Evaluation of the performance of smoothing functions in generalized additive models for spatial variation in disease

US County-Level Variation in Availability and Prevalence of Black Physicians in 1906

Importance

Black physicians are substantially underrepresented in the US health care workforce, with detrimental effects on the health and health care experiences of Black individuals. These contemporary gaps can be traced to the early days of the medical profession using the first edition of the American Medical Directory (AMD).

Objective

To identify state- and county-level patterns related to the training and availability of Black physicians relative to their White counterparts in the 1906 AMD.

Design, Setting, and Participants

For this cross-sectional study, data for 41 828 physician entries in 18 US states in or adjacent to the South as well as the District of Columbia were extracted from the 1906 AMD and aggregated to 1570 counties. Data analysis was performed between September 2023 and January 2024.

Exposures

County-level exposure variables included population density, racial composition, and illiteracy rate among US-born White residents as well as an index of terrain ruggedness and the number of lynchings in the previous decade. Median values of physicians' distance from place of practice to place of medical training (by race of physician) were also used as an exposure variable.

Main Outcomes and Measures

There were 4 county-level outcomes: (1) presence of any Black physician, (2) proportion of Black physicians per Black population, (3) proportion of White physicians per White population, and (4) community representativeness (reported as the community representativeness ratio). The cross-sectional analysis used generalized additive mixed models with state-level random effects.

Results

Across 1570 counties, Black physicians comprised 746 (1.8%) of the 41 828 physicians in the dataset. Black physicians tended to train further from their place of practice than their White counterparts. The proportion of Black physicians per 1000 Black residents was 0.08 compared with 1.62 for White physicians; these proportions varied substantially by state. At the county level, the presence of any Black physician was associated with percentage Black population (odds ratio [OR], 28.94 [95% CI, 9.77 to 85.76]; P ≤ .001), population density (OR, 2.63 [95% CI, 2.03 to 3.40]; P ≤ .001), and distance to the nearest Black medical school (OR, 0.62 [95% CI, 0.42 to 0.92]; P = .02).

Conclusions and Relevance

A variety of structural disadvantages are illustrated in this cross-sectional study of county-level sociodemographic and geographic characteristics associated with the prevalence of Black physicians in the earliest days of the profession. To demonstrate its broader utility for health disparities research, the dataset has been made publicly available with a visualization platform.

Data Imputation and Body Weight Variability Calculation Using Linear and Nonlinear Methods in Data Collected From Digital Smart Scales: Simulation and Validation Study

Background

Body weight variability (BWV) is common in the general population and may act as a risk factor for obesity or diseases. The correct identification of these patterns may have prognostic or predictive value in clinical and research settings. With advancements in technology allowing for the frequent collection of body weight data from electronic smart scales, new opportunities to analyze and identify patterns in body weight data are available.

Objective

This study aims to compare multiple methods of data imputation and BWV calculation using linear and nonlinear approaches

Methods

In total, 50 participants from an ongoing weight loss maintenance study (the NoHoW study) were selected to develop the procedure. We addressed the following aspects of data analysis: cleaning, imputation, detrending, and calculation of total and local BWV. To test imputation, missing data were simulated at random and using real patterns of missingness. A total of 10 imputation strategies were tested. Next, BWV was calculated using linear and nonlinear approaches, and the effects of missing data and data imputation on these estimates were investigated.

Results

Body weight imputation using structural modeling with Kalman smoothing or an exponentially weighted moving average provided the best agreement with observed values (root mean square error range 0.62%-0.64%). Imputation performance decreased with missingness and was similar between random and nonrandom simulations. Errors in BWV estimations from missing simulated data sets were low (2%-7% with 80% missing data or a mean of 67, SD 40.1 available body weights) compared with that of imputation strategies where errors were significantly greater, varying by imputation method.

Conclusions

The decision to impute body weight data depends on the purpose of the analysis. Directions for the best performing imputation methods are provided. For the purpose of estimating BWV, data imputation should not be conducted. Linear and nonlinear methods of estimating BWV provide reasonably accurate estimates under high proportions (80%) of missing data.

Spatial analyses of ALS incidence in Denmark over three decades

OBJECTIVE

Amyotrophic lateral sclerosis (ALS) is a neurodegenerative disease of the motor neuron with very few known risk factors. We conducted a spatial epidemiologic analysis of ALS incidence in Denmark to assess the contribution of sociodemographic determinants to geographic variation.

METHODS

We analyzed 4249 ALS cases (1982-2013), each with 100 controls matched on sex and birth year. Odds ratio and 95% confidence bands at birth and diagnosis/index locations were calculated using generalized additive models. We included a bivariate spatial smooth for location in our conditional logistic regression adjusted for socioeconomic status and marital status. We also conducted analyses adjusted for both birth and diagnosis addresses to separate location effects.

RESULTS

We observed significantly elevated ALS odds near Copenhagen for both the birth and diagnosis period analyses. Sociodemographic factors did not explain the observed patterns. When we further adjusted our spatial analyses by including both birth and diagnosis addresses, the significant area of elevated male ALS odds by birth address shifted to northwest Denmark away from Copenhagen, and there was little evidence of variation among women. Geographic variation at diagnosis differed between male and females, suggesting that patterns are not just due to regional variation in case ascertainment.

CONCLUSION

ALS incidence in Denmark is associated with both location at birth and diagnosis, suggesting that geographic variation may be due to exposures occurring at birth or closer to diagnosis, although the latter could relate to case ascertainment issues.