably, HDL cholesterol levels were lower in subjects aged 40, 45 or 60 years who were taking statins than their counterparts, but HDL cholesterol level was not a significant contributor to the multivariate model when each gender is considered separately. A family history of CHD was associated with the use of statins in all groups; taking into account the family history is recommended by guidelines for primary but not for secondary prevention [1]. Cigarette smokers tended to be less likely to be taking statins than other presumed healthy subjects, although not significantly. Thus, the presence of this important CHD risk factor did not increase the likelihood of treatment, despite the increased risk of smokers.

Not surprisingly in both primary and secondary prevention, subjects were more likely to take statins if they had visited their general practitioner at least once during the previous year. Individuals with a distant history of heart attack or with mild angina may not be aware of benefits of cholesterol-lowering drugs unless they have recently visited their general practitioner. This notion is supported by findings from general practices indicating that a more remote history of ischaemic heart disease was associated with a lower use of statins [12]. The lower rate of statin use for healthy women with a high educational level may reflect a better lifestyle amongst these compared with other women, a reluctance to take drugs, or other factors.

In Great Britain, application of the conservative Sheffield guidelines for statin treatment is estimated to result in 8.2% of the population from 35 to 69 years old receiving treatment [9]; we are not aware of similar calculations published for the Norwegian population. Overall, only 5.7% of our presumed healthy sample aged 40, 45 or 60 years were taking statins, indicating a substantial potential for treatment, in particular in regard to the more liberal European and Scandinavian guidelines that recommend treatment for individuals whose CHD risk is over 20% over 10 years [1, 2]. Amongst people that were not taking statins aged 60 years in the Oslo Health Study, over one-fifth of men had a 10-year Framingham score of over 20% and should be considered for statin use according to guidelines.

The applicability of Framingham risk equation in order to estimate risk in a European population has been questioned. Recently, the SCORE project published a risk scoring system for use in European practice [26]. It has been shown the Framingham risk scores overestimate risk in populations with low CHD rates. In contrast, close agreement has been shown between the Framingham-based risk function and Northern European rates of CHD [27].

In Norway, the mean daily dose of simvastatin and atorvastatin (the most used statins in this survey) are 23 and 18 mg, respectively. These doses would be expected to give an approx. 25% reduction in serum total cholesterol. We expected cholesterol levels to be lower than those that were observed in treated subjects in our survey compared with nonusers of statins. Moreover, with the exception of men with CVD or diabetes, the majority of treated subjects did not reach target total and LDL cholesterol levels. The most frequently used statins were simvastatin and atorvastatin, which are also the most effective drugs in achieving target total and LDL cholesterol levels. Although percentage reductions in cholesterol levels in unselected patients in primary care have been noted to be similar to those in randomized-controlled trials initial levels may be higher than those of subjects in trials, making target levels difficult to reach [28]. In accordance with the excellent tolerability record of statin drugs, only very small proportions of people reported that they had taken a statin previously, but not currently. Thus, adherence to therapy seems to be a lesser problem in our population with nearly full reimbursement for medical care and drug costs than in other reports [29]. Likewise, Danish patients have shown good adherence to statin therapy [30].

In conclusion, patients requiring secondary prevention tended to be undertreated with statins. Use of statins for primary prevention was appropriately associated with antihypertensive and acetylsalicylic acid use and a family history of CHD. However, the proportion of men taking statins was low compared with the proportion of women when 10-year absolute CHD risk (Framingham score) differences are considered. Only a small proportion of apparently eligible men aged 60 years without CVD or diabetes were taking statins, whilst apparently healthy elderly women were most likely to receive statins. The focus on total CHD risk rather than cholesterol levels recommended by guidelines does not seem to have impacted physicians’ practice [1, 2].

Conflict of interest statement

No conflict of interest was declared.

Acknowledgement

The data collection was conducted as part of the Oslo Health Study 2000–2001 in collaboration with the National Health Screening Service of Norway – now the Norwegian Institute of Public Health.

Ancillary