White Blood Count, Albumin, and BMI Enhance VACS Index Prognostic Model, but Nadir CD4 and CD8 Metrics Do Not

Abstract

Background People living with HIV frequently achieve long-term viral suppression necessitating better metrics of disease burden for clinical management and research. The Veterans Aging Cohort Study (VACS) Index predicts hospitalization, mortality, and other outcomes, using routinely available clinical data. We sought to enhance the index by evaluating whether nadir CD4, CD8, CD4/CD8 ratio, white blood count (WBC) or absolute neutrophil count (ANC), albumin, and body mass index (BMI) enhanced prediction. The original index categorized predictors for ease of understanding and calculation of a risk score. We also sought to expand categories and develop a continuous variable model, suitable for use with automated calculation, to provide higher resolution.Methods VACS, includes all HIV infected patients in VA Care. Among those who initiated ART 1996–2013, (excluding any treated for HCV infection), we obtained laboratory values from a randomly selected visit 2000–2014, at least one year after ART initiation. Patients were followed for 5-year, all cause mortality until September 30, 2016. We fit Cox models starting with currently used predictors (age, CD4, HIV-1 RNA, hemoglobin, FIB4, eGFR and HCV status) and decided to include new variables based on model fit, chi-square, strength and significance of individual levels and c-statistic. Functional form for continuous variables was determined graphically. Adequacy of final models was assessed with Kaplan-Meier plots by deciles of risk.ResultsAmong 28,390 patients there were 7,293 deaths (7.2 per 100 person-years) in median 4.1 years of follow-up. Nadir CD4, CD8, CD4:CD8 did not improve prediction. WBC and ANC performed equally but WBC was more widely available. C-statistics improved from 0.776 for the original VACS Index (in this sample) to 0.805.Conclusion Addition of WBC, albumin, and BM enhances utility of the VACS Index as a measure of overall severity of disease both as an outcome for research and for patient monitoring in the clinical setting. Validation in external cohorts is in progress.Disclosures All authors: No reported disclosures.