Leveraging Routinely Collected Program Data to Inform Extrapolated Size Estimates for Key Populations in Namibia: Small Area Estimation Study

BACKGROUND

Estimating the size of key populations, including female sex workers (FSW) and men who have sex with men (MSM), can inform planning and resource allocation for HIV programs at local and national levels. In geographic areas where direct population size estimates (PSEs) for key populations have not been collected, small area estimation (SAE) can help fill in gaps using supplemental data sources known as auxiliary data. However, routinely collected program data have not historically been used as auxiliary data to generate subnational estimates for key populations, including in Namibia.

OBJECTIVE

To systematically generate regional size estimates for FSW and MSM in Namibia, we used a consensus-informed estimation approach with local stakeholders that included the integration of routinely collected HIV program data provided by key populations' HIV service providers.

METHODS

We used quarterly program data reported by key population implementing partners, including counts of the number of individuals accessing HIV services over time, to weight existing PSEs collected through bio-behavioral surveys using a Bayesian triangulation approach. SAEs were generated through simple imputation, stratified imputation, and multivariable Poisson regression models. We selected final estimates using an iterative qualitative ranking process with local key population implementing partners.

RESULTS

Extrapolated national estimates for FSW ranged from 4777 to 13,148 across Namibia, comprising 1.5% to 3.6% of female individuals aged between 15 and 49 years. For MSM, estimates ranged from 4611 to 10,171, comprising 0.7% to 1.5% of male individuals aged between 15 and 49 years. After the inclusion of program data as priors, the estimated proportion of FSW derived from simple imputation increased from 1.9% to 2.8%, and the proportion of MSM decreased from 1.5% to 0.75%. When stratified imputation was implemented using HIV prevalence to inform strata, the inclusion of program data increased the proportion of FSW from 2.6% to 4.0% in regions with high prevalence and decreased the proportion from 1.4% to 1.2% in regions with low prevalence. When population density was used to inform strata, the inclusion of program data also increased the proportion of FSW in high-density regions (from 1.1% to 3.4%) and decreased the proportion of MSM in all regions.

CONCLUSIONS

Using SAE approaches, we combined epidemiologic and program data to generate subnational size estimates for key populations in Namibia. Overall, estimates were highly sensitive to the inclusion of program data. Program data represent a supplemental source of information that can be used to align PSEs with real-world HIV programs, particularly in regions where population-based data collection methods are challenging to implement. Future work is needed to determine how best to include and validate program data in target settings and in key population size estimation studies, ultimately bridging research with practice to support a more comprehensive HIV response.

Exploring the use of experimental small area estimates to examine the relationship between individual-level and area-level community belonging and self-rated health

Background

Small area estimation refers to statistical modelling procedures that leverage information or "borrow strength" from other sources or variables. This is done to enhance the reliability of estimates of characteristics or outcomes for areas that do not contain sufficient sample sizes to provide disaggregated estimates of adequate precision and reliability. There is growing interest in secondary research applications for small area estimates (SAEs). However, it is crucial to assess the analytic value of these estimates when used as proxies for individual-level characteristics or as distinct measures that offer insights at the area level. This study assessed novel area-level community belonging measures derived using small area estimation and examined associations with individual-level measures of community belonging and self-rated health.

Data and methods

SAEs of community belonging within census tracts produced from the 2016-2019 cycles of the Canadian Community Health Survey (CCHS) were merged with respondent data from the 2020 CCHS. Multinomial logistic regression models were run between area-level SAEs, individual-level sense of community belonging, and self-rated health on the study sample of people aged 18 years and older.

Results

Area-level community belonging was associated with individual-level community belonging, even after adjusting for individual-level sociodemographic characteristics, despite limited agreement between individual- and area-level measures. Living in a neighbourhood with low community belonging was associated with higher odds of reporting being in fair or poor health, versus being in very good or excellent health (odds ratio: 1.53; 95% confidence interval: 1.22, 1.91), even after adjusting for other factors such as individual-level sense of community belonging, which was also associated with self-rated health.

Interpretation

Area-level and individual-level sense of community belonging were independently associated with self-rated health. The novel SAEs of community belonging can be used as distinct measures of neighbourhood-level community belonging and should be understood as complementary to, rather than proxies for, individual-level measures of community belonging.

Trends in chronic childhood undernutrition in Bangladesh for small domains

Chronic childhood undernutrition, known as stunting, is an important population health problem with short- and long-term adverse outcomes. Bangladesh has made strides to reduce chronic childhood undernutrition, yet progress is falling short of the 2030 Sustainable Development Goals targets. This study estimates trends in age-specific chronic childhood undernutrition in Bangladesh's 64 districts during 1997-2018, using underlying direct estimates extracted from seven Demographic and Health Surveys in the development of small area time-series models. These models combine cross-sectional, temporal, and spatial data to predict in all districts in both survey and non-survey years. Nationally, there has been a steep decline in stunting from about three in five to one in three children. However, our results highlight significant inequalities in chronic undernutrition, with several districts experiencing less pronounced declines. These differences are more nuanced at the district-by-age level, with only districts in more socio-economically advantaged areas of Bangladesh consistently reporting declines in stunting across all age groups.

Trend estimation of sub-national level daily smoking prevalence by age and sex in Australia

INTRODUCTION

Despite that the smoking prevalence has considerably declined in Australia after successful public health strategies over many decades, smoking is still the leading cause of preventable diseases and death in Australia. These declines have not occurred consistently across all geographical-demographic domains. In order to provide an evidence base for monitoring the trend towards the goal of reducing smoking across all domains in Australia, this study aims to estimate trends of smoking prevalence for small domains cross-classified by seven age groups (18-24, 25-29, 30-39, 40-49, 50-59, 60-69, and ≥70 years), two genders, and eight states and territories over twenty years (2001-2021).

METHODS

Direct estimates of smoking prevalence for the target small domains were calculated from the micro-data of the Australian National Health Surveys conducted in seven rounds during 2001-2021. The obtained direct estimates were then used as input for developing time-series models expressed in a hierarchical Bayesian structure as a form of small-area estimation. The developed models borrow cross-sectional, temporal, and spatial strength in such a way that they can interpolate smoking levels in the non-survey years for all detailed level small domains. Smoothed trends of smoking prevalence for higher aggregation levels are obtained by aggregation of the detailed level trend predictions.

RESULTS

Model-based small area estimators provide consistent and reasonable smoothed trends at both detailed and higher aggregation levels. Results show that the national-level trend exhibits a steeper linear decline over the study period, from 24% in 2001 to 12% in 2021, with a considerable gender difference of around 5% over the period, with males reporting a higher prevalence. Improved model-based estimates at the state level and by age also show steady declines in trends except for the Northern Territory (still above 20%) and older age groups 60-69 and ≥70 years (declining trends remain stable after 2012).

CONCLUSIONS

The findings of the study identify the geographical-demographic groups that had poor improvement over the period 2001-2021, and are still behind the target of achieving lower smoking prevalence. These, in turn, will help health researchers and policymakers deliver targeted programs to the most vulnerable, enabling the nation to meet its health goals in a timely way.

An Introduction to Bayesian Spatial Smoothing Methods for Disease Mapping: Modeling County Firearm Suicide Mortality Rates

This article introduces Bayesian spatial smoothing models for disease mapping, a specific application of small area estimation where the full universe of data is known, to a wider audience of public health professionals using firearm suicide as a motivating example. Besag, York and Mollié (BYM) Poisson spatial and space-time smoothing models were fit to firearm suicide counts for the years 2014-2018. County raw death rates in 2018 ranged from 0-24.81 deaths per 10,000 people. However, the highest mortality rate was highly unstable based on only 2 deaths in a population of approximately 800, and 82.4% of contiguous US counties experienced fewer than 10 firearm suicide deaths and were thus suppressed. Spatially smoothed county firearm suicide mortality estimates ranged from 0.06-4.05 deaths per 10,000 people and could be reported for all counties. The space-time smoothing model produced similar estimates with narrower credible intervals as it allowed counties to gained precision from adjacent neighbors and their own rates in adjacent years. Bayesian spatial smoothing methods are a useful tool for evaluating spatial health disparities in small geographies where small numbers can result in highly variable rate estimates, and new estimation techniques in R have made fitting these models more accessible to researchers.

Spatial Patterning of Spontaneous and Medically Indicated Preterm Birth in Philadelphia

Preterm birth (PTB) remains a key public health issue that disproportionately affects Black individuals. Since spontaneous PTB (sPTB) and medically indicated PTB (mPTB) may have different causes and interventions, we quantified racial disparities for sPTB and mPTB, and we characterized the geographic patterning of these phenotypes, overall and according to race/ethnicity. We examined a pregnancy cohort of 83,952 singleton births at 2 Philadelphia hospitals from 2008-2020, and classified each PTB as sPTB or mPTB. We used binomial regression to quantify the magnitude of racial disparities between non-Hispanic Black and non-Hispanic White individuals, then generated small area estimates by applying a Bayesian model that accounts for small numbers and smooths estimates of PTB risk by borrowing information from neighboring areas. Racial disparities in both sPTB and mPTB were significant (relative risk of sPTB = 1.83, 95% confidence interval: 1.70, 1.98; relative risk of mPTB = 2.20, 95% confidence interval: 2.00, 2.42). The disparity was 20% greater in mPTB than sPTB. There was substantial geographic variation in PTB, sPTB, and mPTB risks and racial disparity. Our findings underscore the importance of distinguishing PTB phenotypes within the context of public health and preventive medicine. Future work should consider social and environmental exposures that may explain geographic differences in PTB risk and disparities.

Subnational estimates of maternal mortality in Nigeria: Secondary Data Analysis of female siblings' survivorship histories

High Maternal Mortality (MM) in Nigeria is complicated by the absence of reliable estimates at subnational levels. Obtaining accurate data at the state and geopolitical region levels is crucial for effective policy-making and targeted interventions. This study employs novel small area estimation techniques to derive plausible estimates of Maternal Mortality rates and ratios for Nigerian states and geopolitical regions. Data from 293,769 female siblings, provided by 114,154 women in the Nigeria Demographic and Health Surveys of 2008, 2013, and 2018, are used. Empirical Bayesian technique and the James-Stein estimator are applied to estimate MM Rates and Ratios, respectively. Maternal Mortality Ratio is highest in rural areas, Northern Nigeria states, and regions. While the South West exhibits lower MMRatio, the Northern regions, particularly the North-East, show consistently higher ratios. Mortality trends have decreased in the North West and South East regions but increased in the South West from 2008 to 2018. Addressing these disparities is essential for achieving sustainable development goals and improving maternal health in Nigeria.

Predicting potential areas at risk of the Dengue Hemorrhagic Fever in Jakarta, Indonesia-analyzing the accuracy of predictive hot spot analysis in the absence of small geographical area data

Dengue Hemorrhagic Fever (DHF), a more severe form of dengue, is one of the most rapidly spreading mosquito-borne diseases in the world. This study is motivated by the rising DHF incidence in Jakarta, the capital city of Indonesia. We mainly utilized hot spot analysis, which employs spatial statistics to find at-risk areas for DHF outbreaks in Jakarta's five municipalities. However, producing informative results from hot spot analysis requires a complete set of data on each of Jakarta's 42 districts, which is not available. We thus propose the idea of using small area estimation (SAE) and machine learning to make up for the lack of data. To evaluate whether this proposed method is effective, we compare the hot spot results from the estimation with the actual data of each district. The results show that the estimated hot spot map is similar to the hot spot map from the actual data. This implies that it is possible to find potential at-risk areas of dengue fever without a complete dataset in every small geographic area. We expect that this research can increase the performance of DHF control measures at the district level, even in the absence of small area data.

Spatio-temporal changes in pre-exposure prophylaxis uptake among MSM in mainland France between 2016 and 2021: a Bayesian small area approach with MSM population estimation

INTRODUCTION

In France, oral pre-exposure prophylaxis (PrEP) for HIV prevention has been publicly available since 2016, mainly targeting at men who have sex with men (MSM). Reliable and robust estimations of the actual PrEP uptake among MSM on a localized level can provide additional insights to identify and better reach marginalized MSM within current HIV prevention service provision. This study used national pharmaco-epidemiology surveillance data and regional MSM population estimations to model the spatio-temporal distribution of PrEP uptake among MSM in France 2016-2021 to identify marginalized MSM at risk for HIV and increase their PrEP uptake.

METHODS

We first applied Bayesian spatial analyses with survey-surveillance-based HIV incidence data as a spatial proxy to estimate the size of (1) regional HIV-negative MSM populations and (2) MSM who could be eligible for PrEP use according to French PrEP guidelines. We then applied Bayesian spatio-temporal ecological regression modelling to estimate the regional prevalence and relative probability of the overall- and new-PrEP uptake from 2016 to 2021 across France.

RESULTS

HIV-negative and PrEP-eligible MSM populations vary regionally across France. Île-de-France was estimated to have the highest MSM density compared to other French regions. According to the final spatio-temporal model, the relative probability of overall PrEP uptake was heterogeneous across France but remained stable over time. Urban areas have higher-than-average probabilities of PrEP uptake. The prevalence of PrEP use increased steadily (ranging from 8.8% [95% credible interval 8.5%;9.0%] in Nouvelle-Aquitaine to 38.2% [36.5%;39.9%] in Centre-Val-de-Loire in 2021).

CONCLUSIONS

Our results show that using Bayesian spatial analysis as a novel methodology to estimate the localized HIV-negative MSM population is feasible and applicable. Spatio-temporal models showed that despite the increasing prevalence of PrEP use in all regions, geographical disparities and inequalities of PrEP uptake continued to exist over time. We identified regions that would benefit from greater tailoring and delivery efforts. Based on our findings, public health policies and HIV prevention strategies could be adjusted to better combat HIV infections and to accelerate ending the HIV epidemic.

A Correlated Network Scale-up Model: Finding the Connection Between Subpopulations

Aggregated relational data (ARD), formed from "How many X's do you know?" questions, is a powerful tool for learning important network characteristics with incomplete network data. Compared to traditional survey methods, ARD is attractive as it does not require a sample from the target population and does not ask respondents to self-reveal their own status. This is helpful for studying hard-to-reach populations like female sex workers who may be hesitant to reveal their status. From December 2008 to February 2009, the Kiev International Institute of Sociology (KIIS) collected ARD from 10,866 respondents to estimate the size of HIV-related groups in Ukraine. To analyze this data, we propose a new ARD model which incorporates respondent and group covariates in a regression framework and includes a bias term that is correlated between groups. We also introduce a new scaling procedure utilizing the correlation structure to further reduce biases. The resulting size estimates of those most-at-risk of HIV infection can improve the HIV response efficiency in Ukraine. Additionally, the proposed model allows us to better understand two network features without the full network data: 1. What characteristics affect who respondents know, and 2. How is knowing someone from one group related to knowing people from other groups. These features can allow researchers to better recruit marginalized individuals into the prevention and treatment programs. Our proposed model and several existing NSUM models are implemented in the networkscaleup R package.

Small Area Forecasting of Opioid-Related Mortality: Bayesian Spatiotemporal Dynamic Modeling Approach

BACKGROUND

Opioid-related overdose mortality has remained at crisis levels across the United States, increasing 5-fold and worsened during the COVID-19 pandemic. The ability to provide forecasts of opioid-related mortality at granular geographical and temporal scales may help guide preemptive public health responses. Current forecasting models focus on prediction on a large geographical scale, such as states or counties, lacking the spatial granularity that local public health officials desire to guide policy decisions and resource allocation.

OBJECTIVE

The overarching objective of our study was to develop Bayesian spatiotemporal dynamic models to predict opioid-related mortality counts and rates at temporally and geographically granular scales (ie, ZIP Code Tabulation Areas [ZCTAs]) for Massachusetts.

METHODS

We obtained decedent data from the Massachusetts Registry of Vital Records and Statistics for 2005 through 2019. We developed Bayesian spatiotemporal dynamic models to predict opioid-related mortality across Massachusetts' 537 ZCTAs. We evaluated the prediction performance of our models using the one-year ahead approach. We investigated the potential improvement of prediction accuracy by incorporating ZCTA-level demographic and socioeconomic determinants. We identified ZCTAs with the highest predicted opioid-related mortality in terms of rates and counts and stratified them by rural and urban areas.

RESULTS

Bayesian dynamic models with the full spatial and temporal dependency performed best. Inclusion of the ZCTA-level demographic and socioeconomic variables as predictors improved the prediction accuracy, but only in the model that did not account for the neighborhood-level spatial dependency of the ZCTAs. Predictions were better for urban areas than for rural areas, which were more sparsely populated. Using the best performing model and the Massachusetts opioid-related mortality data from 2005 through 2019, our models suggested a stabilizing pattern in opioid-related overdose mortality in 2020 and 2021 if there were no disruptive changes to the trends observed for 2005-2019.

CONCLUSIONS

Our Bayesian spatiotemporal models focused on opioid-related overdose mortality data facilitated prediction approaches that can inform preemptive public health decision-making and resource allocation. While sparse data from rural and less populated locales typically pose special challenges in small area predictions, our dynamic Bayesian models, which maximized information borrowing across geographic areas and time points, were used to provide more accurate predictions for small areas. Such approaches can be replicated in other jurisdictions and at varying temporal and geographical levels. We encourage the formation of a modeling consortium for fatal opioid-related overdose predictions, where different modeling techniques could be ensembled to inform public health policy.

Spatial Analysis on Supply and Demand of Adult Surgical Masks in Taipei Metropolitan Areas in the Early Phase of the COVID-19 Pandemic

This study aimed to assess the gap between the supply and demand of adult surgical masks under limited resources. Owing to the implementation of the real-name mask rationing system, the historical inventory data of aggregated mask consumption in a pharmacy during the early period of the COVID-19 outbreak (April and May 2020) in Taiwan were analyzed for supply-side analysis. We applied the Voronoi diagram and areal interpolation methods to delineate the average supply of customer counts from a pharmacy to a village (administrative level). On the other hand, the expected number of demand counts was estimated from the population data. The relative risk (RR) of supply, which is the average number of adults served per day divided by the expected number in a village, was modeled under a Bayesian hierarchical framework, including Poisson, negative binomial, Poisson spatial, and negative binomial spatial models. We observed that the number of pharmacies in a village is associated with an increasing supply, whereas the median annual per capita income of the village has an inverse relationship. Regarding land use percentages, percentages of the residential and the mixed areas in a village are negatively associated, while the school area percentage is positively associated with the supply in the Poisson spatial model. The corresponding uncertainty measurement: villages where the probability exceeds the risk of undersupply, that is, Pr (RR < 1), were also identified. The findings of the study may help health authorities to evaluate the spatial allocation of anti-epidemic resources, such as masks and rapid test kits, in small areas while identifying priority areas with the suspicion of undersupply in the beginning stages of outbreaks.

Multivariate small area modelling of undernutrition prevalence among under-five children in Bangladesh

District-representative data are rarely collected in the surveys for identifying localised disparities in Bangladesh, and so district-level estimates of undernutrition indicators - stunting, wasting and underweight - have remained largely unexplored. This study aims to estimate district-level prevalence of these indicators by employing a multivariate Fay-Herriot (MFH) model which accounts for the underlying correlation among the undernutrition indicators. Direct estimates (DIR) of the target indicators and their variance-covariance matrices calculated from the 2019 Bangladesh Multiple Indicator Cluster Survey microdata have been used as input for developing univariate Fay-Herriot (UFH), bivariate Fay-Herriot (BFH) and MFH models. The comparison of the various model-based estimates and their relative standard errors with the corresponding direct estimates reveals that the MFH estimator provides unbiased estimates with more accuracy than the DIR, UFH and BFH estimators. The MFH model-based district level estimates of stunting, wasting and underweight range between 16 and 43%, 15 and 36%, and 6 and 13% respectively. District level bivariate maps of undernutrition indicators show that districts in north-eastern and south-eastern parts are highly exposed to either form of undernutrition, than the districts in south-western and central parts of the country. In terms of the number of undernourished children, millions of children affected by either form of undernutrition are living in densely populated districts like the capital district Dhaka, though undernutrition indicators (as a proportion) are comparatively lower. These findings can be used to target districts with a concurrence of multiple forms of undernutrition, and in the design of urgent intervention programs to reduce the inequality in child undernutrition at the localised district level.

Geographic Variation in HIV Testing Among Transgender and Nonbinary Adults in the United States

BACKGROUND

Transgender and nonbinary (TNB) populations are disproportionately affected by HIV and few local health departments or HIV surveillance systems collect/report data on TNB identities. Our objective was to estimate the prevalence of HIV testing among TNB adults by US county and state, with a focus on the Ending the HIV Epidemic (EHE) geographies.

METHODS

We applied a Bayesian hierarchical spatial small area estimation model to data from the 2015 US Transgender Survey, a large national cross-sectional Internet-based survey. We estimated the county- and state-level proportion of TNB adults who ever tested or tested for HIV in the last year by gender identity, race/ethnicity, and age.

RESULTS

Our analysis included 26,100 TNB participants with valid zip codes who resided in 1688 counties (54% of all 3141 counties that cover 92% of the US population). The median county-level proportion of TNB adults who ever tested for HIV was 44% (range 10%-80%) and who tested in the last year was 17% (range 4%-44%). Within most counties, testing was highest among transgender women, black respondents, and people aged ≥25 years. HIV testing was lowest among nonbinary people and young adults aged <25 years. The proportion of TNB adults who tested within the last year was very low in most EHE counties and in all 7 rural states.

CONCLUSIONS

HIV testing among TNB adults is likely below national recommendations in the majority of EHE geographies. Geographic variation in HIV testing patterns among TNB adults indicates that testing strategies need to be tailored to local settings.

On the Use of Aggregate Survey Data for Estimating Regional Major Depressive Disorder Prevalence

Major depression is a severe mental disorder that is associated with strongly increased mortality. The quantification of its prevalence on regional levels represents an important indicator for public health reporting. In addition to that, it marks a crucial basis for further explorative studies regarding environmental determinants of the condition. However, assessing the distribution of major depression in the population is challenging. The topic is highly sensitive, and national statistical institutions rarely have administrative records on this matter. Published prevalence figures as well as available auxiliary data are typically derived from survey estimates. These are often subject to high uncertainty due to large sampling variances and do not allow for sound regional analysis. We propose a new area-level Poisson mixed model that accounts for measurement errors in auxiliary data to close this gap. We derive the empirical best predictor under the model and present a parametric bootstrap estimator for the mean squared error. A method of moments algorithm for consistent model parameter estimation is developed. Simulation experiments are conducted to show the effectiveness of the approach. The methodology is applied to estimate the major depression prevalence in Germany on regional levels crossed by sex and age groups.

TB Hackathon: Development and Comparison of Five Models to Predict Subnational Tuberculosis Prevalence in Pakistan

Pakistan's national tuberculosis control programme (NTP) is among the many programmes worldwide that value the importance of subnational tuberculosis (TB) burden estimates to support disease control efforts, but do not have reliable estimates. A hackathon was thus organised to solicit the development and comparison of several models for small area estimation of TB. The TB hackathon was launched in April 2019. Participating teams were requested to produce district-level estimates of bacteriologically positive TB prevalence among adults (over 15 years of age) for 2018. The NTP provided case-based data from their 2010-2011 TB prevalence survey, along with data relating to TB screening, testing and treatment for the period between 2010-2011 and 2018. Five teams submitted district-level TB prevalence estimates, methodological details and programming code. Although the geographical distribution of TB prevalence varied considerably across models, we identified several districts with consistently low notification-to-prevalence ratios. The hackathon highlighted the challenges of generating granular spatiotemporal TB prevalence forecasts based on a cross-sectional prevalence survey data and other data sources. Nevertheless, it provided a range of approaches to subnational disease modelling. The NTP's use and plans for these outputs shows that, limitations notwithstanding, they can be valuable for programme planning.

Ensuring Optimal Community HIV Testing Services in Nigeria Using an Enhanced Community Case-Finding Package (ECCP), October 2019-March 2020: Acceleration to HIV Epidemic Control

Purpose

The 2018 Nigeria HIV/AIDS Indicator and Impact Survey (NAIIS) showed Nigeria’s progress toward the UNAIDS 90-90-90 targets: 47% of HIV-positive individuals knew their status; of these, 96% were receiving antiretroviral therapy (ART); and of these, 81% were virally suppressed. To improve identification of HIV-positive individuals, Nigeria developed an Enhanced Community Case-Finding Package (ECCP). We describe ECCP implementation in nine states and assess its effect.

Methods

ECCP included four core strategies (small area estimation [SAE] of people living with HIV [PLHIV], map of HIV-positive patients by residence, HIV risk-screening tool [HRST], and index testing [IT]) and four supportive strategies (alternative healthcare outlets, performance-based incentives for field testers, Project Extension for Community Healthcare Outcomes, and interactive dashboards). ECCP was deployed in nine of 10 states prioritized for ART scale-up. Weekly program data (October 2019–March 2020) were tracked and analyzed.

Results

Of the total 774 LGAs in Nigeria, using SAE, 103 (13.3%) high-burden LGAs were identified, in which 2605 (28.0%) out of 9,294 hotspots were prioritized by mapping newly identified PLHIV by residential addresses. Over 22 weeks, among 882,449 individuals screened using HRST, 723,993 (82.0%) were eligible and tested for HIV (state range, 43.7–90.4%), out of which 20,616 were positive. Through IT, an additional 3,724 PLHIV were identified. In total, 24,340 PLHIV were identified and 97.4% were linked to life-saving antiretroviral therapy. The number of newly identified PLHIV increased 17-fold over 22 weeks (week 1: 89; week 22: 1,632). Overall mean HIV positivity rate by state was 3.3% (range, 1.8–6.4%).

Conclusion

Using ECCP in nine states in Nigeria increased the number of PLHIV in the community who knew their status, allowing them to access life-saving care and decreasing the risk of HIV transmission.

Thirty Years of The Network Scale-up Method

Estimating the size of hard-to-reach populations is an important problem for many fields. The Network Scale-up Method (NSUM) is a relatively new approach to estimate the size of these hard-to-reach populations by asking respondents the question, "How many X's do you know," where X is the population of interest (e.g. "How many female sex workers do you know?"). The answers to these questions form Aggregated Relational Data (ARD). The NSUM has been used to estimate the size of a variety of subpopulations, including female sex workers, drug users, and even children who have been hospitalized for choking. Within the Network Scale-up methodology, there are a multitude of estimators for the size of the hidden population, including direct estimators, maximum likelihood estimators, and Bayesian estimators. In this article, we first provide an in-depth analysis of ARD properties and the techniques to collect the data. Then, we comprehensively review different estimation methods in terms of the assumptions behind each model, the relationships between the estimators, and the practical considerations of implementing the methods. We apply many of the models discussed in the review to one canonical data set and compare their performance and unique features, presented in the supplementary materials. Finally, we provide a summary of the dominant methods and an extensive list of the applications, and discuss the open problems and potential research directions in this area.

l2-Penalized temporal logit-mixed models for the estimation of regional obesity prevalence over time

Obesity is considered to be one of the primary health risks in modern industrialized societies. Estimating the evolution of its prevalence over time is an essential element of public health reporting. This requires the application of suitable statistical methods on epidemiologic data with substantial local detail. Generalized linear-mixed models with medical treatment records as covariates mark a powerful combination for this purpose. However, the task is methodologically challenging. Disease frequencies are subject to both regional and temporal heterogeneity. Medical treatment records often show strong internal correlation due to diagnosis-related grouping. This frequently causes excessive variance in model parameter estimation due to rank-deficiency problems. Further, generalized linear-mixed models are often estimated via approximate inference methods as their likelihood functions do not have closed forms. These problems combined lead to unacceptable uncertainty in prevalence estimates over time. We propose an l₂-penalized temporal logit-mixed model to solve these issues. We derive empirical best predictors and present a parametric bootstrap to estimate their mean-squared errors. A novel penalized maximum approximate likelihood algorithm for model parameter estimation is stated. With this new methodology, the regional obesity prevalence in Germany from 2009 to 2012 is estimated. We find that the national prevalence ranges between 15 and 16%, with significant regional clustering in eastern Germany.

A new robust Bayesian small area estimation via α -stable model for estimating the proportion of athletic students in California

In the last few years, diabetes mellitus and obesity revealed to be one of the fastest‐growing chronic diseases in youth in the United States. The number of new diabetes cases is dramatically increasing, and, for the moment, effective therapy does not exist. Experts believe that one of the causes of this increase is the decline in exercise behavior. The California Education Code requires local educational agencies (LEAs) to administer the FITNESSGRAM, the Physical Fitness Test (PFT), to Californian students of public schools. This test evaluates six fitness areas, and experts defined that a passing result on all six areas of the test represents a fitness level that offers some protection against the diseases associated with physical inactivity. We consider 2015–2016 data provided by the California Department of Education (CDE): for each Californian county (m=57), we aim at estimating the county‐level proportion of students with a score equal to six. To account for the heterogeneity of the phenomenon and the presence of outlying counties, we extend the standard area‐level model by specifying the random effects as a symmetric α‐stable (SαS) distribution that can accommodate different types of outlying observations. The model can accurately estimate the county‐level proportion of students with a score equal to six. Results highlight some interesting relationships with social and economic situations in each county. The performance of the proposed model is also investigated through an extensive simulation study.

Linking Data on Constituent Health with Elected Officials' Opinions: Associations Between Urban Health Disparities and Mayoral Officials' Beliefs About Health Disparities in Their Cities

Policy Points Mayoral officials' opinions about the existence and fairness of health disparities in their city are positively associated with the magnitude of income-based life expectancy disparity in their city. Associations between mayoral officials' opinions about health disparities in their city and the magnitude of life expectancy disparity in their city are not moderated by the social or fiscal ideology of mayoral officials or the ideology of their constituents. Highly visible and publicized information about mortality disparities, such as that related to COVID-19 disparities, has potential to elevate elected officials' perceptions of the severity of health disparities and influence their opinions about the issue.

CONTEXT

A substantive body of research has explored what factors influence elected officials' opinions about health issues. However, no studies have assessed the potential influence of the health of an elected official's constituents. We assessed whether the magnitude of income-based life expectancy disparity within a city was associated with the opinions of that city's mayoral official (i.e., mayor or deputy mayor) about health disparities in their city.

METHODS

The independent variable was the magnitude of income-based life expectancy disparity in US cities. The magnitude was determined by linking 2010-2015 estimates of life expectancy and median household income for 8,434 census tracts in 224 cities. The dependent variables were mayoral officials' opinions from a 2016 survey about the existence and fairness of health disparities in their city (n = 224, response rate 30.3%). Multivariable logistic regression was used to adjust for characteristics of mayoral officials (e.g., ideology) and city characteristics.

FINDINGS

In cities in the highest income-based life expectancy disparity quartile, 50.0% of mayoral officials "strongly agreed" that health disparities existed and 52.7% believed health disparities were "very unfair." In comparison, among mayoral officials in cities in the lowest disparity quartile 33.9% "strongly agreed" that health disparities existed and 22.2% believed the disparities were "very unfair." A 1-year-larger income-based life expectancy disparity in a city was associated with 25% higher odds that the city's mayoral official would "strongly agree" that health disparities existed (odds ratio [OR] = 1.25; P = .04) and twice the odds that the city's mayoral official would believe that such disparities were "very unfair" (OR = 2.24; P <.001).

CONCLUSIONS

Mayoral officials' opinions about health disparities in their jurisdictions are generally aligned with, and potentially influenced by, information about the magnitude of income-based life expectancy disparities among their constituents.

Creating local estimates from a population health survey: practical application of small area estimation methods

Regular health surveys can produce reliable estimates at higher geographic levels but not for small areas. Alternatives are to aggregate data over several years or use model-based methods. We created and evaluated model-based estimates for four health-related outcomes by gender, for 153 Local Government Areas using data from the New South Wales Population Health Survey. The evaluation examined evidence on bias and determined the covariates available and appropriate for each outcome variable. The evaluation considered the likely precision of the resulting estimates. The bias and precision of results for single years (2006–2008) for each outcome variable using six covariate specifications were compared with direct survey estimates based on a single year's data and those obtained by aggregating over seven years. A practical issue is how to choose covariates to include in the models as the best covariate specification varies between outcome variables. Model-based results had median root mean squared errors between 3.3% and 5.5% (max 5.2% and 11.3% respectively) and median relative root mean squared errors between 6.8% and 24.5% (max 11.7% and 41.5% respectively). The model-based estimates were unbiased compared with direct estimates based on one or seven years of data and when aggregated to a point where direct estimates were reliable. The bias and reliability assessment process provides a way for policymakers to have confidence in model-based estimates.

Spatial Modelling to Inform Public Health Based on Health Surveys: Impact of Unsampled Areas at Lower Geographical Scale

Small area estimation is an important tool to provide area-specific estimates of population characteristics for governmental organizations in the context of education, public health and care. However, many demographic and health surveys are unrepresentative at a small geographical level, as often areas at a lower level are not included in the sample due to financial or logistical reasons. In this paper, we investigated (1) the effect of these unsampled areas on a variety of design-based and hierarchical model-based estimates and (2) the benefits of using auxiliary information in the estimation process by means of an extensive simulation study. The results showed the benefits of hierarchical spatial smoothing models towards obtaining more reliable estimates for areas at the lowest geographical level in case a spatial trend is present in the data. Furthermore, the importance of auxiliary information was highlighted, especially for geographical areas that were not included in the sample. Methods are illustrated on the 2008 Mozambique Poverty and Social Impact Analysis survey, with interest in the district-specific prevalence of school attendance.

Small area estimation and hotspot identification of opioid use disorder among military veterans living in the Southern United States

BACKGROUND

The purpose of this study was to estimate opioid use disorder prevalence rates at the county level among veterans in Alabama and to determine hotspots of said rates. Methods: By combining data from the National Survey on Drug Use and Health and the American Community Survey, we developed a mixed-effects generalized linear model of opioid use disorder and modeled probabilities onto veteran-specific population counts at the county level in Alabama. Results: The average model-based estimate for opioid use disorder prevalence among veterans in Alabama from 2015 to 2017 was 0.79% (SD = 0.16), with a minimum of 0.52% (i.e., Lowndes county, Alabama) and a maximum of 1.10% (Dale county, Alabama). Hotspot analysis revealed a significant cluster of "high-high" veteran opioid use disorder prevalence in neighboring Marion, Winston, and Cullman counties. Conclusions: The application of the statistical technique presented in this study can provide feasible, cost-effective, and practical county-level prevalence estimates of veteran-specific opioid use disorder and should be widely applied by states and counties so that they can more accurately and efficiently allocate resources to caring for veterans with an opioid use disorder.

State-Level Affordability of Factory-Made Cigarettes among Current US Smokers: Findings from the ITC US Survey, 2003-2015

Cigarette affordability measures the price smokers pay for cigarettes in relation to their incomes. Affordability can be measured using the relative income price of cigarettes (RIP), or the price smokers pay to purchase 100 packs of 20 cigarettes divided by their per capita household income. Using longitudinal data from 7046 smokers participating in the International Tobacco Control (ITC) US Survey, the purpose of this study was to test whether affordability significantly changed following the US federal tax increase implemented on 1 April 2009. This study also estimated temporal trends in affordability from 2003-2015 at state and national levels using small area estimation methods and segmented linear mixed effects regression models. RIP increased slightly during 2003-2008. This was followed by a 30% increase during 2008-2010, indicating cigarettes were less affordable after the federal tax increase. RIP continued to increase during 2010-2013 but decreased during 2013-2015, suggesting cigarettes have recently become more affordable for US smokers. State-level trends in RIP were consistent with overall national trends. Controlling for other factors, a $1 increase in the state excise tax was significantly associated with a 9% increase in RIP, indicating state taxes reduced affordability. Tax-induced price increases must keep pace with underlying economic conditions to ensure cigarettes do not become more affordable over time.

Integrating national surveys to estimate small area variations in poor health and limiting long-term illness in Great Britain

OBJECTIVES

This study aims to address, for the first time, the challenges of constructing small area estimates of health status using linked national surveys. The study also seeks to assess the concordance of these small area estimates with data from national censuses.

SETTING

Population level health status in England, Scotland and Wales.

PARTICIPANTS

A linked integrated dataset of 23 374 survey respondents (16+ years) from the 2011 waves of the Health Survey for England (n=8603), the Scottish Health Survey (n=7537) and the Welsh Health Survey (n=7234).

PRIMARY AND SECONDARY OUTCOME MEASURES

Population prevalence of poorer self-rated health and limiting long-term illness. A multilevel small area estimation modelling approach was used to estimate prevalence of these outcomes for middle super output areas in England and Wales and intermediate zones in Scotland. The estimates were then compared with matched measures from the contemporaneous 2011 UK Census.

RESULTS

There was a strong positive association between the small area estimates and matched census measures for all three countries for both poorer self-rated health (r=0.828, 95% CI 0.821 to 0.834) and limiting long-term illness (r=0.831, 95% CI 0.824 to 0.837), although systematic differences were evident, and small area estimation tended to indicate higher prevalences than census data.

CONCLUSIONS

Despite strong concordance, variations in the small area prevalences of poorer self-rated health and limiting long-term illness evident in census data cannot be replicated perfectly using small area estimation with linked national surveys. This reflects a lack of harmonisation between surveys over question wording and design. The nature of small area estimates as 'expected values' also needs to be better understood.

Predicting Periodontitis at State and Local Levels in the United States

The objective of the study was to estimate the prevalence of periodontitis at state and local levels across the United States by using a novel, small area estimation (SAE) method. Extended multilevel regression and poststratification analyses were used to estimate the prevalence of periodontitis among adults aged 30 to 79 y at state, county, congressional district, and census tract levels by using periodontal data from the National Health and Nutrition Examination Survey (NHANES) 2009-2012, population counts from the 2010 US census, and smoking status estimates from the Behavioral Risk Factor Surveillance System in 2012. The SAE method used age, race, gender, smoking, and poverty variables to estimate the prevalence of periodontitis as defined by the Centers for Disease Control and Prevention/American Academy of Periodontology case definitions at the census block levels and aggregated to larger administrative and geographic areas of interest. Model-based SAEs were validated against national estimates directly from NHANES 2009-2012. Estimated prevalence of periodontitis ranged from 37.7% in Utah to 52.8% in New Mexico among the states (mean, 45.1%; median, 44.9%) and from 33.7% to 68% among counties (mean, 46.6%; median, 45.9%). Severe periodontitis ranged from 7.27% in New Hampshire to 10.26% in Louisiana among the states (mean, 8.9%; median, 8.8%) and from 5.2% to 17.9% among counties (mean, 9.2%; median, 8.8%). Overall, the predicted prevalence of periodontitis was highest for southeastern and southwestern states and for geographic areas in the Southeast along the Mississippi Delta, as well as along the US and Mexico border. Aggregated model-based SAEs were consistent with national prevalence estimates from NHANES 2009-2012. This study is the first-ever estimation of periodontitis prevalence at state and local levels in the United States, and this modeling approach complements public health surveillance efforts to identify areas with a high burden of periodontitis.

REGULARIZED BRAIN READING WITH SHRINKAGE AND SMOOTHING

Functional neuroimaging measures how the brain responds to complex stimuli. However, sample sizes are modest, noise is substantial, and stimuli are high dimensional. Hence, direct estimates are inherently imprecise and call for regularization. We compare a suite of approaches which regularize via shrinkage: ridge regression, the elastic net (a generalization of ridge regression and the lasso), and a hierarchical Bayesian model based on small area estimation (SAE). We contrast regularization with spatial smoothing and combinations of smoothing and shrinkage. All methods are tested on functional magnetic resonance imaging (fMRI) data from multiple subjects participating in two different experiments related to reading, for both predicting neural response to stimuli and decoding stimuli from responses. Interestingly, when the regularization parameters are chosen by cross-validation independently for every voxel, low/high regularization is chosen in voxels where the classification accuracy is high/low, indicating that the regularization intensity is a good tool for identification of relevant voxels for the cognitive task. Surprisingly, all the regularization methods work about equally well, suggesting that beating basic smoothing and shrinkage will take not only clever methods, but also careful modeling.

Validation of multilevel regression and poststratification methodology for small area estimation of health indicators from the behavioral risk factor surveillance system

Small area estimation is a statistical technique used to produce reliable estimates for smaller geographic areas than those for which the original surveys were designed. Such small area estimates (SAEs) often lack rigorous external validation. In this study, we validated our multilevel regression and poststratification SAEs from 2011 Behavioral Risk Factor Surveillance System data using direct estimates from 2011 Missouri County-Level Study and American Community Survey data at both the state and county levels. Coefficients for correlation between model-based SAEs and Missouri County-Level Study direct estimates for 115 counties in Missouri were all significantly positive (0.28 for obesity and no health-care coverage, 0.40 for current smoking, 0.51 for diabetes, and 0.69 for chronic obstructive pulmonary disease). Coefficients for correlation between model-based SAEs and American Community Survey direct estimates of no health-care coverage were 0.85 at the county level (811 counties) and 0.95 at the state level. Unweighted and weighted model-based SAEs were compared with direct estimates; unweighted models performed better. External validation results suggest that multilevel regression and poststratification model-based SAEs using single-year Behavioral Risk Factor Surveillance System data are valid and could be used to characterize geographic variations in health indictors at local levels (such as counties) when high-quality local survey data are not available.

A flexible urban health index for small area disparities

Available urban health metrics focus primarily on large area rankings. Less has been done to develop an index that provides information about level of health and health disparities for small geographic areas. Adopting a method used by the Human Development Index, we standardized indicators for small area units on a (0, 1) interval and combined them using their geometric mean to form an Urban Health Index (UHI). Disparities were assessed using the ratio of the highest to lowest decile and measurement of the slope of the eight middle deciles (middle; 80 %) of the data. We examined the sensitivity of the measure to weighting, to changes in the method, to correlation among indicators, and to substitution of indicators. Using seven health determinants and applying these methods to the 128 census tracts in the city of Atlanta, USA, we found a disparity ratio of 5.92 and a disparity slope of 0.54, suggesting substantial inequality and heterogeneity of risk. The component indicators were highly correlated; their systematic removal had a small effect on the results. Except in extreme cases, weighting had a little effect on the rankings. A map of Atlanta census tracts exposed a swath of high disparity. UHI rankings, ratio, and slope were resistant to alteration in composition and to non-extreme weighting schemes. This empirical evaluation was limited to a single realization, but suggests that a flexible tool, whose method rather than content is standardized, may be of use for local evaluation, for decision making, and for area comparison.

Dasymetric Modeling and Uncertainty

Dasymetric models increase the spatial resolution of population data by incorporating related ancillary data layers. The role of uncertainty in dasymetric modeling has not been fully addressed as of yet. Uncertainty is usually present because most population data are themselves uncertain, and/or the geographic processes that connect population and the ancillary data layers are not precisely known. A new dasymetric methodology - the Penalized Maximum Entropy Dasymetric Model (P-MEDM) - is presented that enables these sources of uncertainty to be represented and modeled. The P-MEDM propagates uncertainty through the model and yields fine-resolution population estimates with associated measures of uncertainty. This methodology contains a number of other benefits of theoretical and practical interest. In dasymetric modeling, researchers often struggle with identifying a relationship between population and ancillary data layers. The PEDM model simplifies this step by unifying how ancillary data are included. The P-MEDM also allows a rich array of data to be included, with disparate spatial resolutions, attribute resolutions, and uncertainties. While the P-MEDM does not necessarily produce more precise estimates than do existing approaches, it does help to unify how data enter the dasymetric model, it increases the types of data that may be used, and it allows geographers to characterize the quality of their dasymetric estimates. We present an application of the P-MEDM that includes household-level survey data combined with higher spatial resolution data such as from census tracts, block groups, and land cover classifications.

Small area estimates reveal high cigarette smoking prevalence in low-income cities of Los Angeles county

Los Angeles County has among the lowest smoking rates of large urban counties in the USA. Nevertheless, concerning disparities persist as high smoking prevalence is found among certain subgroups. We calculated adult smoking prevalence in the incorporated cities of Los Angeles County in order to identify cities with high smoking prevalence. The prevalence was estimated by a model-based small area estimation method with utilization of three data sources, including the 2007 Los Angeles County Health Survey, the 2000 Census, and the 2007 Los Angeles County Population Estimates and Projection System. Smoking prevalence varied considerably across cities, with a more than fourfold difference between the lowest (5.3%) and the highest prevalence (21.7%). Higher smoking prevalence was generally found in socioeconomically disadvantaged cities. The disparities identified here add another layer of data to our knowledge of the health inequities experienced by low-income urban communities and provide much sought data for local tobacco control. Our study also demonstrates the feasibility of providing credible local estimates of smoking prevalence using the model-based small area estimation method.

Modelling bias in combining small area prevalence estimates from multiple surveys

Combining information from multiple surveys can improve the quality of small area estimates. Customary approaches, such as the multiple-frame and statistical matching methods, require individual level data, whereas in practice often only multiple aggregate estimates are available. Commercial surveys usually produce such estimates without clear description of the methodology that is used. In this context, bias modelling is crucial, and we propose a series of Bayesian hierarchical models which allow for additive biases. Some of these models can also be fitted in a classical context, by using a mixed effects framework. We apply these methods to obtain estimates of smoking prevalence in local authorities across the east of England from seven surveys. All the surveys provide smoking prevalence estimates and confidence intervals at the local authority level, but they vary by time, sample size and transparency of methodology. Our models adjust for the biases in commercial surveys but incorporate information from all the sources to provide more accurate and precise estimates.

Validation of a model for estimating state and local prevalence of serious mental illness

This study addresses an ongoing problem in mental health needs assessment. This involves estimating the prevalence of an identified problem, specifically serious mental illness (SMI), for local areas in a reliable, valid, and cost-effective manner. The aim of the study is the application and testing of a recently introduced methodology from the field of small area estimation to determining SMI rates in the 48 contiguous US states, and in local areas of Massachusetts. It uses 'regression synthetic estimation fitted using area-level covariates', to estimate a model using data from the 2001-2002 replication of the National Comorbidity Study (n = 5593) and apply it, using 2000 STF-3C Census data, to various state and local areas in the United States. The estimates are then compared with independently collected SMI indicators. The estimates show not only face validity and internal consistency, but also predictive validity. The multiple logistic model has a sensitivity of 21.1% and a specificity of 95.1%, based largely on socio-economic disparities. Pearson r validity coefficients for the area estimates range from 0.43 to 0.75. The model generates a national estimate of SMI adults of 5.5%; for the 48 states, rates ranging from 4.7% to 7.0%; and for Massachusetts towns and cities, 1.1% to 7.5%.