The WIHS design has been described previously. 7
Briefly, WIHS is an ongoing prospective cohort study that enrolled HIV-infected and at-risk seronegative women in five metropolitan areas in the United States. Only HIV-infected women enrolled between October 1994 and November 1995 were included in the current analysis. Demographics and social, medical, and detailed HIV disease history were collected by standardized interview. Prior AIDS-defining illness (ADI) was defined based on the 1993 Centers for Disease Control case definition for AIDS, excluding CD4 <200 cells/mm3
. Laboratory data including CD4, HIV-RNA, hepatitis serologies, hemoglobin, serum albumin, and routine urinalysis were collected at enrollment and semi-annually. Serum creatinine testing was performed at enrollment and annually. Glomerular filtration rate (eGFR) was estimated using the 4-variable Modification of Diet in Renal Disease (MDRD) equation. Because currently available GFR estimates have not been validated in HIV-infected populations, we selected the abbreviated MDRD equation because of the established relationship between decreased MDRD eGFR and increased mortality in other patient populations. 8, 9
The WIHS was approved by the institutional review boards of all sites, and participants provided written informed consent. The current analysis was also approved by the institutional review board of the Mount Sinai School of Medicine.
Urine specimens were collected at study entry and every 6 months. To minimize artifact from cystitis, specimens were excluded from this analysis if the corresponding urinalysis demonstrated pyuria (leukocyte esterase 1+ with a positive or missing urine culture, or leukocyte esterase >1+). By WIHS protocol, urine specimens were not collected during menses. Women with two eligible urine specimens were included in this analysis. The index visit was defined as the first visit at which an eligible specimen was obtained. Specimens with protein <1+ were tested for microalbuminuria. Semi-quantitative microalbuminuria testing was performed in banked urine specimens under standardized conditions using Clinitek Microalbumin Reagent Strips and the Clinitek 50 Analyzer (Siemens Healthcare Diagnostics, Deerfield, IL). This point-of-care dipstick assay has a sensitivity of 86% compared to clinical laboratory testing, and was selected because of its applicability to clinical practice. Frozen specimens (−70C) from the Division of AIDS central repository were thawed completely before testing. Quality assurance testing was performed with standardized controls (Bio-Rad Laboratories, Hercules, CA). Microalbuminuria was defined as an albumin: creatinine ratio > 30 mg/g.
The study exposure was hierarchically defined as follows; 1) Women with urinalysis protein ≥1+ at both the index visit and the next consecutive eligible visit were considered to have “confirmed proteinuria.” 2) Women with microalbuminuria at both visits or with microalbuminuria at one visit and proteinuria at the other were considered to have “confirmed microalbuminuria.” 3) Women with proteinuria or microalbuminuria at only one visit were considered to have “unconfirmed albuminuria.” 4) Women without proteinuria or microalbuminuria on either visit were considered to have no albuminuria.
Data on vital status were collected from medical records, providers, and personal contacts of WIHS participants and from the National Death Index. In cases where the death certificate was not available, the cause of death was ascertained from medical records, providers, or personal contacts, in that order. The cause of death was categorized as AIDS, non-AIDS, or indeterminate, as previously described. 10
Clinical characteristics at the index visit were summarized for each albuminuria group. When data were missing at the index visit, the value was obtained from the most recent prior visit. Continuous and categorical variables were compared using Kruskall-Wallis and chi-square tests, respectively. All p
-values are 2-sided. Kaplan-Meier and proportional hazards models evaluated associations of albuminuria with time to death. Proportional hazards models included baseline characteristics that have previously been associated with short-term mortality in WIHS:1 11
demographics, prior ADI, hepatitis C co-infection, log transformed HIV-RNA, CD4, hemoglobin, and serum albumin. Diabetes, blood pressure, and eGFR were also considered for inclusion because of their known association with albuminuria. Survival analyses were left-truncated at the second eligible study visit, and were censored at 31 October 1997, based on a prior analysis of WIHS demonstrating minimal HAART use and little effect of HAART exposure on the relationship between other covariates and death prior to this date. 11
In addition, a sensitivity analysis including HAART use as a time dependent variable in the multivariate models was performed. All analyses were performed using SAS (version 9.1.3; SAS).