Multilevel Regression With Post-stratification Research Articles

Assessing the National Representativeness of Estimates of Antimicrobial-Resistant Urogenital Neisseria gonorrhoeae in US Men, Gonococcal Isolate Surveillance Project, 2008-2018.

The percentage of Neisseria gonorrhoeae (GC) isolates with resistance or elevated minimum inhibitory concentrations to antimicrobials has steadily increased. Current estimates are based on the Gonococcal Isolate Surveillance Project (GISP), a sentinel surveillance study of male GC in the United States. This analysis seeks to assess for adjustment before treating aggregated GISP estimates as nationally representative of all reported male urogenital infections. We used multilevel regression with poststratification (MRP) to compute national estimates of the proportion of antimicrobial resistance (AMR) (defined as exceeding minimum inhibitory concentration thresholds) in male GC using data from 2008 to 2018 GISP and case reports. Sensitivity analyses investigated the impact of analysis assumptions and unmeasured variables. We additionally produced estimates of 2018 AMR GC cases among US men. National estimates were consistent with unweighted estimates. The estimated proportion of incident AMR GC infections in men with urogenital GC in 2018 was 51.5% (95% confidence interval [CI], 50.1%-52.9%), equating to an estimated 366,300 incident AMR GC infections in US men aged 15 to 39 years. Estimates of AMR for tested antimicrobials in male GC infections in 2018 ranged from 0.16% (95% CI, 0.08%-0.24%) for ceftriaxone to 29.9% (95% CI, 28.6%-31.1%) for ciprofloxacin. Sensitivity analyses revealed that unmeasured data on sex of sex partners could substantially impact weighted estimates. Antimicrobial resistance among reported incident male urogenital GC infections remains rare for ceftriaxone, the current standard of care. Aggregated GISP data are generally representative of men in the US who are reported with urogenital gonorrhea.

Novel estimates reveal subnational heterogeneities in disease-relevant contact patterns in the United States.

Population contact patterns fundamentally determine the spread of directly transmitted airborne pathogens such as SARS-CoV-2 and influenza. Reliable quantitative estimates of contact patterns are therefore critical to modeling and reducing the spread of directly transmitted infectious diseases and to assessing the effectiveness of interventions intended to limit risky contacts. While many countries have used surveys and contact diaries to collect national-level contact data, local-level estimates of age-specific contact patterns remain rare. Yet, these local-level data are critical since disease dynamics and public health policy typically vary by geography. To overcome this challenge, we introduce a flexible model that can estimate age-specific contact patterns at the subnational level by combining national-level interpersonal contact data with other locality-specific data sources using multilevel regression with poststratification (MRP). We estimate daily contact matrices for all 50 US states and Washington DC from April 2020 to May 2021 using national contact data from the US. Our results reveal important state-level heterogeneities in levels and trends of contacts across the US over the course of the COVID-19 pandemic, with implications for the spread of respiratory diseases.

Modeling state preferences for Covid-19 policies: Insights from the first pandemic summer

IntroductionDuring the COVID-19 pandemic, governments have experimented with a wide array of policies to further public health goals. This research offers an application of multilevel regression with post-stratification (MRP) analysis to assess state-level support for commonly implemented policies during the pandemic. MethodsWe conducted a national survey of U.S. adults using The Harris Poll panel from June 17–29, 2020. Respondents reported their support for a set of measures that were being considered in jurisdictions in the U.S. at the time the survey was fielded. MRP analysis was then used to generate estimates of state-level support. ResultsThe research presented here suggests generally high levels of support for mask mandates and social distancing measures in June 2020—support that was consistent throughout the United States. In comparison, support for other policies, such as changes to the road environment to create safer spaces for walking and bicycling, had generally low levels of support throughout the country. This research also provides some evidence that higher support for coronavirus-related policies could be found in more populous states with large urban centers, recognizing that there was low variability across states. ConclusionThis paper provides a unique application of MRP analysis in the public health field, uncovering noteworthy state-level patterns, and offering several avenues for future research. Future research could examine policy support at a small geographic level, such as by counties, to understand the distribution of support for public policies within states.

Improved Multilevel Regression with Poststratification through Machine Learning (autoMrP)

Multilevel regression with poststratification (MrP) has quickly become the gold standard for small area estimation. While the first MrP models did not include context-level information, current app...

Predicting neighbourhood-level psychiatric admission rates using multi-level regression with post-stratification-derived estimates of ecological cognitive social capital

Background with rationaleEcological cognitive social capital, the aggregated beliefs about community and norms of one’s neighbours, is an important protective factor for mental health. However, the techniques currently used in the literature to estimate it are basic. Multi-level regression with post-stratification (MRP) is a technique for making such small area estimates using survey data, and appears a promising alternative to existing methods.
 Main AimTo test the predictive validity of MRP-derived estimates of sense of belonging and generalised trust on psychiatric admission rates across Wales.
 MethodsMRP estimates of two questions measuring social capital, trust and belonging, were created for all middle super output areas in Wales using the National Survey for Wales 2016-17. These estimates were then used to predict rates of psychiatric admission in the same areas during 2017. Estimates of the same two variables were also computed using the simple aggregation approach used elsewhere in the literature, and the predictive validity of the two types of estimate were compared
 ResultsTrust and belonging were both protective factors, with MRP estimates yielding higher risk ratios (.80 and .77 respectively) than simple aggregation equivalents (.91 and .93).
 ConclusionMRP appears to be a useful technique for computing estimates of ecological social capital and other self-report based measures, outperforming techniques currently used in the literature in terms of predictive validity.

Using multilevel regression with poststratification to obtain regional health estimates from a Facebook-recruited sample

PurposeWe assess the effectiveness of multilevel regression with poststratification (MRP) as a tool to mitigate selection bias from online surveys of small geographical regions. MethodsWe collected self-reported health information from an Internet-based sample of adults residing within the St. Louis, MO, metropolitan area in 2017. We created Bayesian hierarchical models with three sets of predictor variables for each of six common health behaviors and outcomes, with results poststratified using the American Community Survey to estimate region and ZIP Code Tabulation Area–level prevalence. ResultsWhen comparing MRP estimates with a population-based sample as a reference, we found that adjustment using MRP can reduce bias in prevalence estimates and provide estimates for local area prevalence. 14 of 18 adjusted estimates were closer to the benchmark than the unadjusted estimates and MRP using all three covariate sets resulted in better overall agreement with the benchmark compared with the unadjusted estimates. ConclusionsMRP can improve prevalence estimates from self-selected Internet-based samples, although a nonnegligible amount of bias may remain. Illustrating the utility and limitations of this method will help researchers develop relevant estimates of the local public health burden, helping local health officials better understand and reduce poor health outcomes.

Estimating ecological social capital using multi-level regression with post-stratification: A spatial analysis of psychiatric admission rates in wales.

Although ecological cognitive social capital is an important predictor of mental health, measurement remains crude. Multi-level regression with post-stratification (MRP) is a technique for computing small area estimates of survey responses, used in political science but hitherto not for estimating ecological protective factors in epidemiology. National Survey for Wales 2016-17 data (N = 10,486) were used to produce MRP estimates of belonging and generalised trust for 410 middle super output areas covering Wales. These estimates were used to predict psychiatric admission rates in 2017 (N = 9978 cases). Low belonging and trust are ecological risk factors for psychiatric admissions, with a 29% (25-33%) and 25% (22-29%) increase in admissions per standard deviation decrease respectively. Equivalent results for using standard simple aggregation of survey data by area suggest 8% (4-12%) and 10% (6-14%) increases in risk per standard deviation. MRP has potential for studying ecological risk factors based on self-report measures, with greater predictive validity for incidence of psychiatric admissions than current methods.

Assessing local chlamydia screening performance by combining survey and administrative data to account for differences in local population characteristics

Reducing health inequalities requires improved understanding of the causes of variation. Local-level variation reflects differences in local population characteristics and health system performance. Identifying low- and high-performing localities allows investigation into these differences. We used Multilevel Regression with Post-stratification (MRP) to synthesise data from multiple sources, using chlamydia testing as our example. We used national probability survey data to identify individual-level characteristics associated with chlamydia testing and combined this with local-level census data to calculate expected levels of testing in each local authority (LA) in England, allowing us to identify LAs where observed chlamydia testing rates were lower or higher than expected, given population characteristics. Taking account of multiple covariates, including age, sex, ethnicity, student and cohabiting status, 5.4% and 3.5% of LAs had testing rates higher than expected for 95% and 99% posterior credible intervals, respectively; 60.9% and 50.8% had rates lower than expected. Residual differences between observed and MRP expected values were smallest for LAs with large proportions of non-white ethnic populations. London boroughs that were markedly different from expected MRP values (≥90% posterior exceedance probability) had actively targeted risk groups. This type of synthesis allows more refined inferences to be made at small-area levels than previously feasible.

Predicting State Presidential Election Results Using National Tracking Polls and Multilevel Regression with Poststratification (MRP)

Predicting State Presidential Election Results Using National Tracking Polls and Multilevel Regression with Poststratification (MRP)