Results
This study included 472,373 participants in the final analyses. Table 1 showed the baseline characteristics of participants by regular PPI use. At baseline, 46 530 (9.85%) participants reported regular use of PPIs. PPI users tended to be deprived, consumed less alcohol, less physically active, with a higher rate of comorbidities (hyperlipidemia, diabetes, and CVD), and were more likely to use other medications (like aspirin, NSAID sandstatin). As expected, PPI users had higher rate of esophagitis/Barretts esophagus, GERD, peptic ulcer, and upper gastrointestinal bleeding. Regular PPI users also had a poorer self-reported overall health rating, and higher rate of longstanding illness.
During a median follow-up of 8.1 years, we documented 1661 incident CKD events among the 46,530 PPI users (4.53 per 1000 person-years), and 5630 events among 425,843 nonusers (1.65 per 1000 person-years). In the initial age and sex-stratified model, regular PPI users had 2.15 times higher risk of incident CKD as compared to non-users (HR 2.15, 95CI 2.03-2.27) (Table 2).The association was attenuated somewhat, but remained significant after adjustment for potential confounders, including sociodemographic factors, lifestyle habits, presence of comorbidities, and use of other medications (HR1.24, 95%CI 1.17 -1.31).The estimated risk was unchanged after further adjustment for clinical indications for PPI use (HR1.24, 95%CI 1.16-1.33).For easy interpretation, NNHs were calculated based on based on the fully adjusted HR and the CKD incidence among non-PPI users. Every 773.1 (95% CI, 690.3 to 946.5), 406.7 (95% CI, 357.8 to 506.5), and 177.5 (95% CI, 152.8 to 226.5) regular PPI users may result in one CKD case over one, two and five years, respectively (Figure 1 ). Regarding different types of PPIs, we found the increased risk of CKD was associated with Omeprazole and Lansoprazole (HR 1.24, 95%CI 1.14 – 1.34 and 1.26, 1.14 – 1.38, respectively), but not with other PPIs, mainly due to the relative low number of cases (Table 3).
We also found that a slightly stronger association between PPI use and acute kidney injury (AKI). In the fully adjusted model, regular PPI users had a 30% higher risk of AKI incident (HR 1.30, 95%CI 1.22 – 1.38) (Table S1 ). Given the similar clinical indications for PPIs and H2RAs, we also evaluated the association between regular use of H2RAs use and CKD risk. In the basic model, regular use of H2RAs was associated with a 69% greater risk of CKD (HR 1.69, 95%CI 1.49 – 1.91). However, this association disappeared after adjustment for potential confounders and clinical indications for H2RAs use (HR 1.03, 95%CI 0.91– 1.16) (Table S2 ). Direct comparison showed that PPI use was associated with 17% higher risk of CKD compared with H2RAs use (HR 1.17, 95% CI 1.00 to 1.36) (Table S3 ).
In subgroup analyses, the estimates for risk of CKD associated with PPI use did not differ by age, BMI, smoking status, alcohol consumption, hypertension, diabetes, GERD and clinical indication for PPI use (Figure 2 ). However, the estimated risk of CKD with PPI use seemed to be higher among women than men (p-interactions =0.003). Our primary results remained unchanged in several sensitivity analyses by excluding CKD incidents and death cases identified during the first 2 years of follow-up, excluding the participants with cardiovascular disease, and limiting our analysis to participants with no missing covariate data (Table S4 ). When we applied inverse treatment probability weights based on the propensity scores, the estimated HR for CKD was 1.40 (95 % CI 1.25–1.57). The results of propensity score–matching analysis were similar.