Our study indicates that LDL-C control among Finnish hypertensive patients is insufficient, especially among younger patients. Without LLM, more than half of patients did not reach LDL-C target and even with medication, one third of patients did not meet the target. Furthermore, the proportion of individuals reaching LDL-C target seems to be lowest among working age patients who might benefit the most from CVD risk reduction over time [17, 18].
It is clear that younger patients have significantly lower total CVD risk than older patients when assessed using conventional short-term (generally 10-year) risk estimates. Due to current emphasis on short-term risk estimates, clinicians often choose not to initiate effective dyslipidemia treatment when short-term risk is low due to young age. It is remarkable, however, that all our study patients had at least one major CVD risk factor (treatment for hypertension), indicating that proper treatment of another major risk factor (hypercholesterolemia) would decrease the lifetime risk of CVD considerably [18].
Furthermore, it is challenging to rationalize why patients who are on LLM treatment are not treated to a relatively easy-to-reach LDL-C target of <3 mmol/l, regardless of age. With these individuals, the question is not “Should we treat cholesterol with drugs or not?” but rather: “Should we use the chosen medication properly or not?”. Poor medication adherence often forms a barrier for successful therapy, together with clinical inertia [3, 24, 25]. We argue, however, that lack of sufficient, individual physician feed-back and robust leadership engagement to overcome clinical inertia are also major, but modifiable reasons for this failure. Computerized decision support systems could offer one way to drive change for the better, but feedback alone is not sufficient for system-wide change [26, 27].
Strengths and limitations
This study has several strengths. To our knowledge, this is the first article to focus on age dependence in LDL-C control among hypertensive patients. Furthermore, Finland has robust public health care and majority of hypertensive patients are treated in public primary health care [28]. To conduct the study, we were able to rely on comprehensive public health care health records of a total population of over 155 000 individuals living in Central Finland [29]. Hence, the EHR database used in our study includes the majority of all hypertensive patients treated in this area.
Our study has also some limitations that are worth discussion. This was an observational cross-sectional study using routinely collected health care data together with laboratory data. These data sources have naturally several limitations. First, they do not provide sufficient information to assess total individual CVD risk. Therefore, we focused only in hypertensive population without CVD, diabetes or severe renal dysfunction. It is therefore reasonable to assume that LDL-C treatment targets of the general population are used in these patients [3, 9, 23]. Second, these data sources lack trustworthy information on smoking and current blood pressure status. This would be a major problem in prognostic study setting, but is not essential when studying the treatment status of an independent risk factor, such as LDL-C. Smoking and the blood pressure level of hypertensive patients do not change the LDL-C target levels, either. Third, the coverage and accuracy of diagnostic codes is never perfect, and may have resulted in misclassification of some patients, especially diabetics. However, the EHR data used in the study has also been a basis of a rigorous quality measurement system for several years and the accuracy of diagnostic codes has therefore enhanced remarkably. When analysing the data, we were also able to perform multiple chart reviews and found no signs of major misclassification. It is also possible that better target achievement in older age groups is partly due to more frequent use of health services due to increased multimorbidity, but we were not able to detect and compare our findings with the use of health services. Furthermore, it is probable, that some working age patients visit both primary and occupational health care and their LDL-C level may thus be treated to target in occupational health care after being first measured in primary health care. Finally, the status of LLM treatment in this study was based on up-to-date information of LLM prescription but our data does not provide reliable information on the adherence to medication or non-medical treatment of dyslipidemia.
Comparison with existing literature
Some previous studies have observed a similar age dependent trend in the proportion of individuals reaching LDL-C target [11-14]. It might seem reasonable to assume that physiological changes of aging could explain the better LDL-C control among older patients. However, it has been shown that both plasma LDL-C and total cholesterol levels increase progressively after age of 20 years [30, 31]. One explanation seems to be better LLM adherence among older patients [32]. However, the pattern of age-dependent increase in LDL-C is different between men and women and may thus, at least partly, explain the somewhat weaker association of age and LDL-C -levels among women [33].
Earlier, the gender differences in the proportion of men and women reaching LDL-C target have already raised a need to pay special attention to treatment of dyslipidemia in women [14]. Our results suggest that it is now time to pay more attention to younger dyslipidemia patients, as well. This is further emphasized by earlier research findings indicating that younger age is also associated with lower awareness and treatment rates of elevated blood pressure in Finland [34] . Lifetime risk estimates of CVD are developing rapidly and open access lifetime risk calculators are already available [17, 35]. At least one on-going randomised, controlled clinical study is currently investigating 10-year benefits of statin treatment in 35–59 year-old patients with LDL-C > 1.8 mmol/L and at least one risk factor other than dyslipidemia [16].