County-Level Spatiotemporal Patterns of New HIV Diagnoses and Pre-exposure Prophylaxis (PrEP) Use in Mississippi, 2014–2018: A Bayesian Analysis of Publicly Accessible Censored Data

Figures & data

Figure 1 Counties with truly elevated S_i (i.e., spatial hot spots, stippled) and δ_i (i.e., spatiotemporal hot spots, hatched) in two different scenarios.

Table 1. Cross-classification based on a county’s hot spot or cold spot status of new HIV diagnoses and PrEP use

	New HIV diagnoses			PrEP use			Counties
	Hot spot	Cold spot	Neither	Hot spot	Cold spot	Neither
Category 1	✖			✖			Hinds, Madison, Warren, Rankin, Copiah, Harrison
Category 2	✖				✖		Lee
Category 3	✖					✖	Washington, Bolivar, Leflore, Pike, Forrest, Coahoma
Category 4		✖		✖			Jackson
Category 5		✖			✖		Prentiss, Tishomingo
Category 6		✖				✖	Pearl River, Pontotoc, Union, Tippah, Itawamba
Category 7			✖	✖			Lawrence, Simpson, Hancock
Category 8			✖		✖		Alcorn
Category 9			✖			✖	Others

Note: PrEP = pre-exposure prophylaxis.

Figure 2 Mississippi state and county boundaries.

Figure 3 High proportions of suppressed values (shaded areas) in new HIV diagnosis and pre-exposure prophylaxis (PrEP) use in Mississippi, 2014–2018, using publicly accessible data from AIDSVu. First row: New HIV diagnoses per 10,000 population. Second row: Pre-exposure prophylaxis (PrEP) use per 10,000 population. Third row: Crude PrEP-to-need ratio (PnR).

Figure 4 Main spatial patterns of new HIV diagnosis and pre-exposure prophylaxis (PrEP) use. (A1) Posterior mean (PM) of new HIV diagnosis risk (exp($S_i^{k=1}$)). (A2) Posterior probability (PP) of having a higher risk (PP(exp($S_i^{k=1}$) > 1)). (B1) PM of PrEP use (exp($S_i^{k=2}$)). (B2) PP of having a higher PrEP use (PP(exp($S_i^{k=2}$) > 1)). Ceased hot spots and cold spots cross-hatched in yellow; new hot spots stippled; new cold spots hatched in black.

Figure 5 State-level main temporal trends of new HIV diagnoses (stable) and pre-exposure prophylaxis (PrEP) use (sharply increasing) in Mississippi (exp(${\beta }_{}^k{t}^{*}+{\varphi }_t^k$)), 2014–2018.

Figure 6 County-level departure trends of new HIV diagnoses and pre-exposure prophylaxis (PrEP) use from state-level trends in Mississippi (${\delta }_i^k$): (A1) Posterior mean (PM) of ${\delta }_i^{k=1}.$ (A2) Posterior probability (PP) of having a positive ${\delta }_i^{k=1}.$ (B1) PM of ${\delta }_i^{k=2}.$ (B2) PP of having a positive ${\delta }_i^{k=2}.$

Figure 7 Temporal trends of new HIV diagnosis in Hinds and DeSoto counties, 2014–2018. Solid point: observed new HIV diagnosis rate per 10,000 population; circle: predicted new HIV diagnoses per 10,000 population; gray region: 95 percent CrI of predicted new HIV diagnosis rate; dashed line: predicted county-specific trend of new HIV diagnoses; solid line: state-level trend of new HIV diagnosis.

Figure 8 County classification based on hot spot/cold spot status of new HIV diagnoses and pre-exposure prophylaxis (PrEP) use. The colored dots represent PrEP-to-need ratio (PnR) hot spot or cold spot status in 2018.

Table 2.  The associations between new HIV diagnoses, pre-exposure prophylaxis (PrEP) use, and social determinants of health variables

	New HIV diagnosis	PrEP use
Socioeconomic deprivation	1.45 [1.19, 1.75]	1.06 [0.93, 1.19]
Income inequality	0.92 [0.82, 1.03]	0.99 [0.92, 1.08]
Percentage of population without health insurance	0.88 [0.76, 1.02]	0.96 [0.87, 1.05]
Residential segregation	0.95 [0.80, 1.12]	1.02 [0.93, 1.12]

Note:: Posterior mean (PM) followed by 95 percent credible interval in brackets. Statistically significant coefficients are shown in bold.