Although malaria has been eliminated from the US since the early 1950s, about 1,500 cases of malaria are reported in the US every year. Few studies have comprehensively quantified the disease and economic burden of malaria in the US, and the domestic impact has not been well described. Epidemiological study of malaria using hospital data has rarely been explored, and can supplement the findings from current nationwide surveillance system data to elucidate the malaria disease and economic burden. Identification of predictors of increased costs and severe malaria may be useful in developing targeted interventions.

Discharge records for malaria-related hospitalizations from the 2000-2012 Nationwide Inpatient Sample (NIS) were examined. In chapter 1, the disease burden was quantified by examining the frequencies and population rates of malaria-related hospitalization by demographics, infecting species, clinical, financial, institutional, geographic, and seasonal characteristics. Trends over the period in the counts of malaria cases by different patient characteristics were assessed using negative binomial regression. In chapter 2, the economic burden was quantified by examining the mean and total hospital days, hospital charges, and hospital costs for malaria-related hospitalizations by demographics, species, clinical, financial, geographic, and institutional characteristics. Trends and potential predictors for hospital charges and costs were identified using linear regression. Trends and potential predictors for length of stay were identified using negative binomial regression. In chapter 3, a subset analysis on severe malaria cases was conducted, and the frequencies and population rates of severe malaria-related hospitalizations by demographics, species, clinical, financial, geographic, and institutional characteristics were examined. Trends in the rates of severe malaria hospitalizations over the study period were assessed by negative binomial regression. Multiple logistic regression models were used to identify potential predictors for severe disease (cerebral malaria, acute respiratory distress [ARDS], severe anemia, renal failure, or jaundice) and death among those with malaria-related hospitalizations.

From 2000-2012, there were an estimated 19,189 malaria-related hospitalizations (4.95 per 1 million population) in the US, including 147 in-hospital deaths and 3,888 severe malaria cases. On average, malaria patients were hospitalized for 4.39 days with charges of $25,116. From 2000-2012, malaria-related hospitalizations accounted for 84,213 hospital days, $151,825,389 in total hospital costs, and $470,102,584 in total charges. The most frequent malaria complication was renal failure (40.5%), followed by severe anemia (34.6%), ARDS (19.9%), cerebral malaria (18.7%), and jaundice (15.4%). P. falciparum accounted for the majority of malaria- and severe malaria-related hospitalizations. Malaria-related hospitalizations occurred disproportionately among patients who were male, Black, or aged 25-44 years. After controlling for potential confounders, older age was associated with higher odds of severe malaria, ARDS, severe anemia, and renal failure. Males had higher odds of developing renal failure and jaundice, while females had higher odds of developing severe anemia and ARDS. HIV infection was associated with increased odds of severe malaria. Having severe malaria was associated with a longer length of stay. Older age, severe malaria, HIV infection, and longer lengths of stay were associated with higher charges and costs. Mean charges increased significantly over the study period. Patients with a malaria diagnosis were more often hospitalized in the Middle Atlantic and South Atlantic census divisions, urban teaching, private not-for-profit, and large bed size hospitals. Malaria patients who were self-payers or had Medicaid were at increased odds of having renal failure, compared to those with Medicare.

This dissertation demonstrates that malaria imposes a substantial disease and economic burden in the US, and underscores the need for improved primary and secondary prevention measures, especially among high-risk groups.

Currently, the most effective method to reduce further transmission and optimize the well-being of those with HIV is to achieve and maintain viral suppression. The HIV Care Continuum is commonly used as a framework to monitor progress towards this end goal. As such, it is important to understand the dynamics of progressing through each stage of the continuum in order to implement effective targeted interventions.

This study was, to the best of our knowledge, the first to evaluate a cohort of HIV-positive individuals through the entire course of the HIV care continuum: from diagnosis through viral suppression. Thus, factors associated with overall viral suppression and progression to each intermediate continuum stage could be evaluated within the same population, allowing for a more standardized comparison of the relevant determinants between the stages. This type of comparison was lacking in the current literature, in which information about the continuum stages must be derived from different studies based on different study populations.

As such, our study (Chapter 2) was able to evaluate associations at each continuum stage and provide insight into the consistency of these associations throughout the continuum. We observed that certain factors associated with overall viral suppression were not always significant at every stage. Conversely, some factors that were associated with a certain intermediate stage were not associated with overall viral suppression. For example, race/ethnicity was significantly associated with viral suppression among everyone diagnosed. However, there was no difference between the ethnic groups with regards to linkage and retention. Therefore, within this population, interventions and policies aimed towards increasing linkage/retention in these ethnic groups in hopes of improving viral suppression rate would be misdirected and ineffectual. Our results demonstrate the importance of understanding the continuum as a whole in any given population to best direct public health resources.

Another aspect relevant to the appropriateness of the HIV care continuum is the designated definitions of the intermediate stages. The intermediate stage with the most variation in definition is retention in care. For any given population, any retention measure used must be significantly associated with increased viral suppression (among the retained). However, another perhaps more clinically practical perspective in deciding the most appropriate retention measure definition would be determining the measure that best predicts viral suppression. This study (Chapter 3) was also the first to evaluate eight different measures of retention, with measures involving kept visits, missed visits, and HIV laboratory tests, among a single population. Further, our study was among the few to apply a non-traditional statistical-learning method (k-folds cross validation) to estimate the accuracy of the retention measures in predicting viral suppression among new patients. The results demonstrated that although no gold standard retention measure may exist, each retention measure evaluated was significantly associated with increased viral suppression as well as positive predictor of viral suppression. However, the difference between the ranking of strength of association (odd ratio) and predictive accuracy (from cross-validation) indicates that traditional statistical methods alone may not best determine prognostic ability of the measures. Methods such as those in statistical learning can be utilized in evaluating the most appropriate measure for each population.

Aside from viral suppression, another goal of HIV counseling and care is to reduce the sexual behaviors that promote transmission. The literature is mixed with regards to the direction of change in high-risk sexual behaviors at different points after HIV diagnosis: 1) following linkage to care; and 2) following initial viral suppression. Our analyses (Chapter 4) also evaluated the direction of any change in sexual behavior at these two time points using STD incidence as a proxy. The results provided more support for an attenuation in high-risk behavior possibly related to treatment adherence than for an increase due to “HAART complacency”. In addition, the findings were more consistent with a reduction in high-risk behaviors observed following HIV diagnosis. Ethnicity was also associated with reduction in STD incidence. Further sub-analysis based on the association with ethnicity revealed a subgroup of clients that were at potentially high-risk for HIV transmission (no reduction in high-risk sexual behavior, concurrent STD infection, and less likely to be virally suppressed). Future studies and programs should focus on identifying these high-risk subgroups in addition to the broader analyses on sexual behavior change.

Syphilis is a curable disease, yet 10 million persons worldwide are infected each year, of which 1.4 million are pregnant women. Syphilis is an infection caused by the spirochete Treponema pallidum and frequently has atypical presentations that may be difficult to differentiate from other sexually transmitted infections (STIs) making effective diagnostics essential to the identification of infection. Additionally, the manifestations of the disease vary depending on the duration of infection and time of presentation. There are four stages-primary, secondary, latent (early and late), and tertiary.

Syphilis screening and treatment programs that utilize laboratory-based testing have been hampered by limited laboratory access, long turn-around time for results, and loss to follow-up of syphilis infected individuals. When syphilis diagnostic testing involves multiple tests performed off-site, only a proportion of infected individuals receive treatment and continued transmission occurs.

A rapid test is a simple point-of-care test that can be used in a variety of settings and provides a result to guide clinical management during the time of the initial consultation (ideally within 30 minutes or less). The advent of rapid point-of-care tests for syphilis has reduced barriers and allowed for new health systems approaches to syphilis prevention, including same day testing and treatment. Dual rapid tests that have multiple analytes for the detection of antibodies for both HIV and syphilis infections are now available.

Our first study addressed preferences for dual HIV/syphilis tests through a conjoint survey analysis. Conjoint analysis is a method for systematically estimating consumer preferences across discrete attributes. Conjoint analysis has been used extensively in marketing research and measures the value that consumers place on each feature of a product. Conjoint analysis has recently begun to be applied in health research. We recruited 298 men and women 18 years of age and over seeking testing or care at GHESKIO (Haitian Study Group for Kaposi’s sarcoma and Opportunistic Infections) clinics. We created 8 hypothetical dual test profiles varying across six dichotomous attributes: cost (free versus $4), accuracy (no false positive versus false positive), time-to-result (20 minutes versus 1 week), blood draw method (finger prick versus venipuncture), number of draws (1 versus 2), and test type (rapid versus laboratory). Cost (free vs. $4; impact score=27.2, SD=36.6, p<.0001) had the highest impact on likelihood of testing, followed by number of blood draws (1 versus 2; impact score=17.5, SD=29.8, p<.0001), blood draw method (fingerprick versus venipuncture; impact score=9.7, SD=26.5, p<.0001), test type (rapid versus laboratory; impact score= -4.5, SD=21.9 , P=.0005), and time-to-result (20 minutes versus 1 week; impact score=3.6, SD=25.6, p=.0139). This analysis showed that implementation of a low cost dual rapid test in the laboratory for HIV and syphilis could be used to improve screening uptake and accessibility to accelerate time to treatment.

Our second study was a field study of a dual HIV and syphilis rapid test in Port-au-Prince, Haiti. Field studies using whole blood fingerprick specimens are essential to understand how the test will perform in real-world settings. We found that for the HIV test component, sensitivity and specificity were 99.2% (95% CI: 95.8%, 100%) and 97.0% (95% CI: 93.2%, 99.0%), respectively; and for the syphilis component were 96.5% (95% CI: 91.2%, 99.0%) and 90.8% (95% CI: 85.7%, 94.6%), respectively. This test performed well and could be considered for wider use to increase rates of HIV and syphilis screening.

Our final analysis was an epidemiological bias analysis using estimates of potential misclassification by use of an imperfect gold standard test. We used regression calibration methods to assess the effect of reference standard misclassification bias using the syphilis component of the dual HIV and syphilis rapid test as an example. We concluded by making recommendations for reporting diagnostic performance ranges instead of point estimates when using an imperfect reference test.

In conclusion, dual HIV and syphilis testing is a preferred screening modality that can be used to streamline HIV and syphilis prevention and case finding. These tests perform well in both laboratory and point-of-care settings and should be considered for wider implementation. Use of regression calibration methodology can provide accurate information on the potential performance characteristics of these tests when a perfect gold standard test is unavailable.

Tuberculosis is a global public health issue with more than 2 billion people infected worldwide. It is also a serious public health concern within the United States with 9,557 cases of active disease diagnosed in 2015 alone. In the U.S., specific sub-groups, such as foreign-born persons, persons with diabetes or persons living with HIV or other immunocompromising conditions are known to be at higher risk of TB disease. Among foreign-born residents in the U.S., persons born in high-morbidity countries are known to be at even higher risk of developing the disease. Yet, TB disease incidence rates by country of birth are not reported at the local, state or national level despite these large, known differences in risk by country of birth. This is part due to the complications of using country-of-specific population estimates and technical challenges of using standard distributions to model a communicable disease. This thesis aims to call attention to this notable gap and, in part, to fill it.

Data on 5,447 diagnosed TB cases from the Los Angeles County Department of Public Health TB Control Program were combined with stratified population estimates available from the Public Use Microdata Survey to calculate incidence rate of TB disease for the years 2005 through 2011, stratifying by country of birth and other demographic factors. Bayesian models were used to account for the uncertainty in the number of diagnoses and the population estimates. Extending these models into spatial analysis required the use of a hierarchical Bayesian model. Prediction models were constructed using bootstrap backward elimination and stochastic variable selection.

We estimated that the unadjusted incidence rate among persons born in the Philippines was 44.3 per 100,000 person-years and among persons born in Vietnam 38.7 per 100,000 person-years in comparison to 2.3 per 100,000 for U.S.-born persons. In spatial analysis, TB disease incidence was found to be spatially heterogeneous within Los Angeles County and remained so within high-risk countries of birth and when accounting for age, sex and years in residence. In prediction modeling, we found the addition of PUMA-level ecological variables did not improve the prediction of TB disease incidence beyond models using age, sex, country of birth and years in residence. With these three analytical approaches–non-spatial, spatial and prediction–we confirmed that TB disease incidence rates varied markedly by country of birth and showed that issues arising from the technical challenges of dependent outcomes, sparse data and uncertainty in population estimates can be ameliorated.

Background: Cysticercosis, an infection caused by the larval stage of the pork tapeworm Taenia solium, has become increasingly important both in the United States and globally in recent decades. Neurocysticercosis is one of the most common parasitic diseases of the human nervous system and a leading cause of acquired epilepsy worldwide. Despite its potential impact, there is a lack of comprehensive information on cysticercosis infection in the US. This study was designed to estimate the impact of cysticercosis nationwide, in terms of the number of people hospitalized and the economic burden of hospitalizations, as well as explore potential risk factors for cysticercosis in-hospital fatality. Methods: The Nationwide Inpatient Sample (NIS), representing a 20% sample of annual hospital discharge records in the US, was used for analysis from 1998-2009. National estimates of cysticercosis-related hospitalizations and rates, patient and hospital stay characteristics, and hospitalization-related charges were calculated using NIS sample weights and US Census Bureau data. Risk factors for fatality were evaluated using bivariate and multivariate logistic regression. Results: There were 28,565 cysticercosis-related hospitalizations nationwide estimated during the study period, representing a hospitalization rate of 8.16 per million population. The highest proportion of cases were reported in males, Hispanic populations, hospitals in the West and among patients aged 18-44, consistent with previous studies. There were 364 total in-hospital deaths estimated, representing an overall case-fatality rate of 1.28% and a nationwide in-hospital mortality rate of 0.1 deaths per million population. National estimates of associated charges among cysticercosis-related hospitalizations amounted to approximately $996 million. Male sex, hospital location in the West, shunting procedures and diagnoses of obstructive hydrocephalus were associated with in-hospital fatality in bivariate regression analysis (OR: 1.18, 2.36, 1.79 and 2.85 respectively). Discussion: National estimates of cases, deaths and associated charges reinforce the importance of cysticercosis in the US. Despite improvements in management strategies, the burden of cysticercosis-related hospitalizations remains substantial, indicating the need for significant progress towards reducing infection nationwide. Increased surveillance and identification of cases can serve to better define the burden of disease, potentially allowing for targeted prevention and control campaigns in populations at greatest risk.

Purpose

Methods

Results

Conclusions