Response bias to a randomised controlled trial of a lifestyle intervention in people at high risk of cardiovascular disease: a cross-sectional analysis

Research evaluating lifestyle interventions for prevention of cardiovascular disease (CVD) may not reach those most at risk. We compared the response rate to a randomised controlled trial (RCT) of a lifestyle intervention by CVD risk, ethnicity and level of deprivation. Primary care patients with a QRisk2 score ≥ 20% were invited to participate in a RCT of an intensive lifestyle intervention versus usual care. This cross-sectional analysis compares anonymised data of responders and non-responders with multiple logistic regression, using adjusted odds ratios (AORs) for QRisk2 score, ethnicity, Index of Multiple Deprivation (IMD 2010) quintile, age and sex. From 60 general practices, 8902 patients were invited and 1489 responded. The mean age was 67.3 years and 21.0% were female. Of all patients invited, 69.9% were of white ethnic background, 13.9% ethnic minority backgrounds and 16.2% had no ethnicity data recorded in their medical records. Likelihood of response decreased as QRisk2 score increased (AOR 0.82 per 5 percentage points, 95% CI 0.77-0.88). Black African or Caribbean patients (AOR 0.67; 95% CI 0.45-0.98) and those with missing ethnicity data (AOR 0.55; 95% CI 0.46-0.66) were less likely to respond compared to participants of white ethnicity, but there was no difference in the response rates between south Asian and white ethnicity (AOR 1.08; 95% CI 0.84-1.38). Patients residing in the fourth (AOR 0.70; 95% CI 0.56-0.87) and fifth (AOR 0.52; 95% CI 0.40-0.68) most deprived IMD quintile were less likely to respond compared to the least deprived quintile. Evaluations of interventions intended for those at high risk of CVD may fail to reach those at highest risk. Hard to reach patient groups may require different recruitment strategies to maximise participation in future trials. Improvements in primary care ethnicity data recording is required to aid understanding of how successfully study samples represent the target population. ISRCTN, ISRCTN84864870. Registered 15 May 2012, https://doi.org/10.1186/ISRCTN84864870 .