Sign in Subscribe
Columbia University

Improving Inference for HIV Health Outcomes During the COVID-19 Pandemic

Strategic Insights

— 1 min read

Research at Columbia University has utilized multilevel regression and poststratification (MRP) to estimate the proportion of viral load suppression and means of mental and physical health scales among persons with HIV in New York City. The study addressed a challenge in MRP methods by adapting to scenarios where the joint distribution of poststratifiers is not known. The research applied the adapted MRP to a community health survey disrupted by the COVID-19 pandemic. Results from the study have implications for improving inference for HIV health outcomes during the pandemic.

Key Takeaways:

  • The study utilized multilevel regression and poststratification (MRP) to estimate HIV health outcomes in New York City, addressing a gap in existing MRP methods that assume joint distribution of poststratifiers is known.
  • The adapted MRP model considered both the survey outcome and population sizes of subgroups formed by poststratifiers, using Poisson and negative binomial models for small numbers of poststratifiers and Bayesian additive regression trees for larger numbers.
  • The study applied the adapted MRP to a community health survey from 2018-2021, disrupted by the COVID-19 pandemic, to estimate proportion of viral load suppression and means of mental and physical health scales.
  • Researchers at Columbia University utilized funding from the United States Health Resources & Service Administration (HRSA) to support this research.
  • The study proposes a novel approach to improving inference for HIV health outcomes during the pandemic, with potential applications in epidemiology and public health.

Statistics:

  • The study estimated the proportion of viral load suppression among persons with HIV in New York City.
  • The adapted MRP model used Bayesian additive regression trees for larger numbers of poststratifying variables.
  • The study analyzed data from the 2018-2021 wave of the Community Health Advisory and Information Network survey.
  • The pandemic disrupted sampling and in-person data collection for this survey.

Sources:

  • NewsRx. Findings from Columbia University Yields New Findings on COVID-19 (Multilevel Regression and Poststratification Using Margins of Poststratifiers: Improving Inference for Hiv Health Outcomes During the Covid-19 Pandemic). Medical Letter on the CDC & FDA. November 2, 2025; p 323.
  • Statistics in Medicine. Multilevel Regression and Poststratification Using Margins of Poststratifiers: Improving Inference for Hiv Health Outcomes During the Covid-19 Pandemic, 2025;44.
  • Wiley. 111 River St, Hoboken 07030-5774, NJ, USA.

Read more