HIV criminalization exacerbates subpar diagnosis and treatment across the US: response to the ‘Association of HIV diagnosis rates and laws criminalizing HIV exposure in the United States’

In their article, “Association of HIV diagnosis rates and laws criminalizing HIV exposure in the United States” Sweeney et al. [1] find no association between a state’s criminal exposure laws and the rates of HIV or AIDS diagnosis. Thirty-three states in the United States have implemented laws criminalizing behaviour, including needle sharing and sexual contact, that could put others at risk of transmission [2]. As highlighted by Sweeney et al. [1], the public health impact of these laws should be assessed. However, it was not considered that the annual number of diagnoses alone is uninformative without taking into account epidemiological trajectories. If an epidemic is growing, a constant number of annual diagnoses would actually correspond to a reduced rate of diagnosis among PLHIV. Conversely, the number of diagnoses may remain constant as an epidemic is brought under control if the percentage of PLHIV diagnosed increases. In fact, these inverse associations would be expected. Given that diagnosis is an integral component of treatment-as-prevention strategies, higher rates of diagnosis should be associated with curtailing of HIV epidemics.

We conducted the analysis described in Sweeney et al. [1], but stratified the diagnosis rate into two response variables: i) the proportion of PLHIV diagnosed, and ii) annual percentage change in HIV prevalence. The data required for the replication of the results, and the relevant analysis code, are provided at https://github.com/prathasah/US-law-and-HIV. All socio-economic factors described in [1] were used as explanatory variables. Since our first outcome normalized the total HIV diagnoses with the number of PLHIV (instead of the population size), we included population sizes of states as an additional explanatory variable in our model. Here, we present the results of the full models with all explanatory variables, and the subset of predictors that best explain the response variables (Table 1).

Table 1.

Full (i.e., with all the predictors) and subset (i.e., the best subset of predictors that explain the response) model results for HIV diagnoses and percentage change in HIV prevalence by HIV criminal exposure law and state-specific demographic characteristics in the United States. Bold numbers indicate a significant association (P < 0.05).

	Models
	Full model		Subset model
	β (±SE)	P	β (±SE)	P
Proportion of HIV diagnoses, states, 2008-2012
HIV criminal exposure law	-0.042 (0.017)	0.016	-0.042 (0.016)	0.010
Median household income	0.000 (0.002)	0.873	0.000 (0.002)	0.886
Unemployment rate	0.004 (0.001)	<0.001	0.003 (0.001)	<0.001
Population size	0.028 (0.007)	<0.001	0.027 (0.007)	<0.001
Percentage of population
Less than high school education	-0.047 (0.005)	<0.001	-0.045 (0.005)	<0.001
Residing in urban areas	-0.016 (0.009)	0.083
Below poverty level	0.001(0.001)	0.646	0.001 (0.001)	0.340
Hispanic or Latino	-0.024 (0.022)	0.283	-0.026 (0.019)	0.164
Non-Hispanic black	-0.019(0.015)	0.202	-0.017 (0.013)	0.177
Non-Hispanic white	-0.038(0.020)	0.059	-0.032 (0.018)	0.086
Annual percentage change in HIV prevalence, states, 2009-2012
HIV criminal exposure law	0.625 (0.305)	0.040	0.712 (0.280)	0.011
Median household income	-0.153 (0.192)	0.425
Unemployment rate	-0.343 (0.113)	0.002	-0.311(0.113)	0.006
Population size	0.135 (0.164)	0.409
Percentage of population
Less than high school education	0.199 (0.144)	0.167	0.326 (0.136)	0.017
Residing in urban areas	0.338 (0.159)	0.034	0.351 (0.133)	0.008
Below poverty level	-0.102 (0.180)	0.574
Hispanic or Latino	-0.284 (0.157)	0.071	-0.376 (0.145)	0.010
Non-Hispanic black	0.185 (0.128)	0.149
Non-Hispanic white	0.083 (0.110)	0.449

Counter to the conclusions of Sweeney et al., our analyses indicate that laws criminalizing HIV exposure are associated with lower proportion of HIV diagnosis (full model: ${\chi }_1^2=5.82$, P = 0.016; subset model: ${\chi }_1^2=6.72$, P = 0.009), and increased HIV prevalence (full model: ${\chi }_1^2=4.21$, P = 0.04; subset model: ${\chi }_1^2=6.46$, P = 0.011 ). Educational attainment is associated with declining HIV prevalence and higher diagnosis rates. State population size and urbanicity are associated with higher proportions of PLHIV diagnosed and increasing prevalence, respectively.

As the authors and others have argued [1,3], laws criminalizing HIV exposure can deter people from seeking diagnosis. Given the effectiveness of current antiretrovirals in preventing HIV transmission, diagnosis and treatment are fundamental to both improving individual health outcomes as well as protecting others. Our analyses here underscore the importance of distinguishing between the impact of laws on HIV diagnosis and HIV transmission, as their combined effect on HIV diagnosis rate could be confounding and misleading. Our evaluations of these distinct outcomes demonstrate that laws criminalizing HIV exposure have a negative association with HIV testing, and a positive association with increasing HIV prevalence. Consequently, these laws may be exacerbating HIV transmission, as advocates for legal reform have argued [4]. Our results are consistent with studies that have documented the ramifications of HIV criminalization [5–7]. Our analyses demonstrate that laws criminalizing HIV exposure are not only ineffective, but counterproductive.