(1) The Risk of Acute Infection in Association with Newly Diagnosed Depression: A Cohort Study

Publication Number: 675

Background: Two cohort studies suggested an up to 60% increased risk of infections in patients with depression compared to patients without depression. However, confounding and surveillance bias may have played a role.

Objectives: To assess the risk of acute infections in patients with diagnosed depression compared to patients without diagnosed depression accounting for surveillance bias and confounding.

Methods: We conducted a cohort study using the UK primary care Clinical Practice Research Datalink (CPRD, 2000-2019). We identified patients with an incident diagnosis of depression (exposed cohort)). For each exposed patient, we identified one comparison patient of the same age and sex without a diagnosis of depression prior to the cohort entry date (CED) of the corresponding exposed patient. Comparison patients had to have ≥1GP visit within 14 days prior to the CED. We followed patients from day 1 after the CED for a maximum of 2 years or until censoring. The primary outcome was a composite of diagnosed acute infection (respiratory, gastrointestinal, urogenital, or septicemia). We used fine stratification based on propensity score (PS) to control for 32 baseline covariates including lifestyle factors, comorbidities, comedications, and health-care utilization. We estimated incidence rates (IR) and IR ratios (IRR) of acute infections comparing diagnosed depression to no depression using negative binomial regression.

Results: After PS-fine stratification, we included 285 922 patients with newly diagnosed depression and 285 921 comparison patients. We quantified a weighted IR of 97.32 infections/1000 person-years (py, 32 965 infections in 338 725.2 py) in patients with diagnosed depression, and IR 83.67 /1000 py (40 840 infections in 488 121.5 py) in patients without depression. After PS-matching, the risk of diagnosed acute infection in patients with depression (vs no depression) was slightly increased (IRR, 1.18; 95% CI, 1.16-1.20). Excluding patients with any of the recorded baseline comorbidities or comedications reduced the PS-weighted IRR to 1.07 (95% CI, 1.04-1.09). Not requiring ≥1GP visit within 14 days before CED for comparison patients resulted in a PS-weighted IRR of 1.54 (95% CI, 1.52-1.55).

Conclusions: After controlling for surveillance bias and confounding, we observed a less increased incidence of diagnosed acute infections in patients with diagnosed depression (vs no depression) than previously suggested. The slight relative risk increase further diminished after excluding patients with underlying comorbidities, suggesting residual confounding by disease severity.