Bao Duong and Jean Jacques Brou
June 25, 2026
# Load the packages and data
library(tidyverse)
library(here)
library(janitor)
library(tidyverse)
library(tmap)
library(sf)
library(tigris)
library(leaflet)
knitr::opts_chunk$set(
warning = FALSE,
message = FALSE,
fig.path = "figs/",
fig.width = 7.252,
fig.height = 4,
comment = "#>",
fig.retina = 3
)Topic : Identifying California counties for targeted HIV prevention services: A County-level Analysis of HIV Burden, PrEP Coverage, and Socioeconomic Vulnerability
Data-driven approaches can help inform targeted HIV interventions in communities with a high burden of HIV infections and low PrEP utilization. Therefore, this project seeks to identify potential priority counties in California for targeted HIV prevention services using the 2023 AIDSVu datasets. We will explore county-level patterns of HIV burden, PrEP utilization, and socioeconomic vulnerability by answering the following research questions:
1. How do HIV diagnosis rates vary across California counties? Do counties with higher HIV diagnosis rates also have higher PrEP utilization? 2. Which county-level socioeconomic factors (e.g., percent living in poverty, education attainment, uninsured rate, and unemployment) are associated with higher HIV diagnosis rates in California?
3. Do racial disparities in HIV diagnosis rates exist across California counties?
Source of the data: We will use AIDSVu Dataset 2024 (By Rollins Emory University) for three different datasets
AIDSVu_County_NewDX_2023-20250726 AIDSVu_County_PrEP_2023_20250501 AIDSVu_County_SDOH_2023-20250726
Data source here
We use 2023 AIDSVu Data collected by state and local health departments, and de-duplicated and processed by the U.S. Centers for Disease Control and Prevention (CDC) to meet data quality standards for comparability and reliability. All 50 states, the District of Columbia (DC), and U.S. territories collect comparable confidential, name-based case reports of HIV infection and social determinants of health.
We downloaded the Social Determinants of Health, County New Diagnoses, and County PrEP Data for county level data in America and its associated new diagnosis rates for the county and for different racial groups. It also had related SDOH data. We restricted our analysis to California, remove unnecessary variables and merged the three data by keys such: “year”, “geoid”, “state”, “state_abbreviation”, “county_name”. We downloaded 2023 California shape files and merged it to the clean data.
There were also rate stability variables given to indicate if the related diagnosis or prEP rates were reliable or not. NAs for diagnosis or prEP rates were added if the rate stability was indicated as not reliable.
The interactive map examine geographic variation in new HIV diagnosis rates and PrEP use by county. We visualize HIV new diagnoses rate per 100,000 and represented county PrEP use rate per 100,000 with dots to show high or low PrEP use. Higher new hiv diagnoses rate are concentrated in larger counties in the south and middle of California. Counties with higher new hiv diagnoses rate have lower PrEP use rate compared to smaller counties with lower new hiv diagnosis rate. The interactive map shows that Kern county has highest new HIV diagnoses rate (29) compared to Imperial county with highest PrEp users rate and a lower new HIV diagnoses rate (18).
This plot shows cards that show the average percents for each county for poverty, less than high school education, uninsured rate, and unemployment rate for 2023. Also, it shows a radar chart that normalized these soceocnomic factor from 0 to 1 across all counties. Specific California counties can compare how they do with other counties this way. The visualization also compares to a high and low HIV diagnosis group (seperated by median). There appears to be a relationship between having a less than High School education and higher HIV diagnosis rates for California counties. There appears to not be a strong relationship for poverty, unemployment, and uninsured rates for HIV diagnosis.
The interactive map examine geographic variation in new HIV diagnosis rates and PrEP use by county. We visualize HIV new diagnoses rate per 100,000 and represented county PrEP use rate per 100,000 with dots to show high or low PrEP use. Higher new hiv diagnoses rate are concentrated in larger counties in the south and middle of California. Counties with higher new hiv diagnoses rate have lower PrEP use rate compared to smaller counties with lower new hiv diagnosis rate. The interactive map shows that Kern county has highest new HIV diagnoses rate (29) compared to Imperial county with highest PrEp users rate and a lower new HIV diagnoses rate (18).
This grouped bar chart shows the average of new HIV diagnosis rates per 100,000 among different racial groups in the State of California.The bar chart shows a higher average rate for black and Hispanic compared to white and other minorities.
Counties bar chart also shows the different HIV diagnosis rates by race for each county. For most counties, Black groups experience much higher HIV diagnosis rates where it was about 3-4 times more than their White counterparts. Hispanic groups had twice as high diagnosis rates as their White Counterparts. For some counties, Multiple Races, and NH Pacific Islander also were at least twice as high as their White counterparts. Asian groups had less HIV diagnosis rates than their White counterparts.
[Click the shiny dashboard to view counties specific average] (https://bao-duong17.shinyapps.io/shiny-app/)
#Shiny app: [HIV Diagnosis Analysis Dashboard — California] Identifying California counties for targeted HIV prevention services: A County-level Analysis of HIV burden, PrEP Coverage, and socioeconomic vulnerability
We had to use 2023 as that was the most recent year available for AIDSVu for their HIV diagnosis, PrEP, and socioeconomic factors. A good amount of the data was suppressed due to cases being too low to share in a dataset or the state department requested to not share to AIDSVu. The dataset also shared rates for HIV diagnosis and PrEP, but the rates had a related rate stability variable that directed us to view the rate with caution due to cases being too low (for example, American Indian for many counties) and so we mutated the dataset to remove those low confidence values from the visualizations. We also did not have PrEP rate by race and so we were unable to make a stronger analysis to consider potential interplay of race to PrEP to HIV diagnosis. Finally, our chloropleth map analyzed by counties, but perhaps there was a better way to analyze rates and relationships more meaningfully by population hubs in California.
This analysis identified important county-level patterns in HIV burden, PrEP utilization, and socioeconomic vulnerability across California. Counties with higher HIV diagnosis rates generally had lower high school educational attainment and substantial racial disparities, with Black and Hispanic populations experiencing disproportionately higher HIV diagnosis rates than White populations. Higher new HIV diagnoses rates are concentrated in larger counties in the south and middle of California. PrEP utilization rates vary across the state; counties with higher new HIV diagnosis rates have lower PrEP use rates compared to smaller counties.
These findings can help public health agencies prioritize resources and target HIV prevention strategies, including increasing PrEP access and addressing underlying social determinants of health in the communities most affected by HIV. Future research should examine more recent data and incorporate race-specific PrEP utilization measures to understand better and address these disparities.