Owusu-Edusei, Kwame Jr PhD, PMP; Doshi, Sonal R. MS, MPH
Chlamydia, gonorrhea, and syphilis are the 3 major curable sexually transmitted diseases (STDs) that continue to pose considerable medical problems in the United States.1 In the last few years, chlamydia rates have increased, whereas gonorrhea rates have somewhat decreased.1 Over the same time period, syphilis has shown a resurgence due, in part, to an increase in rates among men who have sex with men.1
In the United States, the diagnosis, treatment, and prevention of STDs have traditionally been the domain of publicly funded STD and family planning (FP) clinics, although more and more services are shifting to the private sector.2 A substantially large proportion of STD cases are diagnosed in the private sector at both the national and local levels.1 With the impending healthcare reform, the corollary to the declining number of STD/FP clinics may be an increased speed in the shift in STD care from the STD/FP clinics to the private sector providers, eventually making up for the decrease in case detection from the STD/FP clinics. Recent reports show that for many areas in the United States, STD clinics diagnosed roughly 25% to 50% of primary and secondary (P & S) syphilis cases, 15% to 35% of gonorrhea cases, and 5% to 15% of chlamydia cases.3 Additionally, according to recent national surveillance reports, STD clinics have been the source of at least 10% of the total number of cases of chlamydia, gonorrhea, and P & S syphilis reported over the last few years.1,4–6
Although chlamydia, gonorrhea, and syphilis are notifiable STDs in the United States, there are differences in the regulations and reporting policies established and/or used by each state and local STD program.1 In Texas, various individuals and entities (including STD and FP clinics) are required by law to report chlamydia, gonorrhea, and syphilis cases to the health authorities.7 Data from the National Electronic Telecommunications System for Surveillance covering the last decade indicated that the proportion of STDs diagnosed by STD clinics in Texas were quite substantial and similar to the numbers reported nationwide (30%–47% of P & S syphilis, 20%–36% of gonorrhea, and 5%–16% of chlamydia).
In an impact analysis conducted after the closure of an STD clinic in the District of Columbia, a marked decrease in syphilis cases was reported.8 This finding suggests that the STD clinics are important for detection. The closure of the STD clinic may have had other potential outcomes such as the long-term adverse impact on the burden of disease within the community because infected individuals and their partners were not identified and treated (and/or counseled) to curb transmission.3,9
Reports suggest that some jurisdictions discontinued the provision of direct clinical services in 1993 when national health reform seemed imminent.10,11 The discontinuation of STD services is occurring presently in light of the impending implementation of the 2009 health reform bill and because of budget shortfalls at both the federal and state levels.3 A recent study reported a 10% decrease in the number of STD clinics in the United States over the last decade.12 This discontinuation of provision of direct services including STD care, may impact the detection and control of STDs at both the local and national levels.
Currently, information on the potential impact of the presence of an STD and/or FP clinic on STD detection and control is scant. In view of this, the objective of this study was to examine the association between the presence of STD clinics and/or publicly funded Title V, X, and XX FP clinics and county-level detection and control of 3 reportable STDs (chlamydia, gonorrhea, and P & S syphilis) in the state of Texas using 2000 and 2007 data. Specifically, we examined the following 2 hypotheses:
1. Are the counties with STD and/or FP clinics reporting relatively more cases of STDs?
2. Is having an STD service(s) in a county associated with decreases in the burden of STD between 2000 and 2007?
The results from our study can inform policy and decision-making in regards to the presence of, and access to, STD services. Finally, these results may be useful in assessing the potential public health impacts/benefits, as projections for resources are made for STD surveillance and control efforts with health reform on the horizon.
Data on chlamydia, gonorrhea, and P & S syphilis cases (for all ages and gender) were obtained from the National Electronic Telecommunications System for Surveillance for 1999, 2000, 2001, 2006, 2007, and 2008 for all 254 counties in Texas. We chose Texas for this analysis because it has the largest number of counties (n = 254). Second, it is a southern state with relatively high STD rates.1 Nonetheless, the methods can be applied to any state with lower STD rates as long as the fifth rule of multiple regression analyses is not flouted—the units of observation (i.e., the number of counties) are greater than the total number of covariates used.13,14 However, the higher the number of observations, the better the statistical power of the analysis and consequently, the more robust the results.13,14
Following a previously published study,15 we computed temporally smoothed STD rates using the cases and their respective population for 2000 and 2007. Due to zero cases in some counties (especially for P & S syphilis) and for the benefit of uniformity in result interpretation, we transformed all the STD rates as shown below:
A similar transformation was done for 2007 using data from 2006, 2007, and 2008. The resulting smoothed rates were used to represent a more reliable estimate of the burden of STDs for both years (i.e., 2000 and 2007).
Equation (Uncited)Image Tools
A list of STD clinics in Texas was obtained from the HIV/STD Prevention Care Branch of the Texas Department of State Health Services. The list of Title V-, X-, and XX-funded FP clinics was obtained from the Family Planning Clinic Locator on the Texas Department of State Health Services website.16 On the basis of a recent closely related spatial analysis study on STD morbidity and services,17 we mapped 2000 and 2007 county-level smoothed rates and overlaid it with geocoded STD and FP clinics, which allowed us to identify the counties with STD service(s). A relatively small proportion of either type of clinic, less than 9%, could not be geocoded due to discrepancies between the street addresses and street maps. Consequently, manual geocoding was used to identify the counties within which the clinics were located.
Demographic and socioeconomic data obtained from the 2000 and 2007 editions of the City and County Data Book were used as control variables.18,19 We selected independent (or control) variables based on availability and results from published literature.15,20–22 However, we excluded independent variables that were highly correlated with other variables. As an example, we dropped median household income because it was significantly correlated with poverty level. The covariates used include percent black, percent Hispanic, percent American Indian, percent American Asian, log of per capita income, percent aged 15 to 24 years, commute index (i.e., percent commuting to neighboring counties for work), log of population density, log of men per 100 women, log of birth rate, log of crime rate, unemployment rate, and percent below poverty line.
We tested multicollinearity by computing the variance inflation factors using 10 as the cutoff point (Belsley et al.23).
Following methodology used in previously published spatial analyses studies on county-level STDs in Texas,15,24 we used a spatial autoregressive model to account for spatial autocorrelation (or spatial dependence). To do this, we included spatial lags (i.e., a measure of the degree of spatial dependence) as independent variables in each regression analysis. We then generated dichotomous variables for the presence of STD service(s) (equals 1 if present, 0 otherwise) for the TX counties with at least 1 STD clinic (n = 46 counties); with at least 1 FP clinic (n = 113 counties); and having at least 1 STD or FP clinic (n = 116 counties). We adopted these 3 broad categories because a proportion (about 12%) of the clinics had 1 name with 2 or more street addresses. Additionally, there was an overlap between the STD clinics and the FP clinics—over 80% of the STD clinics had similar addresses to FP clinics on the list and therefore may be colocated or the same facility. Due to the high correlation between the dichotomous variables for at least 1 FP clinic and at least 1 STD or FP clinic (correlation coefficient ≈ 0.97), only one of them (i.e., the one with the lower P value) was used in each of the final regression models.
The first hypothesis (STD service[s] impact on STD detection) was tested using the cross-sectional data for each STD and for both years (2000 and 2007). The second hypothesis (STD service[s] location and association with STD control) was tested using the relative changes between 2000 and 2007. Thus, for our first hypothesis, regardless of year (2000 or 2007), if the coefficient of the STD service(s) dichotomous variable was positive and significant, we would conclude that on average, counties with STD service(s) reported relatively higher transformed STD rates than those counties without any STD service(s). For the second hypothesis, if the coefficient of the STD service(s) dichotomous variable was negative and significant, we would conclude that, on average, counties with STD service(s) were associated with a decrease in transformed STD rates between 2000 and 2007 when compared with counties without any STD service(s).
We used a mixed log-log/linear form to reduce the problem of overdispersion in the raw data.15
The dependent variables were the transformed smoothed chlamydia, gonorrhea, and P & S syphilis rates. Independent variables included STD service(s) dichotomous variables, percent black, percent Hispanic, percent American Indian, percent American Asian, log of per capita income, percent aged 15 to 24 years, commute index (i.e., percent commuting to neighboring counties for work), log of population density, log of males per 100 females, log of birth rate, log of crime rate, unemployment rate, and percent below poverty line.15 We used a 3-equation (1 for each STD) seemingly unrelated regression estimation suggested by Zellner25 to improve efficiency. Preliminary analyses indicated that most of the changes in the covariates were not significant in the second hypothesis model. Consequently, following Kilmarx et al.,22 we used a stepwise regression procedure with a P value threshold of 0.15 to arrive at a more parsimonious model and to obtain information on partial R2s.
ArcGIS version 9.3 (ESRI, Redland, CA) was used to obtain the polygons representing counties in the state of Texas. GeoDa version 0.9.5-i26 was used to create spatial lag variables as well as perform preliminary spatial regression analyses. SAS version 9.2 (SAS Institute, Cary, NC) and STATA version 9 (StataCorp LP, College Station, TX) were used for regression results validation and diagnostics.
Examining STD Service(s) Impact on STD Detection
Results for the regression analyses used to examine the impact of STD service(s) on detection for all 3 diseases in 2000 and 2007 are presented in Tables 1 and 2, respectively. The maximum computed variance inflation factor for all the models was 4.9, which was within the limit suggested by Belsley et al.23 The estimated coefficients for most of the control variables were significant and had the expected qualitative relationships (i.e., positive or negative). The system-weighted R2s were 0.63 for the 2000 data and 0.68 for the 2007 data. This implies that 63% and 68% of the variations in the transformed STD rates were accounted for by the STD service(s) dichotomous and control variables. On the basis of the purpose and focus of this study, only the major findings on covariates are discussed. Detail discussion on the control variables can be found in previously published studies.15,21,22
For both years, the spatial lags were statistically significant (P < 0.05) in the chlamydia and gonorrhea models, but not in the syphilis model. However, for both years and all diseases, percent population black was significantly (P < 0.01) associated with transformed incidence rates. A 1% increase in percent black was associated with at least 1.4% increase, on average, in the transformed rates of all STDs when compared with percent white (Tables 1, 2). Also, for both years, percent aged 15 to 24 years were statistically significant in the chlamydia (P < 0.01) and gonorrhea (P < 0.05) models, but not in the syphilis models. A 1% increase in the percent aged 15 to 24 years was associated with a 2.6% increase in chlamydia and 1.8% increase in gonorrhea in 2000 and with a 1.7% increase in chlamydia and a 1.2% increase in gonorrhea in 2007 (Tables 1, 2).
Dichotomous variables representing the existence of STD service(s) within a county were found to be positive and significant (Tables 1, 2). For 2000, having at least 1 STD clinic in the county was associated with, on an average, a 10% (P < 0.05) increase in transformed rates of chlamydia, 24% (P < 0.01) increase in transformed rates of gonorrhea, and 20% (P < 0.01) increase in transformed P & S syphilis rates compared with having no STD clinic in the county. Having at least 1 FP clinic in the county was associated with a 9% (P < 0.05) increase in gonorrhea compared with having no FP service in the county. Finally, having at least 1 STD or FP clinic in the county was associated with an 8% (P < 0.01) increase in transformed rates of chlamydia compared with counties with no service (Table 1).
The results for the year 2007 were somewhat similar to 2000 data for chlamydia and gonorrhea, but different for P & S syphilis. None of the service dichotomous variables was significant in the P & S syphilis model. In 2007, having an STD clinic in the county was associated with a 12% (P < 0.01) increase in the transformed chlamydia rate compared with having no STD clinic within the county; 6% (P < 0.05) increase if the county had at least 1 FP clinic compared with having no FP clinic. Having an STD clinic in a county was associated with a 29% (P < 0.01) increase in the transformed gonorrhea rate compared with having no STD clinic in the county (Table 2).
Examining Association Between STD Service(s) and Changes in STD Burden Between 2000 and 2007
On the basis of the P value limit we used (0.15) in the stepwise regression, only variables with P values less than 0.15 were presented. The spatial lags were statistically significant in all STD models (Table 3). Having at least 1 service (STD or FP clinic) in the county was associated with a 4% (P < 0.10) and 8% (P < 0.01) decrease, on an average, in the transformed chlamydia and gonorrhea rates between 2000 and 2007, respectively, when compared with counties with no STD service. However, for P & S syphilis, having at least 1 STD clinic in the county was associated with an 8% (P < 0.05) decrease, on average, in the transformed P & S syphilis rate compared with having no STD clinic in the county (Table 3).
This study examined the association between the presence of STD services (as provided by STD clinics and publicly-funded Title V, X, and XX FP clinics) and detection of 3 reportable STDs (chlamydia, gonorrhea, and P & S syphilis) in the state of Texas using county-level data for 2000 and 2007. Our results indicated that except for syphilis in 2007, counties with STD and/or FP clinics were associated with at least 6% (P < 0.05) higher transformed incidence rates of each of the 3 STDs examined, suggesting that these entities play an important role in STD detection in Texas. These results are consistent with the findings reported in Washington, DC, several years ago.8 Second, our results showed that having at least 1 STD/FP clinic was associated with at least a 4% (P < 0.10) decrease in the rates of disease between 2000 and 2007. When considered in relation to the overall proportion of cases of each STD reported by STD and/or FP clinics, these estimated associations are quite substantial.
The statistically significant associations found in this study suggest that in addition to identifying STDs, STD and FP clinics also play an important positive role in their control. We are not aware of any study that examined the impact of STD/FP clinics on the detection and control of STDs at this scale (i.e., county-level); hence, comparison with previous studies is not possible. The relative impacts of STD service(s) on control is consistent with results from a recent study that showed that relatively more cases of gonorrhea and syphilis are diagnosed in STD clinics than chlamydia.3 Additionally, it was shown in a previously published study that funding for syphilis elimination activities (which was implemented through STD clinics) have a notable positive impact on early syphilis control.21
In general, the results from our spatial regression analyses were consistent with findings from previous studies.15,20,21 There were differences in the magnitude and statistical significance of the coefficients of the spatial lag between the STDs. For instance, the spatial lags that were used to account for spatial dependence were significant in the chlamydia and gonorrhea models, but were not significant in the P & S syphilis model. This is consistent with the large distribution of the diseases (chlamydia and gonorrhea) as was found in a previous study.29 Chlamydia and gonorrhea are more widespread within each county. Thus, one would expect the neighbors of each county to have similar rates which are what the spatial lag captured. Also, consistent with results obtained in a state-level analysis,21 percent aged 15 to 24 years was not significant in the P & S syphilis model. This is expected, given that the highest rates of P & S syphilis is among older age group (25–29 years) than chlamydia and gonorrhea (20–24 years).1
Our study has several limitations. All limitations associated with ecological analyses are applicable. Second, there is incomplete and/or underreporting of cases.1 Thus, differences in the rate of incomplete and underreporting may affect our results if this limitation is exclusively associated with STD/FP clinics. Additionally, the STD incidence data we used were obtained from reported cases of infection which, to a large extent, depend on medical providers' and laboratories' adherence to the testing and reporting requirements in Texas. There is inherent heterogeneity in the screening processes used in different locations. For example, different locations may have substantially different screening mandates and/or efforts. Some jurisdictions may concentrate their screening efforts on target populations while others use different criteria for screening or outreach activities. In view of this, generalization of our results is not possible.
Because of the substantial overlap between the STD and FP clinic sites, we used a crude approximation measure (i.e., the presence of STD/FP clinics, or the lack thereof) as an indication of county residents' access to STD care. A more appropriate measure that is based on the number of unique STD and FP that also accounts for the population in the county (such as, number of STD/FP clinics per 100,000 residents) may provide results that are more robust. Additionally, due to lack of data, screening and treatment rates, the existing regulatory environment as well as funding levels for the periods examined, including their changes overtime (particularly over the period covered by our analysis) were not controlled for in our analysis.
In addition, the criteria for race/ethnicity classification are not perfect. There are overlaps. However, it is difficult to assess how these limitations affected our results. Our analysis did not include the presence/absence of other sources of surveillance data such as private sector/managed care providers, hospitals (emergency room), community health centers, and military base clinics. However, to the extent that we were interested in STD and FP clinics only, the remaining sources of data were implicitly included in the referent group. Thus, the omission of all or some of the other sources of data should not substantially affect our results. Finally, because we limited the scope of this study to the state of Texas, we did not account for spatial effects from counties that are contiguous to border counties and are located in neighboring states.15
Our results suggest that STD/FP clinics played important roles in the county-level detection and control of STDs in Texas between 2000 and 2007. The results of this study highlight associations and do not establish causality between having an STD/FP clinic and improved detection and/or control of STDs at the county-level in Texas. Finer level analyses (such as city/census blocks) that controls for screening and treatment rates may provide more detailed information and should be explored in future studies. A critical examination of the current role of STD/FP clinics and their potential impact (or the lack thereof) can inform the decision-making process and planning for future STD detection and control.
Previous reports show that most high-income countries (such as the Netherlands and Australia) with lower STD burden have had universal health insurance for decades but they have continued to support their STD clinics as well.3,30 In the United States, STD clinics often serve uninsured and marginalized individuals such as men who have sex with men and racial/ethnic minorities3 who are disproportionately affected by STDs.1 Our results provide support for the suggestion that efforts should be made to save the categorical STD clinics,31 or begin discussing available options for future STD detection and control here in the United States. Future studies should consider conducting a comprehensive economic evaluation of these clinics to determine the health and economic benefits they confer on the public to justify support.
1. Centers for Disease Control and Prevention. Sexually transmitted disease surveillance, 2009. Atlanta, GA: U. S. Department of Health and Human Services, 2010.
2. Chorba T, Scholes D, BlueSpruce J, et al. Sexually transmitted diseases and managed care: An inquiry and review of issues affecting service delivery. Am J Med Qual 2004; 19:145–156.
3. Golden MR, Kerndt PR. Improving clinical operations: Can we and should we save our STD clinics? Sex Transm Dis 2010; 37:264–265.
4. Centers for Disease Control and Prevention. Sexually transmitted disease surveillance, 2006. Atlanta, GA: Centers for Disease Control and Prevention, 2007.
5. Centers for Disease Control and Prevention. Sexually Transmitted Disease Surveillance, 2007. Atlanta, GA: U. S. Department of Health and Human Services, 2008.
6. Centers for Disease Control and Prevention. Sexually Transmitted Disease Surveillance, 2008. Atlanta, GA: U. S. Department of Health and Human Services, 2009.
8. Centers for Disease Control and Prevention. Impact of closure of a sexually transmitted disease clinic on public health surveillance of sexually transmitted diseases—Washington, D.C., 1995. Morb Mortal Wkly Rep 1998; 47:1067–1069.
9. White PJ, Ward H, Cassell JA, et al. Vicious and virtuous circles in the dynamics of infectious disease and the provision of health care: Gonorrhea in Britain as an example. J Infect Dis 2005; 192:824–836.
10. Breslow L. Public health and managed care: A California perspective. Health Aff (Millwood) 1996; 15:92–99.
11. Kimball AM, Lafferty WE, Kassler WJ, et al. The impact of health care market changes on local decision making and STD care: Experience in three counties. Am J Prev Med 1997; 13:75–84.
12. Wong W. Fact Sheet: STD program capacity and preparedness in the United States: Results of a National Survey. Washington, DC: National Coalition of STD Directors, 2009.
13. Kennedy P. A Guide to Econometrics. 6th ed. Malden, MA: Blackwell Publishing, 2008.
14. Greene WH. Econometric Analysis. 4th ed. Upper Saddle River, NJ: Prentice-Hall, 2003.
15. Owusu-Edusei K, Chesson HW. Using spatial regression methods to examine the association between county-level racial/ethnic composition and reported cases of chlamydia and gonorrhea: An illustration with data from the state of Texas. Sex Transm Dis 2009; 36:657–664.
17. Owusu-Edusei K Jr, Doshi SR. Use of geographic information systems (GIS) tools to assess potential spatial gaps in sexually transmitted diseases (STD) service supply and demand in Texas: 2007 data. In: National STD conference; 2010; Atlanta, GA.
20. Semaan S, Sternberg M, Zaidi A, et al. Social capital and rates of gonorrhea and syphilis in the United States: Spatial regression analyses of state-level associations. Soc Sci Med 2007; 64:2324–2341.
21. Chesson H, Owusu-Edusei K Jr. Examining the impact of federally-funded syphilis elimination activities in the USA. Soc Sci Med 2008; 67:2059–2062.
22. Kilmarx PH, Zaidi AA, Thomas JC, et al. Sociodemographic factors and the variation in syphilis rates among US counties, 1984 through 1993: An ecological analysis. Am J Public Health 1997; 87:1937–1943.
23. Belsley DA, Kuh E, Welsch RE. Regression Diagnostics. New York, NY: John Wiley & Sons, 1980.
24. Owusu-Edusei K Jr, Owens CJ. Monitoring county-level chlamydia incidence in Texas, 2004–2005: Application of empirical Bayesian smoothing and Exploratory Spatial Data Analysis (ESDA) methods. Int J Health Geogr 2009; 8:12.
25. Zellner A. An efficient method for estimating seemingly unrelated regressions and tests of aggregation bias. J Am Stat Assoc 1962; 28:977–992.
26. Anselin L, Syabri I, Kho Y. GeoDa: An introduction to spatial data analysis. Geogr Anal 2006; 38:5–22.
27. Huber PJ. The behavior of maximum likelihood estimates under nonstandard conditions. In: Proceedings of the Fifth Berkeley Symposium on Mathematical Statistics and Probability. Berkeley, CA: University of California Press, 1967:221–233.
28. White H. Maximum likelihood estimation of misspecified models. Econometrica 1982; 50:1–25.
29. Chesson H, Leichliter JS, Aral SO. The geographic distribution of three STDs in the United States. In: 18th Meeting of the Society for International STD Research; 2009; London, United Kingdom.
31. Rietmeijer CA, Mettenbrink C. Why we should save our STD clinics. Sex Transm Dis 2010; 37:591.