Data Analysis
We calculated person-years from the recruitment date to the date of first diagnosis of CKD, death, or the last date of follow-up, whichever happened first. We fitted Cox regression models with age as the timescale to estimate hazard ratios (HR) and 95% confidence intervals (CI) of PPI use on CKD incidence. First, we stratified the analyses by sex and age at baseline (in years, 37-55, 55-65, ≥65 years). We then fitted multivariable-adjusted models with additional adjustment for ethnicity, socioeconomic status, smoking status, alcohol consumption, physical activity, fruit and vegetable intake, red and process meat intake, BMI, sleep time, multivitamin use, mineral supplements intake, current medical status (systolic blood pressure,hyperlipidemia, diabetes, CVD, overall health rating, longstanding illness ), and concomitant medication use (including aspirin, non-aspirin NSAIDs, acetaminophen, ACEIs, ARBs, beta-blockers, calcium channel blockers, thiazide diuretics, statin, metformin )(model1). To address the possible confounding effect of clinical indications for PPI use, we additionally adjusted for esophagitis/Barretts esophagus, GERD, peptic ulcer, and upper gastrointestinal bleeding, H2RA use and anticoagulants/antiplatelets drugs use. Proportional hazards assumption was checked using Schoenfeld’s tests and no violation was shown. For covariates with selections of ‘do not know ’ and ‘prefer not to answer ’, or with missing data, we included an “unknown/missing” value indicator. Given that PPIs and H2RAs have similar clinical indications, we investigated the association between regular H2RA use and CKD risk and performed a head-to-head comparison between PPIs andH2RAs.To present possible association in a clinically translatable way, we calculated the number needed to harm (NNH) based on the method described by Altman D.G and Andersen P.K.23
To evaluate potential effect modifiers, we conducted additional stratified analyses according to sex, age, BMI, smoking status, alcohol consumption, hypertension, diabetes, GERD and clinical indication for PPI use. We performed a number of sensitivity analyses to check the robustness of the primary results. First, we excluded CKD incidents and death cases during the first two years of follow-up to minimize reverse causality. Second, we excluded participants with cardiovascular disease to minimize potential influence of the medical condition. Third, we restricted the analyses to participants with no missing data on any covariates. Lastly, we also performed propensity score matching and inverse treatment probability weights analyses to control for the influence of confounding variables. All analyses were conducted using the R software (version 3.5.0, R Foundation for Statistical Computing, Vienna, Austria).