Validity of self-reported high cholesterol in the city of Sao Paulo, Brazil, and factors associated with this information's sensitivity

Abstract

The study aimed to validate self-report of high cholesterol in Sao Paulo, Brazil, and verify factors associated with this information's sensitivity. Data were used from the Health Survey of the City of Sao Paulo 2015, a cross-sectional population-based study with a probabilistic sample of the city's residents. The sample included 886 individuals with information from the structured questionnaire, blood measurements of total cholesterol and fractions or who reported being on medication for high cholesterol. The validity of self-reported information on high cholesterol was measured according to sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and kappa index, considering total cholesterol and LDL-c and pharmacological treatment as the gold standard. Logistic regression models were developed to investigate factors associated with the sensitivity of this information. The sensitivity of information on high cholesterol using total cholesterol as the reference was 50.6%, specificity 90.19%, PPV 51.64%, NPV 89.82%, and kappa 0.41. Taking LDL-c as the gold standard, sensitivity was 53.52%, specificity 89.93%, PPV 49.22%, NPV 91.39%, and kappa 0.43. Using total cholesterol as the reference, age (OR = 1.69; 95% CI: 1.24-2.29) and having a private health plan (OR = 2.91; 95% CI: 1.06-7.99) were associated with the information's sensitivity. With LDL-c as the gold standard, age (OR = 1.71; 95% CI: 1.132.57), smoking (OR = 3.33; 95% CI: 1.08-10.27), and having a private health plan (OR = 3.64; 95% CI: 1.10-12.08) were associated with the information's sensitivity. The results suggest low sensitivity and low PPV of self-reported high cholesterol in residents of Sao Paulo.