We conducted a retrospective cohort study. All consecutive Cardiology outpatients visited from May 2014 to July 2016 in a primary healthcare network that gives support to a 250,000-population area were included. We selected LDL-C from 2013 as baseline, and LDL-C from 2016 and 2018, the period in which the aforementioned landmark randomized clinical trials were published. We included all patients who had a MI before the first LDL-C measurement in 2013 and who had all 3 LDL-C measurements available from clinical records. For patients who were dead at time of data collection no LDL-C could be obtained due to technical difficulties, so they were excluded from the analysis. No restrictions existed in this area during the study period for the prescription of high-intensity statin or ezetimibe, and proprotein convertase subtilisin/kexin 9 antagonists were available for those patients with LDL-C ≥ 100mg/dL under maximum tolerated LDL-C lowering treatment. Data collection was authorized for our local ethics committee (Hospital Universitari Vall d’Hebron).

Demographic and clinical characteristics at the time of the index visit were collected retrospectively from clinical records, including age, sex, weight, height, hypertension, dyslipidemia, diabetes mellitus, peripheral artery disease, cerebrovascular disease, age at first MI, years after first MI, atrial fibrillation and left ventricle ejection fraction. LDL-C values according to Friedewald formula were obtained from blood tests performed in 2013, 2016 and 2018. If no LDL-C was available from the selected year, we selected a blood test from 6 months earlier or after, but always at least 3 months after the first MI to ensure proper secondary prevention measures were established.

Results are presented as absolute and relative frequencies for categorical variables and as mean and standard deviation or, when appropriate according to Saphiro–Wilk test, as median and interquartile range for continuous variables. Depending of normality, Student t, U-Mann Whitney or Wilcoxon tests (when applying for matched variables such as LDL-C in 2013 and 2018) were performed when comparing continuous variables, and chi-square with Bonferroni-like adjustment was performed for the comparison of categorical variables. Multivariate analysis was performed with multiple linear regression for quantitative dependent variables and binary logistic regression for dichotomous variables. All dependent variables that showed a P<.1 in univariate comparisons in any of the 3 years assessed were included in the multivariate analysis. Statistical significance was established as a two-tail P<.05. All statistical analysis was performed with Stata 15 (StataCorp. 2017. Stata Statistical Software: Release 15. College Station, TX: StataCorp LLC).

Our results show that LDL-C control in a very high-risk outpatient population was poor in 2013 and is still low in 2018. There is still a little achievement of recommended LDL-C target, with only a quarter and a third of patients with an LDL-C < 70mg/dL in 2013 and 2018, respectively, and only 8% and 11.7% of patients with an LDL-C < 55mg/dL in the same years. Patients with an LDL-C ≥ 100mg/dL, a cut off where an important increase in MACE has been observed and where combined lipid-lowering therapy may be more useful,1 still remained around 20.5% in 2018.

These data are in concordance with multicentric registries.13,14,16,18,20 EUROASPIRE V, an international registry conducted among 27 European countries in 2016–2017 and including patients with coronary artery disease, showed that only 29% of them had an LDL-C < 70mg/dL. Among the very high-risk patients from the DYSIS I study, only 21.5% had an LDL-C < 70mg/dL. DYSIS II registry also carried out among several countries in Europe, Asia, and the Middle East between 2012 and 2014, showed that among patients with stable coronary artery disease only 28.3% of patients had an LDL-C < 70mg/dL. In the REPAR Spanish registry, which included patients with very high cardiovascular risk from 2013 to 2014, 26.4% of the cohort had an LDL-C < 70mg/dL. A recent follow-up study including patients with coronary artery bypass graft surgery in Israel showed that 44% of patients had an LDL < 70mg/dL, although the year of data collection is not described.23 In almost all of the registries mentioned, a vast majority of patients (80–90%) are reported to be receiving some dose of statin. Given that impact on cardiovascular outcomes arises from LDL-C reduction independently of the treatment used, our aim was to assess individual trends in LDL-C variation and goal attainments in LDL-C control irrespective of the drugs used to achieve it, so we did not collect data on the treatment used.

When looking at the temporal trend, a better LDL-C control was achieved along the years, with a median 6mg/dL decrease from 2013 to 2018, a 9.2% absolute increase in the percentage of patients with an LDL-C < 70mg/dL. To our knowledge, this is the first register reporting longitudinal LDL-C variation from the same cohort in the recent years, since many of the published data comes from studies reporting transversal LDL-C levels in a selected time period.

The increase in the percentage of patients with LDL-C < 70mg/dL and LDL-C < 55mg/dL in 2018 is probably related to the release of the aforementioned clinical trials and the spreading notion of “the lower the better” when targeting LDL-C. Although we have not specifically assessed which treatment was used in every patient, there was no restriction in lipid-lowering drugs prescription in our area. Possibly, this improvement reflects the increased risk perception physicians have regarding their patients and their efforts to lower LDL-C by any means, not only by prescribing non-statin drugs but also by intensifying statin treatment.

When assessing maintenance of LDL-C at the desired target, it is remarkable that only 10.0% and 1.38% of patients had all LDL-C measurements < 70mg/dL and < 55mg/dL, respectively. In a similar way, 47.4% and 78.8% had all 3 LDL-C measurements ≥ 70 and 55mg/dL, respectively, male sex and diabetes being significant predictors for this endpoint after adjusting for covariates with a strong odds ratio. This reflects that there is a high prevalence of patients who do not reach LDL-C target at any time and may be exposed to greater risk.

In all pivotal trials assessing lipid-lowering drugs in secondary prevention,7–9,11,12 no subgroup effect depending on sex or baseline risk factor has been identified in LDL-C reduction, except for a greater reduction in LDL-C in diabetic patients with ezetimibe.10 Previous registries have shown that LDL-C lowering therapy is intensified and better LDL-C control is achieved in groups that are probably judged by the treating physician as being at higher risk for MACE,18,22,24 even when no difference between groups in LDL-C goal is made by ESC and ACC/AHA guidelines when cardiovascular risk is so high.4,5 This may explain the fact that male sex, diabetes or hypertension showed in our study an association with better LDL-C control, since the treating physician may consider these subgroups as being at higher risk for recurrent MACE.

Despite having been an increase in LDL-C target attainment from 2013 to 2018, male sex and diabetes remained predictors for adequate LDL-C control in 2018, that may reflect the fact that risk perception of these specific groups have remained unchanged from 2013 to 2018 ?physicians may consider male or diabetic patients as being exposed to higher risk for recurrent MACE both in 2013 and in 2018.

No subgroup has obtained a specific benefit in terms of LDL-C levels along the assessed years, given that improvement has been due to better LDL-C levels in all patients in the cohort. These data are in concordance with the hypothesis that recent scientific literature may have boosted a change in risk perception globally among treating physicians that has affected equally all subgroups of patients.

It is well known that male sex is a risk factor for developing coronary heart disease, but once it is established, no difference in risk for recurrent events should be made based on sex. This fact may wrongly influence the propensity of the treating physician when adjusting lipid-lowering drugs and based on patient sex,25 which may lead to different outcomes in the short26–28 and in the long term,29–32 although little is known about the correlation of proper LDL-C control, drug prescription and recurrent events in this population. We consider that the present data should warn us against instinctive, non-evidenced based risk perception and should prompt further research to clarify the reasons for a worse LDL-C control in women and non-diabetics. Nonetheless, a systematic review points to a better statin adherence in primary prevention in male patients and diabetes, a fact that may also explain our findings.33

We present the first registry to show data regarding the new target set by 2019 ESC guidelines. Although it was not the aim of the treating physician to lower LDL-C levels below 55 mg/dL, only a minority of patients reached an LDL-C < 55mg/dL despite being at very high risk. We have no data on drug use, but regardless of the medications used on our patients, it is certain that a very important treatment intensification will be necessary to reach the new target, so new studies considering drug prescription and cost-effectiveness should be conducted to assess the feasibility of lowering LDL-C < 55mg/dL.

This study has several limitations. Because of a retrospective design, we could not control all inherent biases. Patients with higher LDL-C may have a higher mortality rate and we may be overestimating the impact of LDL-C reduction. We excluded 204 patients because there was no LDL-C available in at least one of the selected years. It could be considered that patients with 3 LDL-C measurements available are probably better controlled than patients with no LDL-C measured in any of the 3 selected years. Nevertheless, LDL-C attainment is still poor in this cohort with theoretically better LDL-C control.

All patients included in our study came from a single primary healthcare network in Spain, and results may not be inferred to other European countries, given regional differences seen in multinational registries as EUROASPIRE IV and V.13,20 Also, since we have only selected patients diagnosed with MI at least 5 years before 2018 and with out-of-hospital follow-up, our data may not be extrapolated to patients with recent MI or ischemic heart disease without MI.

When assessing clinical predictors we did not include in our analysis chronic kidney disease or current smoking, both recently associated with better LDL-C control,18 but seems unlikely that its inclusion would have modified our results when, in the same study, both male sex and diabetes remained independent predictors for an LDL-C below target in the single year assessed when multivariate analysis was performed.

Also, given that we have not collected clinical outcomes, one might think that patients considered as being at higher risk, such as male or diabetic patients, may have had more MI during follow-up and lipid-lowering therapy may have been adjusted after the new MI. However, since there were no differences in the prevalence of LDL-C below target in 2013 between patients with a MI in 2013 and patients with a MI before 2013, we consider that the effect of a new MI in LDL-C reduction in patients with already known coronary heart disease is probably not significant. Instead, we believe that better LDL-C control is attained in this population because of a different perception in risk by the treating physician.