Sexually transmitted infections in association with area-level prostitution and drug-related arrests

Abstract

Objectives:

Examine the mediators and moderators of area-level prostitution arrests and sexually transmitted infections (STI) using population level data.

Methods:

Using justice and public health STI/HIV data in Marion County (Indianapolis), Indiana, over an 18-year period, we assessed the overall association of area-level prostitution and drug-related arrests and STI /HIV, and mediators and moderators of the relationship. Point-level arrests were geocoded and aggregated by census block group.

Results:

Results indicate a positive relationship between numbers of prostitution arrests and area-level STI rates. There was a dose-response relationship between prostitution arrests and STI rates when accounting for drug-related arrests. The highest quintile block groups had significantly higher rates of reported chlamydia (IRR: 3.29, 95% CI: 2.82, 3.84), gonorrhea (IRR: 4.73, 95% CI: 3.90, 5.57), syphilis (IRR: 4.28, 95% CI: 3:47, 5.29), and HIV (IRR: 2.76, 95% CI: 2.24, 3.39) compared with the lowest quintile. When including drug arrests, the second (IRR: 1.19, 95% CI: 1.03, 1.38) and the third (IRR: 1.20, 95% CI: 1.02, 1.41) highest quintile block groups had lower IRR for reported rates of chlamydia, indicating that drug arrests mediated the prostitution arrest effect.

Conclusions:

These findings inform public health agencies and community-based organizations that conduct outreach in these areas to expand their efforts to include harm reduction and HIV/STI testing for both sex workers and individuals experiencing substance use disorder. Another implication of these data is the importance of greater collaboration in public health and policing efforts to address overlapping epidemics that engage both health and legal intervention.

Short Summary

Our study demonstrates the overlapping but distinctive ways by which communities are differentially represented in STI surveillance data: both prostitution arrests and drug-related arrests are associated with area-level STI rates.

Introduction

Nearly 20 million new sexually transmitted infections (STIs) occur annually in the United States, and they disproportionately affect the incarcerated population, female sex workers, and people who use drugs (PWUD).^1–4Female sex workers and men who have sex with men are among the populations most affected by STIs and HIV globally and domestically.¹ Among female sex workers, rates of STIs range from 9 to 60 times that of the general population,^5–7 similarly, as many as one in five female sex workers in the United States is HIV positive (95% CI: 13.5%−21.9%).⁸ Female sex workers also commonly participate in injection and non-injection drug use.⁹ Among PWUD via injection, exchanging sex for money or drugs within the past year increased the likelihood of STI nearly fourfold.⁴ For instance, injection drug use among female sex workers increased the likelihood of an active syphilis infection by three times in two US-Mexico border cities.¹⁰ Injecting and using drugs places female sex workers at elevated risk for acquiring and transmitting STIs within their network, as engaging in high-risk activities such as condom less sex is more likely when actively using drugs.^4,11,12

Neighborhood factors associated with higher rates of STIs include high incarceration rates, economic disadvantage, and low education levels.^13–17 Drug use and dealing are associated with higher rates of STIs at the individual and neighborhood level.^16–18

Neighborhoods with known drug markets have highly interconnected network structures which help maintain disease transmission and are also associated with selecting a high-risk sexual partner and having a current infection with a bacterial STI.¹⁶ Research into the spatial association of neighborhood drug markets and STIs demonstrated an 11% increase in gonorrhea in the local neighborhood and a 27% increase of gonorrhea in adjacent neighborhoods.¹⁷ These results suggest that drug markets and their sexual networks extend beyond the boundaries of census block groups.¹⁷ These studies, however, largely examined only gonorrhea and other STIs may have a different association with drug markets and neighborhoods.

The transmission of STIs among sexual networks is largely driven by core transmitters and key social meeting locations (e.g., drug and sex markets).¹⁹ Core transmitters maintain disease rates within populations as they repeatedly acquire and transmit the disease among their network.²⁰ Core groups have largely been defined by the number of sex partners and network connections within core transmitters networks.^21,22 In addition to highly connected sexual networks, neighborhoods help explain variations in STI transmission.^22,23 Variations in STI rates and counts have been observed across and within cities.^24–26 Recent research examined social meeting places characterized as drug markets, sex markets or a combination of drug and sex markets. Sex markets have been defined as spaces where sex is exchanged for money or drugs, venues included bars, street corners, parks and schools.¹⁹ Results indicate drug markets, sex markets, and drug/sex markets are more likely to be frequented by core transmitters most likely to transmit STIs; and these risk environments may be key drivers in STI transmission within the communities.¹⁹ Less is known, however, whether drug and sex markets have independent influences on STI rates, and whether those influences pertain to multiple STI within the same geographic space. Addressing these issues could improve focusing of resources in specific neighborhoods, allowing more effective and efficient resource allocations.²⁷

Understanding how prostitution and drug-related arrests are associated with STI rates may inform public health and community policing strategies. Therefore, the objectives of this study were to assess the association of area-level sex work and drug-related arrests and STI infection to better understand if they act as mediators or moderators of this relationship. Using population-level justice and public health STI data, this study expands prior research by including multiple STIs, examining the relationship of both prostitution and drug-related arrests over an 18-year time period, and discusses how potential public health and police collaboration can improve outcomes for PWUD and sex workers.

Methods

Study design and population

We conducted a retrospective cohort study of area-level criminal arrest and STI morbidity for chlamydia, gonorrhea, syphilis, and HIV in Marion County (Indianapolis), Indiana from 2000–2018. Criminal arrest and STI morbidity data were obtained in collaboration with the Indianapolis Metropolitan Police Department and Marion County Public Health Department, respectively.

Measures

The primary predictor measures were area-level prostitution and drug-related arrests. We obtained all arrest data from 2000 – 2018 from the Indianapolis Metropolitan Police Department. The primary exposures of interest were prostitution (n=19,430) and drug arrests (n=103,088). There was an average of 1,022.6 prostitution arrests per year and 5,425.7 drug-related arrests per year. Prostitution arrests were defined as any arrest for engaging in transactional sex, and drug-related arrests were defined as any arrest for drug distribution, paraphernalia, and drug possession for any controlled substance (e.g., marijuana, heroin, cocaine, etc.). Dual arrests were categorized when an individual was arrested for both prostitution and drug-related offense during the same incident and were divided into a binary measure of lowest (n=475) and highest quartiles (n=157) due to low numbers of dual arrests across census block groups. Point-level prostitution and drug-related arrests were geocoded and aggregated by Census block group (n=597). Census block groups were divided into quintiles by prostitution (lowest – 82 and highest – 156)and drug-related arrests (lowest – 105 and highest – 126), using the first quintile (lowest arrests) as the reference category.

The outcome measure for this study was STI incidence rate ratios, defined as the number of new chlamydia, gonorrhea, syphilis, and HIV diagnoses within Census block groups. We obtained all reported chlamydia, gonorrhea, syphilis, and HIV cases (n=266,868) diagnosed in Marion County from 2000–2018. When two positive tests for the same STI organism occurred for an individual within 30 days of each other,^28–30 only the initial positive test was included to avoid double counting. We defined co-infection rates as any individual with a positive test for two different STIs within 14 days of each other.

Census data moderating factors were aggregated at Census block group level and included three binary measures of race, ethnicity and socioeconomic status. Race was defined as Black if over >75% of residents identified as Black individuals (n=64) and ethnicity was defined as Latino if over >20% of residents identified as Latino (n=64) and socioeconomic status was defined as >60% living below 200% of federal poverty line (n=42). A measure was created to indicate time (2000–2003; 2004–2009; 2010–2013; 2014–2018) to adjust for potential temporal trends.

Geocoding

Addresses from both STI and arrest data sources were geocoded to street location using ArcGIS v10.8 and Marion County base maps. Among the STI data, 83% of residential addresses (n=222,118) were successfully geocoded, geotagged, and aggregated to their associated Census block group. Cases that did not geocode contained missing address information, were outside Marion County, or the individual was listed as homeless.

A total of 98% (n=115,443) of prostitution and drug-related arrest locations were successfully geocoded and aggregated to their associated Census block group. In addition to arrest locations, we also geocoded the residential address of the arrested individual listed in the arrest report and successfully geocoded 75%. The cases that did not geocode contained a non-street address (e.g., Mexico, 123 Main St.), recorded refusal to provide address to the arresting officer, or were listed as homeless.

Analysis

We calculated population based STI rates per 100,000 for each census block group overall and by year. We performed negative binomial and zero-inflated negative binomial regression models to estimate incident rate ratios (IRR) of each STI.^31s Incidence rate ratios were stratified by 4-year time periods (2000–2003, 2004–2008, 2009–2013,2014–2018), minority and ethnic composition, and poverty level of each Census block group.

Density maps were created using a kernel density function (KDF) of point-level data, categorizing along a color gradient (red=highest concentration; blue=lowest concentration). The kernel density algorithm examines each incident point and calculated intensity rates based on how many incidents are clustered near the given incident point. Near incidents are defined as those failing within a predetermined search radius that extends out from the incident point under examination. The KDF was color-coded into quintiles for prostitution arrests, drug arrests, and chlamydia rates. Gonorrhea, syphilis, and HIV rates are presented based on chlamydia decile cutoffs.

Results

Yearly population rates of STIs indicate trends over the study period. Across all STIs, rates were higher in the early 2000s, dropped between 2009–2012, and have since increased to similar rates seen in the early 2000s (Table 1). Density maps show clusters of prostitution and drug arrests and rates of chlamydia, gonorrhea, syphilis and HIV in Marion County (Figure 1). Comparing the arrests density maps to the STI density maps, several patterns emerge that suggest associations between prostitution and drug-related arrests and STI incidence rates.

Table 1 –

Yearly Sexually Transmitted Infection Rates per 100,000 population, Marion County, Indiana, 2000 – 2018

Year	Chlamydia	Gonorrhea	Syphilis	HIV
2000	918.3	517.8	194.7	-
2001	1114.2	576.4	195.9	-
2002	1176.4	609.0	140.0	-
2003	1084.2	516.5	124.3	-
2004	1032.2	563.2	108.2	-
2005	1252.3	692.2	96.3	-
2006	1229.7	762.5	92.2	-
2007	1298.7	799.9	78.6	21.7
2008	1219.4	649.9	111.9	22.9
2009	487.6	222.1	44.2	21.0
2010	623.0	240.1	21.7	20.8
2011	894.0	273.9	16.6	22.7
2012	1036.3	338.8	21.7	22.0
2013	1053.1	356.8	25.3	24.4
2014	1023.1	367.1	19.9	24.4
2015	977.6	354.1	30.6	21.4
2016	1077.0	439.8	30.3	22.2
2017	1154.8	511.7	32.7	25.6
2018	1072.7	418.5	34.8	23.0

Figure 1 –

Density maps* of Marion County, Indiana: Prostitution/drug arrests and chlamydia, gonorrhea, syphilis, and HIV rates

*Density maps were created using a kernel density function (KDF) of point-level data, categorized along a color gradient (red=highest concentration; blue = lowest concentration)

The number of prostitution arrests within a census block group was positively correlated with the IRR for all types of STI (Table 2). In an unadjusted model for prostitution arrests, the highest quintile block groups (i.e., those with the highest density of prostitution arrests) had significantly higher rates of reported chlamydia (IRR: 3.29, 95% CI: 2.82, 3.84), gonorrhea (IRR: 4.73, 95% CI: 3.90, 5.57), syphilis (IRR: 4.28, 95% CI: 3:47, 5.29), and HIV (IRR: 2.76, 95% CI: 2.24, 3.39) compared with the lowest quintile.

Table 2 –

Incident rate ratios of STIs by prostitution and drug arrest quintiles, Indianapolis, Indiana, 2000 – 2018

	Prostitution Arrests				Drug Arrests				Prostitution & Drug Arrests				Prostitution Arrests/ Drug Arrests/Race/Ethnicity/Poverty
	CT	GC	SPYH	HIV	CT	GC	SPYH	HIV	CT	GC	SPYH	HIV	CT	GC	SPYH	HIV
Prostitution arrest quintiles	IRR (95% CI)	IRR (95% CI)	IRR (95% CI)	IRR (95% CI)	IRR (95% CI)	IRR (95% CI)	IRR (95% CI)	IRR (95% CI)	IRR (95% CI)	IRR (95% CI)	IRR (95% CI)	IRR (95% CI)	IRR (95% CI)	IRR (95% CI)	IRR (95% CI)	IRR (95% CI)
1 (lowest – n=156)	ref	ref	ref	ref	ref	ref	ref	ref	ref	ref	ref	ref	ref	ref	ref	ref
2 (n=137)	1.92 (1.60, 2.29)	2.26 (1.79, 2.86)	1.75 (1.39, 2.21)	1.43 (1.18, 1.74)					1.19 (1.03, 1.38)	1.18 (0.98, 1.42)	1.10 (0.89, 1.38)	1.18 (0.98, 1.42)	1.12 (0.97, 1.30)	1.18 (0.93, 1.34)	1.05 (0.85, 1.31)	1.16 (0.96, 1.40)
3 (n=86)	2.62 (2.16, 3.17)	3.45 (2.69, 4.41)	2.75 (2.12, 3.57)	2.06 (1.69, 2.52)					1.20 (1.02, 1.41)	1.17 (0.95, 1.44)	1.22 (0.95, 1.56)	1.51 (1.22, 1.87)	1.15 (0.99, 1.34)	1.13 (0.93, 1.38)	1.19 (0.93, 1.52)	1.49 (1.20, 1.84)
4 (n=100)	3.08 (2.55, 3.71)	4.25 (3.36, 5.37)	3.71 (2.83, 4.84)	2.60 (2.09, 3.22)					1.05 (0.91, 1.23)	1.04 (0.85, 1.28)	1.28 (0.93, 1.75)	1.68 (1.31, 2.15)	1.11 (0.95, 1.29)	1.10 (0.89, 1.35)	1.37 (1.00, 1.89)	1.70 (1.32, 2.18)
5 (highest – n=118)	3.29 (2.82, 3.84)	4.73 (3.90, 5.75)	4.28 (3.47, 5.29)	2.76 (2.24, 3.39)					0.87 (0.73, 1.03)	0.85 (0.69, 1.14)	1.21 (0.89, 1.62)	1.54 (1.20, 1.99)	1.06 (0.89, 1.27)	1.11 (0.88, 1.39)	1.52 (1.11, 2.07)	1.67 (1.28, 2.17)
Drug arrest quintiles
1 (lowest – n=121)	ref	ref	ref	ref	ref	ref	ref	ref	ref	ref	ref	ref	ref	ref	ref	ref
2 (n=119)					2.08 (1.78, 2.44)	2.49 (2.04, 3.04)	1.76 (1.39, 2.22)	1.68 (1.34, 2.11)	2.00 (1.71, 2.35)	2.41 (1.95, 2.95)	1.69 (1.33, 2.14)	1.55 (1.25, 1.92)	1.95 (1.67, 2.29)	2.36 (1.92, 2.89)	1.69 (1.33, 2.14)	1.54 (1.24, 1.91)
3 (n=119)					2.60 (2.22, 3.04)	3.49 (2.87, 4.24)	2.57 (2.00, 3.30)	1.93 (1.55, 2.40)	2.47 (2.10, 2.90)	3.34 (2.70, 4.10)	2.31 (1.78, 2.99)	1.53 (1.24, 1.91)	2.23 (1.92, 2.61)	2.98 (2.44, 3.64)	2.11 (1.63, 2.73)	1.51 (1.21, 1.87)
4 (n=119)					4.50 (3.90, 5.18)	7.19 (6.01, 8.59)	4.72 (3.74, 5.97)	2.60 (2.09, 3.24)	4.46 (3.80, 5.23)	7.18 (5.86, 8.79)	4.06 (3.08, 5.34)	1.92 (1.51, 2.42)	3.63 (3.09, 4.27)	5.67 (4.58, 7.02)	3.32 (2.49, 4.42)	1.80 (1.40, 2.32)
5 (highest – n=119)					6.37 (5.54, 7.32)	11.2 (9.41, 13.2)	7.23 (5.88, 8.89)	3.56 (2.89, 4.39)	7.02 (5.92, 8.32)	12.4 (10.0, 15.3)	6.16 (4.59, 8.27)	2.49 (1.93, 3.21)	4.27 (3.50, 5.21)	7.00 (5.44, 8.99)	3.78 (2.66, 5.37)	2.12 (1.56, 2.88)
>75% Black (n=64)													1.92 (1.72, 2.13)	2.11 (1.85, 2.39)	1.87 (1.54, 2.27)	1.26 (1.03, 1.55)
>20% Hispanic (n=64)													1.34 (1.18, 1.52)	1.17 (1.00, 1.37)	1.06 (0.87, 1.28)	1.04 (0.86, 1.25)
>60% Below Poverty Line (n=42)													1.23 (1.09, 1.39)	1.31 (1.13, 1.51)	1.22 (0.95, 1.55)	1.11 (0.79, 1.55)

^*CT – chlamydia, GC – gonorrhea, SYPH – syphilis, IRR – incident rate ratio, CI – confidence interval

^**Bolded numbers indicate significant incident rate ratios (p<0.05).

In an unadjusted model for drug arrests only, the highest quintile block groups had significantly higher rates of reported chlamydia (IRR: 6.37, 95% CI: 5.54, 7.32) gonorrhea (IRR: 11.2, 95% CI: 9.41,13.2), syphilis (IRR: 7.23, 95% CI: 5.88, 8.89), and HIV (IRR: 3.56, 95% CI: 2.89, 4.39). When prostitution and drug arrests were included in the model, the second (IRR: 1.19, 95% CI: 1.03, 1.38) quintile block groups and the third (IRR: 1.20, 95% CI: 1.02, 1.41) highest quintile block groups had lower IRR for reported rates of chlamydia, indicating that drug arrests mediated the prostitution arrest effect. The top three prostitution quintiles were associated with increased rates of HIV (IRR: 1.51, 95% CI:1.22,1.87; 1.68, 95% CI: 1.31,2.15; 1.54, 95% CI:1.20,1.99) but are also mediated by drug arrests. Other prostitution quintiles were not associated with higher incidence rate ratios for chlamydia, gonorrhea or syphilis when controlling for drug-related arrests. There was evidence of moderation by area racial and ethnic composition, however, differences were observed across STIs. The percentage of Black composition in a census block group moderated the association between sex work and STI incidence rate ratios in the fourth and fifth quintile census block groups for syphilis (IRR: 1.37, 95% CI:1.00,1.89; 1.52, 95% CI: 1.11, 2.07) and HIV (IRR: 1.70, 95% CI:1.32,2.18; 1.67, 95% CI: 1.28, 2.17) when adjusting for area-level drug arrests and socioeconomic status, however, the percent of Latino composition in a census block group showed no consistent association.

As displayed in Table 3, the top two quintiles of dual (prostitution and drug) arrests were positively associated with all STIs. IRR were stronger across all STIs for dual arrests than for prostitution and drug-related arrests (IRR: chlamydia 1.80, 95% CI:1.59, 2.03, gonorrhea 2.11, 95% CI:1.82, 2.45, syphilis 2.09, 95% CI:1.75, 2.49, HIV 1.85, 95% CI: 1.58, 2.16). This relationship was also moderated by census block group composition for Black, Latino, and poverty levels in the top quintile (IRR: chlamydia 1.60, 95% CI: 1.43, 1.78, gonorrhea 1.92, 95% CI: 1.67, 2.21, syphilis 2.00, 95% CI: 1.66, 2.40, HIV 1.77, 95% CI: 1.51, 2.09) (Table 3). We also examined STI rates across four time periods and results were consistent with prior findings presented indicating no variation across time.

Table 3 –

Incident rate ratios of STIs by top half of dual arrests, Indianapolis, Indiana, 2000–2018

Dual arrest census block groups
	CT	GC	SPYH	HIV	CT	GC	SPYH	HIV
	IRR (95% CI)	IRR (95% CI)	IRR (95% CI)	IRR (95% CI)	IRR (95% CI)	IRR (95% CI)	IRR (95% CI)	IRR (95% CI)
Lowest (n=475)	ref	ref	ref	ref	ref	ref	ref	ref
Highest (n=157)	1.80 (1.59, 2.03)	2.11 (1.82, 2.45)	2.09 (1.75, 2.49)	1.85 (1.58, 2.16)	1.60 (1.43, 1.78)	1.92 (1.67, 2.21)	2.00 (1.66, 2.40)	1.77 (1.51, 2.09)
>75% Black (n=64)					2.95 (2.66, 3.27)	3.61 (3.17, 4.12)	2.65 (2.22, 3.15)	1.59 (1.36, 1.86)
>20% Hispanic (n=64)					1.59 (1.40, 1.81)	1.43 (1.21, 1.70)	1.18 (0.98, 1.43)	1.12 (0.95, 1.32)
>60% Below Poverty Line (n=42)					1.64 (1.40, 1.92)	1.87 (1.54, 2.28)	1.79 (1.25, 2.56)	1.55 (1.11, 2.15)

^*CT – chlamydia, GC – gonorrhea, SYPH – syphilis, IRR – incident rate ratio, CI – confidence interval

^**Bolded numbers indicate significant incident rate ratios (p<0.05)

Discussion

Using population-level arrest and public health STI data, we examined prostitution and drug-related arrests as they related to risk of area-level STI rates over an 18-year period. STI rates were considerably higher than national rates reported by the CDC. For instance, in 2018 the national rate for chlamydia was 539.9 per 100,000; whereas the Marion County rate was 1,072.7 per 100,000. Marion County had the second highest syphilis rate in the United States in 2000 but dropped to the fiftieth percentile in 2018 and has not reached rates as high as the early 2000s.¹ Our data demonstrate the overlapping but distinctive ways by which communities are differentially represented in STI surveillance data: both prostitution arrests and drug-related arrests are associated with area-level STI rates. However, drug-related arrests substantially mediate the relationship of prostitution arrests and STI area-level infection rates. Our data also showed the importance of poverty and race in moderating the relationships of arrests and STI. When stratifying by the top half of dual arrests for both prostitution and drug use, these, dual arrests were significantly associated with higher risk of all STIs and the risk for chlamydia, gonorrhea and syphilis is even higher in predominantly Black census block groups. These findings suggest that arrests associated with drug markets and drug use are key determinants in STI transmission^4,10,16,17 and support prior research that drug markets, sex markets and both drug and sex markets are associated with higher rates of STI transmission risk.¹⁹

Since strategies for drug market policing are largely determined at local levels, this indicates – at least from a public health perspective – that STI prevention would benefit from approaches such as formal police-public health partnerships.^32s Our findings suggest that important relationships exist between STI infection and sex work and, to a greater degree, drug-related arrests within specific communities, especially for chlamydia, gonorrhea and HIV. This finding can inform public health agencies and community-based organizations that conduct outreach in these areas to expand their efforts to include harm reduction and HIV/STI testing for both sex workers and individuals experiencing substance use disorder. Co-location of STI screening in jails,^33s community courts,^34s,35s drug courts,^36s and syringe services programs^37s results in case finding and treatment and are promising approaches for accessing hard to reach populations who are vulnerable to STI. Such an effort could result in conserved resources and better health outcomes.

Another implication of these findings is the importance of greater collaboration in public health and policing efforts to address overlapping epidemics that engage both health and legal intervention. Although police and public health collaborations are not a new concept, the recent social movement has called for police reform to better assist the communities it serves and may be a key opportunity for new public health and police partnerships. Police agencies could provide data to local health departments on areas of high sex work and drug arrests, helping these agencies to better target STI testing and prevention efforts on their already stretched work force.^32s Police and public health co-response has proven successful in crisis intervention teams, which comprise police officers and mental health professionals to assist individuals suffering from mental illness.^38s Other police-public health partnerships, such as the Cardiff model, have generated new policies and place-based initiatives in violence prevention.^32s,39s

Understanding current police practices and partnering with law enforcement has important implications for sex workers, PWUD, and STI transmission, as certain policing practices may force PWUD to avoid carrying clean syringes, inject quickly or with unsterile needles which can increase the likelihood of STI transmission.^40s Neighborhoods with higher STI rates are also often times the same communities with higher violent crime rates;^41s,42s which has implementations regarding police enforcement. Policing methods such as ‘hot spot’ policing directs more police resources to communities with higher rates of crime;^43s,44s however, individuals may not receive STI testing or other medical services they need within the criminal justice system.^45s In partnering with public health officials, police need to also move more towards social harm policing efforts for communities experiencing co-occurring crime, medical emergencies, and drug use, compared to just crime^46s and models of procedural justice policing.^47s Procedural justice policing emphasizes treating residents with dignity and respect and has demonstrated decreases in complaints against the police, use of force, and indicates improved legitimacy between residents and the police. This improved legitimacy with police is important when designing co-response collaborations as public health workers and researchers have long worked to gain and maintain trust with these typically hard to reach populations.^48s,49s

There have been additional advances in policing practices in recent years that seek to link individuals to needed services and not into the justice system. For instance, arrest diversion programs, such as Seattle’s Law Enforcement Assisted Diversion program diverts individuals suspected of low-level drug and prostitution charges to social, medical, and legal services compared to arrest. Results indicate a reduction in future arrest for individuals involved in low-level drug and prostitution activity by nearly 60 percent.^50s Such diversion programs may be an opportunity to implement STI testing among PWUD and sex workers. For example, extragenital testing for chlamydia and gonorrhea may be avenues for improved testing.^51s Additionally, syringe service programs have been implemented by many states and aim to reduce infectious disease and improve outcomes for people who inject drugs. Studies suggest police support such programs; however, officer education, training, and understanding legal constraints are needed when implementing such programs._52s

Given that individuals arrested for prostitution and drug-related offenses may not necessarily live in the neighborhood in which they are arrested, we examined both the arrest incident address and the residential address of the arrestee listed in the arrest report. Findings were nearly identical, indicating no difference in risk levels regarding place, possibly because individuals reside in the area in which they are arrested or that they live in neighborhoods with similar STI infection rates. Lastly, we stratified models by time periods to examine differences across time and results were consistent across different historical intervals within the 18-year study timeframe. All these stratified analyses, which showed consistent relationships between area-level arrests and STI risk, point to the robustness of these findings.

We did not examine access to health care, so it is unclear whether individuals and communities in this study had equal access to STI testing and clinical care. Studies have shown sex workers often lack health care due to fear of arrest, lack knowledge of testing availability, and life distractions, such as the need to meet basic needs for food, shelter and safety.^53s Furthermore, living in a disadvantaged neighborhood often limits access to health care providers and decreases the likelihood of preventative care.^54s A recent study in Detroit highlighted the complexities of accessing healthcare services when intersecting sex work, drug use, and poverty within individuals lives.^55s

There are several limitations of this study. It is an ecological study design that precludes knowing whether individual-level associations in fact exist (ecological fallacy); however, our findings are important for STI surveillance and directing public health resources to specific communities regardless of the causal mechanisms. Our data only includes one metropolitan area and therefore may lack generalizability to smaller or larger metropolitan areas. Marion County incorporates three smaller cities with their own police departments. We used arrest data from Indianapolis Metropolitan Police Department, which serves over 90% of Marion County. Census block groups from the three smaller cities were removed from these analyses. Additionally, prostitution and drug-related arrests do not necessarily reflect sex work and drug activity within an area and does not allow us to examine heterosexual versus homosexual activity, which may or may not have different risk profiles. Differences in routine yearly screening and symptomatic screening across STIs was not accounted for; similarly, we do not know how testing was done within these communities during our study timeframe and HIV data were only available for the later part of the study. Although we had a population level of drug-related arrest, we did not assess drug type, quantity, or quality, and differences may exist in regard to STI transmission risk. Furthermore, census block groups may not be the best indicator of area-level effect and stronger effects may be observed at smaller levels of geography. Lastly, we were not able to measure access to health care, possible bias in STI testing, or account for community programs which may have been implemented during our study.

Conclusion

Prostitution arrests are associated with STI risk; however, this relationship is mediated by drug arrests. These data suggest that important relationships exist between STI risk and sex work and, to a greater degree, drug arrests within communities that could better inform intervention activities. Improved collaborations between public health and policing are needed to address overlapping epidemics that require both health and legal intervention.