Factors predicting the prescribing of statins for the primary preventing of cardiovascular disease: an historical cohort study

Talk Code: 
Samuel Finnikin
B H Willis, T Marshall
Author institutions: 
Institute of Applied Health Research, University of Birmingham


Estimating CVD risk is central to determining potential benefits from statins and communicating this to patients, but statin prescribing often does not accord with estimated CVD risk. This study investigates how calculation of CVD risk influences statin prescribing, and what influences statin prescribing if a CVD risk is not calculated.


A historical cohort study was undertaken using data from a large database of primary care records (THIN). Included patients were eligible for statins for primary prevention of CVD and had a lipid test between 2012 and 2016. The outcome was prescription of a statin within 60 days of the lipid result being recorded. Recorded QRISK2 scores (coded ten-year CVD risks) were extracted, along with variables used in calculating QRISK2 and others that might affect statin prescribing. If QRISK2 was not recorded, a post-hoc QRISK2 score was calculated. The cohort was described and the recorded or calculated QRISK2 score was used to describe statin prescribing in relation to NICE guideline eligibility. A mixed effects model (with each practice included as a random intercept) was used to establish the predictors of statin prescribing in in patients with and without a recorded QRISK2 score. Non-linear predictor effects were modelled using multiple fractional polynomials.


There were 686,560 entries into the cohort. 146,693 (21.4%) with a recorded QRISK2 score (the ‘recorded’ group). Statins were initiated in 6.6% of the recorded and 4.1% of the unrecorded groups respectively (p<0.001). Among patients in whom statins were indicated, more were prescribed statins if QRISK2 was recorded than unrecorded (19.2% v 6.2%, P<0.001). Among patients in whom QRISK2 was recorded, 85.0% of statin initiations met NICE eligibility criteria compared with 44.2% were QRISK2 was unrecorded (P<0.001). Among patients in whom QRISK2 was recorded, QRISK2 score was the most important predictor of statin initiation, but when QRISK2 was not recorded, total cholesterol was the main predictor of statin initiation, followed by diabetes.


Estimated CVD risk (QRISK2) is frequently not recorded prior to initiating statins in UK primary care. When QRISK2 is recorded, prescribing is more directed at patients with increased risk and hence those most likely to benefit. When QRISK2 is unrecorded, prescribing is mainly based on total cholesterol levels and therefore less associated with predicted benefit. When risk estimation is not used, patients will not have the required information to make treatment decision. Promoting the routine use of CVD risk estimation is essential to guide optimum statin prescribing and ensure patients have the necessary information to make shared decisions with their doctor. With increased focus on CVD prevention planned for early in the 2020’s, we suggest that focusing on accurate risk assessment is the key to effective patient-centred care.

Submitted by: 
Samuel Finnikin
Funding acknowledgement: 
SF was supported by an NIHR In-practice fellowship for part of this project