The Centers for AIDS Research Network of Integrated Clinic Systems (CNICS) is a prospective observational cohort study of adult PWH in clinical care at academic institutions across the United States. The study cohort included all PWH in care, defined as those who attended 1 or more in-person or virtual HIV primary care visits between September 1, 2018, and December 31, 2020, at 7 CNICS sites: Johns Hopkins University, Case Western Reserve University, Fenway Health, University of Alabama at Birmingham, University of California-San Diego, University of North Carolina at Chapel Hill, and University of Washington. CNICS research has been approved by the institutional review boards at each site.
Methods of data collection in CNICS were as previously reported. In brief, comprehensive clinical data including diagnoses, laboratory test results, and medications collected through electronic medical records and institutional data systems undergo rigorous quality assessment and are harmonized in a central repository that is updated quarterly. Demographic data, including birth sex, patient-reported racial/ethnic identity, and risk factor of HIV acquisition, were collected at cohort enrollment. Patient-reported outcome measures of smoking (current vs. former vs. never use of tobacco) and unstable housing/homelessness were collected through tablet-based surveys every 4–6 months during primary care visits[22,23] and from medical records.
SARS-CoV-2 Infections and Outcomes
Candidate SARS-CoV-2 infections and COVID-19 cases were identified through laboratory test results and provider-documented diagnoses (ICD-10 codes) recorded between March 1, 2020, and December 31, 2020, and verified through a medical record review. SARS-CoV-2 testing results and hospitalization records electronically imported to CNICS sites from external health systems were reviewed for data completeness. Medical records for all cases identified by diagnosis code without supporting laboratory tests performed within the CNICS system, including externally reported diagnoses and test results, were reviewed by clinicians at each site using a standardized protocol.
Disease severity indicated by hospitalization for COVID-19, requirement for supplemental oxygen, intensive care admission, and invasive mechanical ventilation were verified by site clinicians and central clinician review of hospitalization discharge summaries. A patient with a positive SARS-CoV-2 test obtained while hospitalized for reasons unrelated to COVID-19 (eg, preoperative/admission screening or other incidental diagnosis) was considered a verified case but not as hospitalized due to COVID-19. Deaths occurring within 30 days of COVID-19 diagnosis or discharge from a hospitalization due to COVID-19 were attributed to COVID-19.
We examined the following chronic comorbid conditions: diabetes defined using a previously validated approach as hemoglobin A1c (HbA1c) ≥6.5%, a prescription of a diabetes-specific medication, or a diagnosis of diabetes with associated prescription; treated hypertension as a diagnosis of hypertension and prescription of an antihypertensive medication; obesity as body mass index (BMI) ≥30 kg/m2; hepatitis C virus (HCV) coinfection by the presence of positive HCV antibody or detectable RNA or genotype; and chronic obstructive pulmonary disease (COPD) using a previously validated approach as a diagnosis of COPD and ≥90-day continuous supply of long-acting controller medications. We used laboratory test data to compute clinical measures of chronic kidney disease (CKD) defined as last estimated glomerular filtration rate (eGFR) <60 using CKD-EPI without race adjustment[26,27] and risk scores for atherosclerotic cardiovascular disease (ASCVD) and hepatic fibrosis (fibrosis-4: FIB-4[29–31]). We examined CD4 counts (cells/mm3) as both lowest historical value, as a proxy for CD4 nadir, and current CD4 count, as well as HIV VL (copies/mL). All laboratory test results were censored 1 week before COVID-19 diagnosis to avoid confounding.
Relative risks for COVID-related hospitalization were calculated using relative risk regression. We accounted for potential confounding by adjusting models with disease risk scores (DRSs),[33,34] a prognostic analog of propensity scores useful when studying a limited number of exposed patients and outcomes and a relatively large number of confounders. DRSs were constructed independently for each exposure of interest using logistic regression in the full cohort for all nonduplicative covariates (eg, FIB-4 not adjusted for age because age is a component of FIB-4 calculation) including age, birth sex, race/ethnicity, smoking status, diabetes, hypertension, and CNICS site. We conducted sensitivity analyses adding comorbid conditions to the DRS to further evaluate the specific effects of race/ethnicity and age. All analyses were conducted in Stata version 17 (StataCorp, College Station, TX).
J Acquir Immune Defic Syndr. 2022;90(4):369-376. © 2022 Lippincott Williams & Wilkins