Population-Level Predictors of STI Rate Changes in Missouri From 2008-2017

Sexually transmitted infection rates continue to increase across the US, further developing health disparities and economic burdens of disease, especially as migration occurs. In this study, we assessed the relationships among STI rates and migration at the county-level in Missouri from 2008–2017. Methods Two data sources were used: STI rates of chlamydia, gonorrhea, syphilis, HIV reported to Missouri DHSS and ACS 1-year county population estimates. Linear regression models and ANOVA tests were conducted in SPSS for each STI from year-to-year and 2008–2017. Covariates included in the analyzes were county-level income, employment rate, race, ethnicity, age, and percent poverty. Further, Akaike Information Criterion tests were performed to indicate the best predictor models and averaged standardized beta values.


Abstract
Background Sexually transmitted infection rates continue to increase across the US, further developing health disparities and economic burdens of disease, especially as migration occurs. In this study, we assessed the relationships among STI rates and migration at the county-level in Missouri from 2008-2017.

Methods
Two data sources were used: STI rates of chlamydia, gonorrhea, syphilis, HIV reported to Missouri DHSS and ACS 1-year county population estimates. Linear regression models and ANOVA tests were conducted in SPSS for each STI from year-to-year and 2008-2017. Covariates included in the analyzes were countylevel income, employment rate, race, ethnicity, age, and percent poverty. Further, Akaike Information Criterion tests were performed to indicate the best predictor models and averaged standardized beta values.

Results
Signi cant relationships among STI rates and population growth were identi ed. Chlamydia, syphilis, and HIV were positively associated with population growth from 2008-2017 (β = 0.15; β = 0.01; β = 0.05, respectively). Gonorrhea was negative associated with population growth (β= -0.02) but positively associated with unemployment rates (β = 0.01) highlighting the need to address population growth, as well as other variables in a population.

Conclusions
There seems to be a positive relationship among population change and rates of STIs. As populations change, rates of STIs change. Moving forward, quantitative work should be conducted in various states and the nation to understand this relationship in different contexts. Qualitative studies should assess individual county health departments, identifying strengths and weaknesses, implementation of community health improvement plans. Lastly, public policy should be implemented to buffer the impact of migration on health outcomes.

Background
Sexually transmitted infections (STI) continue to pose a complex, signi cant, and constantly evolving public health concern in the United States. 1 In the Midwest from 2016-2017, the chlamydia rate increased 5.6%, the gonorrhea rate 19.5%, and the syphilis rate 8.8%. 1,2,3 This increase is attributed to more widespread screening, reporting, and cases altogether. 1 Other factors include more sensitive and accurate diagnostic tests, like nucleic acid ampli cation tests (NAATs). 1 A small percentage increase indicates a signi cant amount of new cases and subsequent healthcare costs as more are being screened and that impact STI rates. Macro effects on disease and STI transmission are often ignored and instead the focus is put on individual risk factors. 7 Social epidemiology de nes three levels to the spread of STIs throughout society: 1) individual components, 2) social components, and 3) structural components. 14 Individual components involve biological susceptibility and risky behaviors. Social components involve networks, communities, and how disease diffuses across populations. 14 The structural components are grouped into 1) cultural context, 2) social networks, 3) neighborhood effects, and 4) social capital. Social epidemiology can be used as a lens to understand the impact of population change and STI rates within a state. Social epidemiology lacks the insight of an individualized point of view on STIs, but the majority of data focuses on individual trends. 7 It instead provides insight into large-scale trends affecting many with STIs, where data may be less extensive but equally compelling due to the vast amount of individuals it can then reach. 7 Given the social complexity and nancial cost of STIs, it becomes critical to monitor changing rates of these infections not only throughout static populations but also as moving populations change proportions of age, gender, ethnicity, and sexuality. With these changing dynamics, estimated risks of STI transmission uctuate as different groups have different risk factors surrounding STIs. 7 As populations grow and move, the incidence of STIs is expected to grow, showing the importance of population-level predictors alongside individual interventions. Moving, changing populations face problems of isolation and loneliness after relocating. 12 It has been observed that moving populations typically experience a higher risk of contracting STIs. 15 Characteristics of a moving population include the social disruption that accompanies geographic relocation and a lack of access to health resources. 15 STIs have remained persistent within society, in part, due to failure to contact trace all sexual partners of STI patients. 9 This is possibly due to geographic or networking barriers, consistent with a moving population. Young people and those seeking secondary education most often have to relocate, particularly from rural to urban areas. 16 In the past, the urban population of the world has increased much more quickly than those of the rural population. 12 This is expected to remain the trend into the future as the United Nations expects the world population to increase to 9.1 billion by 2100 with urban population continuing to increase in population size, and rural populations moving more towards urban areas. 12 Urban populations were considered the only "high-risk" areas for STI spread in the past, but new pockets of widespread infection in less densely populated areas have begun to form. 8 In this study, we analyze the relationship among common STIs and population change at the county level in the state of Missouri from 2008-2017.

Methods
Data were accessed from the Missouri Information for Community Assessment (MICA) by the Missouri Department of Health and Senior Services (DHSS) regarding county-level data on age between 15-44, sex, race, and ethnicity every year from 2008-2017. 17 Non-census year data comes in conjunction with the Federal State Cooperative Program for Population Estimates and the National Center for Health Statistics, alongside the US Census Bureau. Census year data, 2010 only, were taken directly from the US Census Bureau. Incidence of the three STIs: chlamydia, gonorrhea, and syphilis was compiled from the DHSS for the same time. 18 Chlamydia, gonorrhea, and syphilis are all infections that, when detected as positive in a hospital or clinic, must be reported to the county health department that the positive test originated from.
These rates are then reported to the Missouri state health department for monthly and yearly reports. 18 Incidence and prevalence of HIV was obtained through Geographic Information System programming by the Centers for Disease Control and Prevention from the years 2008-2016. 19  Using population change as the independent variable, unadjusted multi-variate linear regression models were conducted with HIV prevalence and incidence of HIV, syphilis, gonorrhea, and chlamydia as dependent variable year-to-year and from 2008-2017 in SPSS (version 24).
Further statistical analysis was conducted in the R project in which Akaike Information Criterion (AIC) tests were conducted to determine if population growth was contained within the best change predictor models for each STI over the period of 2008 to 2017. The Akaike information criterion is a statistical test used to nd the best change predictor models in datasets when multiple possible variables are being examined. Use of the AIC as a model t test has been used previously in research surrounding STIs propelling it to be selected for further analysis in this study in order to nd the best t predictor model for the change in STI rates over the 10-year timeframe. 21 These variable combinations are considered signi cant and equally quali ed predictor models if the ΔAIC, or change in Akaike Information Criterion, value is below two. Of those found to be similarly quali ed change predictor models, the weights, or w i , and R 2 values can then be further calculated to predict the likelihood of the signi cant models being classi ed as the best change predictor model in the dataset. 22 All models predicted to be of best t gave similar w i and R 2 values, showing that each model can be a good predictor in the future change of STIs.
Note that a small alteration was made to the analysis with HIV being reported as total cases for the years of 2008 and 2017, instead of the two distinctions of incidence and prevalence to allow for a larger, more signi cant sample size Results Table 1 presents the percent change of study variables from 2008 to 2017, including: race/ethnicity, average household income, average unemployment rate, and population size. Most counties showed a negative percent change in white individuals and average unemployment rates and a positive percent change in African American individuals and Latinx individuals. Table 2 shows the percent change in STIs from 2008 to 2017. Most counties showed considerable percent increases in chlamydia, gonorrhea, and syphilis. The counties show mixed positive and negative percent changes regarding HIV, possibly due to the limited sample sizes. Table 3 shows the results of the linear regression among STIs and population change between 2008-2017. The association between population change and chlamydia (β=0.60, p<0.05), gonorrhea (β=0.50, p<0.05) and syphilis (β=0.2, p<0.05) showed a signi cant positive association. HIV diagnoses from 2015-2016 was positively associated with population change (β=0.30, p<0.05). HIV prevalence showed a similar positive association (β =0.03, p<0.05). As populations increase or decrease in counties in Missouri, rates of STIs increase and decrease too. Table 4 displays the results from the Akaike information criterion test (AIC) with the signi cant change predictors bolded for each STI. Table 5 further shows the model average beta values which can be interpreted to show the future change in STIs per unit of the variables included in the models. Population growth showed a positive value in chlamydia (β=0.15), syphilis (β=0.01), and HIV (β=0.05) and a negative value in gonorrhea (β= -0.03).

Conclusions
This study identi ed several relationships between common STI rates and population change across Missouri at the county level. A positive relationship exists between population change and STI rates. This study also shows that population growth and size are as integral to predicting the change in STIs as other structural-level functions, like income and unemployment which may already be considered when evaluating future STI rates and change. As populations change, rates of STIs change.
Population growth and total HIV cases showed a positive relationship through model average standardized beta values (β=0.05205). If we extrapolate these results to predict future cases, we would expected 52 newly-diagnosed or existing-when-moving-into-the-county cases of HIV if the population of a county grew by 1,000 people. The lifetime cost of a single case of HIV is estimated as $304,500 in medical expenses. 9 An increase in 52 new cases of HIV would cost a county $15.83 million more in healthcare expenditure.
Population growth was positively associated with chlamydia such that an increase of 152 cases may occur per additional 1,000 people (β=0.15206). This adds $29,550 total in additional costs, if the cases are treated successfully after initial infection. 9 Costs increase to $3.5 total million when considering sequalae costs in asymptomatic cases. A large cost disparity exists between male and female cases, highlighting the need to intervene on those infected before chlamydia can spread in a county. 9 Population growth was positively associated with syphilis such that an increase of 7 cases may occur per additional 1,000 people. This adds $4.963 in additional costs per 1,000 people for treatment of syphilis (β=0.00664). 9 Costs increase signi cantly and vary in re ection to the syphilis progression. The cheapest option after syphilis contraction, is immediate treatment before possible disease progression.
Population growth was negatively associated with gonorrhea such that a decrease in 27 cases may occur per additional 1,000 people (β=-0.02705). Interestingly, unemployment was positively associated with gonorrhea such that an increase in 12 cases may occur for 1,000 additional unemployed individuals (β=0.01224). This adds $2,598 in direct medical care. Sequalae costs are estimated near $3.5 million with a large cost disparity existing between men and women. 9 This cost should be investigated further with the emergence of multi-drug resistant gonorrhea. 10 This difference in conclusions between variables highlights the need to consider multiple variables when evaluating the future of STIs. There is no single variable that can accurately predict the future change in STIs without fail. Population growth and size as well as unemployment and income must all be considered when predicting the future of STI rates. Table 1 shows the percent change in STI rates from 2008 to 2017 while controlling for population change. The change is predominantly positive, meaning that even as the population is increasing, STIs are increasing at faster rates than the population. Generally larger counties, like Jackson or St. Louis county, displayed smaller percent changes. For example, the most prevalent STI, chlamydia, had a 0.15% and 20.13% increase in Jackson and St. Louis county respectively. Smaller counties displayed signi cantly larger percent increases. For example, chlamydia cases increased 246.54% and 142.05% in Ste. Genevieve and Stone county, respectively. This shows that STIs are a problem for both small and large counties and are growing at disproportionately high rates in some smaller, more rural counties.
We continue to see populations moving for various reasons, commonly pursuing higher education in younger aged individuals. 15 Individuals who are moving from one place to the other may lack social support and consistent healthcare during and after the move. 14 Public health professionals should consider developing interventions within vulnerable moving populations.
Based on these data, we suggest the following action steps. First, public health researchers should assess how population-level changes may be related to other communicable diseases in Missouri. Second, a yearly analysis should be conducted by the Missouri Department of Health and Senior Services to understand how population change may be impacting communicable diseases in Missouri and recommendations for clinical interventions should be developed and implemented by delving into the qualitative side of the data. Lastly, clinicians and public health practitioners in areas with growing, changing populations need to screen more patients for STIs, especially those who have recently moved. On the national level, a similar analysis should be conducted to evaluate the effects of population growth, among other variables, on the nation's STI rates.
The data for this study does not allow assessment of immigration or emigration or differences due to births/deaths, showing a limitation. Population change is a complex variable. Individuals are likely moving within the state. Additionally, rural counties could be losing population due to fewer births. These population-level data do not measure within state population change. In the future, this assessment should be conducting at the national level, while also accounting for and quantifying immigration and emigration within overall population change. This is a population-level study and results are generalizable only to the state of Missouri. The two most populous areas border other states: St Louis bordering Illinois and Kansas City bordering Kansas, making it a unique environment to study. Additionally, there may be other external variables that in uence the relationships that were not accounted for in the methodology. This further shows the need for this analysis to be produced at the national level.
This study uses multiple data sets from the state of Missouri. By integrating many datasets, we are able to study how changing populations may impact health outcomes. This allows for a socioeconomic understanding of health outcomes. If healthcare and public health professionals can quantify how populations will increase and change with population change, precision public health can create targeted interventions. This may take place through evaluation and integration of this study or similar studies into Community Health Improvement Plans (CHIPs) which then develops to include intervention plans based off the ndings.
This work follows the social epidemiology network, focused around the social capital subset and some neighborhood effects. Direct neighborhood effects from social isolation of a moving population may explain increasing rates of STIs in growing communities with social ties disrupted. Indirect neighborhood effects of increased unemployment may explain the further movement of populations in search of work, with STIs spreading alongside. More research should be done on these possible implications for social theory. Authors of this manuscript plan to conduct further analyzes using additional covariate data to test the relationships among social capital, neighbor effects, and communicable disease. More interventions and research must also be conducted on policies, Community Health Improvement Plans (CHIPs), laws, and economic factors involved in potential prevention methods for STIs. Population-wide plans should be developed and implemented to bene t the community and reduce the health and economic burden of STIs. STIs continue to pose a growing problem across the United States, Midwest region, and the state of Missouri. Prevention and response are of utmost importance as HIV continues to spread and multi-drug resistance in gonorrhea becomes more apparent in the population. Population-level predictors are imperative to understanding STI rate changes and predicting when they may occur so that intervention may be put into place. As populations change in Missouri, due to movement, unemployment rates, or other variables, STIs change with them. Health care and public health providers should respond to moving populations and individuals through screening for STIs and implementing preventative measures. Missouri law and policy makers should develop legislation to better support moving populations and those facing unemployment in order to limit the spread of STIs. Further, legislation should be drafted to prevent unemployment rates from increasing and making preventative measures, like condoms and sexual education, more widespread. This way, the health and wealth of Missourians and United States residents may be bettered.

Declarations
Ethics Approval: This study was approved through the University of Missouri -Kansas City under IRB project number 2015743.
Consent: The author's give them manuscript consent to be published. There was not needed consent from other individuals throughout this process.
Availability of data and materials: The datasets generated or analyzed during this study are available from the corresponding author on reasonable request.
Competing Interests: The authors declare that they have no competing interests.
Funding: The was no funding utilized in this study.
Authors contributions: EV analyzed linear regressions and was a major contributor in writing the manuscript. JL was a major contributor in writing the manuscript. AR was a major contributor to running and analyzing the AIC statistics.    *w i and R 2 were calculated for only those models considered to be of best t Where AIC is Akaike Information Criterion value, ΔAIC is the change in Akaike Information Criterion value, W i is the Akaike weights value, and R 2 is the correlational coe cient.