Table 1 Biases that can occur in epidemiological studies of kidney disease populations as a result of barriers to health care
From: COVID-19 and kidney disease: insights from epidemiology to inform clinical practice
Type of bias | Barriers to accessing health care pre-pandemic | Barriers to SARS-CoV-2 testing | Barriers to accessing health care during the COVID-19 pandemic |
|---|---|---|---|
Information bias (misclassification of the study exposure or outcome) | Misclassification of CKD as non-CKD (absence of testing or coding for CKD) Misclassification of subclinical CKD as AKI and vice versa | Misclassification of infection as non-infection | Misclassification of cause of hospital admission |
Confounding (when a measured or unmeasured variable, not on the causal pathway, influences both the study exposure and the study outcome) | Reasons for getting a particular type of care will confound results Applies to all kidney disease populations | Reasons for getting a test will confound results Applies to all kidney disease populations | Reasons for getting a particular type of care will confound results Applies to all kidney disease populations |
Collider bias (when the study sample collected is conditional on a variable on the causal pathway between study exposure and study outcome) | When a retrospective health record study only includes those with baseline and repeat creatinine test results (in a conventional cohort study people with missing test results would be logged as ‘lost to follow-up’ and appropriately censored/analysed) | Access to SARS-CoV-2 testing may depend on COVID-19 severity and varies between kidney disease populations; comparisons of outcomes amongst those tested may be biased | Access to hospital/ICU care depends on COVID-19 severity and underlying chances of survival/comorbidity, which varies between kidney disease populations; therefore, analyses of hospitalized/ICU patient populations may not always provide information on pathobiology |
Selection bias (for a cohort study, selection bias is a systematic bias in the ability to capture study outcomes dependent on the (unobserved) study outcome; for a case–control study, selection bias occurs when cases are drawn from a different source population than the controls) | Not applicable for studies where COVID-19 is an outcome For studies that consider kidney outcomes after COVID-19, barriers to kidney care/dialysis/transplantation can introduce selection bias | Especially for transplant, home dialysis, CKD and AKI populations where there was no systematic screening for COVID-19 | Applies to all kidney disease populations Some populations may have had fewer barriers than others (e.g. in-centre haemodialysis or transplant recipients) Some populations potentially may have had less access to ICU care |
