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.
Expert Panel on the Identification, Evaluation, and Treatment of Overweight in Adults.
Clinical guidelines on the identification, evaluation, and treatment of overweight and obesity in adults: executive summary. Am J Clin Nutr1998;
68: 899–917.
2.
LongfordN.On selection and composition in small area and mapping problems. Stat Methods Med Res2005;
14: 3–16.
3.
TzavidisNRanalliMSalvatiN, et al.Robust small area prediction for counts.Stat Methods Med Res2015;
24: 373–395.
4.
BurgardJKrauseJMünnichR.Adjusting selection bias in German health insurance records for regional prevalence estimation. Popul Health Metrics2019;
17: 1–13.
5.
LiuBParsonsVFeuerE, et al.
Small area estimation of cancer risk factors and screening behaviors in US counties by combining two large national health surveys. Prevent Chronic Disease2019;
16: E119.
6.
MontoyaIEsnaolaSCalvoM, et al.Small area estimation of health indicators based on data from the Basque Country Health Survey. Gaceta Sanit2019;
33: 289–292.
7.
BurgardJKrauseJMünnichR. An elastic net penalized small area model combining unit- and area-level data for regional hypertension prevalence estimation. J Appl Stat. Epub ahead of print 14 May 2020. DOI: 10.1080/02664763.2020.1765323.
8.
RaoJMolinaI. Small area estimation. In: Wiley series in survey methodology. Hoboken, New Jersey: John Wiley & Sons, Inc., 2015.
9.
PfeffermannD.Small area estimation – new developments and directions. Int Stat Rev2002;
70: 55–76.
10.
PfeffermannD.New important developments in small area estimation. Stat Sci2013;
28: 40–68.
11.
PremKPhengSTeoA, et al.Spatial and temporal projections of the prevalence of active tuberculosis in Cambodia. BMJ Global Health2019;
4: e001083.
12.
Pérez-CrespoLCanals-SansJSuades-GonzálezE, et al.Temporal trends and geographical variability of the prevalence and incidence of attention deficit/hyperactivity disorder diagnoses among children in Catalonia, Spain.Sci Rep2020;
10: 6397.
13.
MolinaISaeiALombardíaM.Small area estimates of labour force participation under a multinomial logit mixed model. J R Stat Soc Ser A (Stat Soc)2007;
170: 975–1000.
14.
GhoshMKimDSinhaK, et al.Hierarchical and empirical Bayes small domain estimation and proportion of persons without health insurance for minority subpopulations. Survey Methodol2009;
35: 53–66.
15.
ChenSLahiriP.Inferences on small area proportions.J Ind Soc Agric Stat2012;
66: 121–124.
16.
Erciulescu A and Fuller W. Small area prediction of the mean of a binomial random variable. In: JSM2013conference proceedings -- surveyresearch methods section, Montreal, Canada, 2013, pp.855--863.
17.
López-VizcaínoELombardíaMMoralesD.Multinomial-based small area estimation of labour force indicators. Stat Model2013;
13: 153–178.
18.
López-VizcaínoELombardíaMMoralesD.Small area estimation of labour force indicators under a multinomial model with correlated time and area effects. J R Stat Assoc Ser A (Stat Soc)2015;
178: 535–565.
19.
MilitinoAUgarteMGoicoaT.Deriving small area estimates from information technology business surveys. J R Stat Assoc Ser A (Stat Soc)2015;
178: 1051–1067.
20.
ChambersRSalvatiNTzavidisN.Semiparametric small area estimation for binary outcomes with application to unemployment estimation for Local Authorities in the UK. J R Stat Assoc Ser A (Stat Soc)2016;
179: 453–479.
21.
HobzaTMoralesD.Empirical best prediction under unit-level logit mixed models. J Off Stat2016;
32: 661–692.
22.
LiuBLahiriP.Adaptive hierarchical Bayes estimation of small area proportions. Calcutta Stat Assoc Bull2017;
69: 150–164.
23.
HobzaTMoralesDSantamaríaL.Small area estimation of poverty proportions under unit-level temporal binomial-logit mixed models. Test2018;
27: 270–294.
24.
ChambersRDreassiESalvatiN.Disease mapping via negative binomial regression M-quantiles.Stat Med2014;
33: 4805–4824.
25.
DreassiERanalliMSalvatiN.Semiparametric M-quantile regression for count data. Stat Methods Med Res2014;
23: 591–610.
26.
BoubetaMLombardíaMMoralesD.Empirical best prediction under area-level Poisson mixed models. Test2016;
25: 548–569.
27.
BoubetaMLombardíaMMoralesD.Poisson mixed models for studying the poverty in small areas. Comput Stat Data Anal2017;
107: 32–47.
28.
ShokoohiFTorabiM.Semi-parametric small-area estimation by combining time-series and cross-sectional data methods. Aust New Zealand J Stat2018;
60: 323–342.
29.
World Health Organization. International classification of diseases for mortality and morbidity statistics (11th Revision), 2018.
30.
KendirCvan den AkkerMVosR, et al.Cardiovascular disease patients have increased risk for comorbidity: a cross-sectional study in the Netherlands.Eur J Gen Pract2018;
24: 45–50.
31.
BreslowNClaytonD.Approximate inference in generalized linear mixed models. J Am Stat Assoc1993;
88: 9–25.
32.
HoerlAKennardR.Ridge regression: biased estimation for nonorthogonal problems. Techometrics1970;
12: 55–67.
33.
SchaeferRRoiLWolfeR.A ridge logistic estimator. Commun Stat – Theory Methods1984;
13: 99–113.
34.
Le CessieSVan HouwelingenJ. Ridge estimators in logistic regression. J R Stat Soc Ser C (Appl Stat)1992;
41: 191–201.
35.
PereiraJBastoMda SilvaA.The logistic lasso and ridge regression in predicting corporate failure. Proc Econom Finance2016;
39: 634–641.
LangeCJentschFAllenJ, et al.Data resource profile: German health update (GEDA) – the health interview survey for adults in Germany.Int J Epidemiol2015;
44: 442–450.
38.
Robert Koch Institute. German Health Update 2010 (GEDA 2010). Public Use File third Version, 2013. DOI: 10.7797/27-200910-1-1-3.
39.
Robert Koch Institute. German Health Update 2009 (GEDA 2009). Public Use File second Version, 2014. DOI: 10.7797/26-200809-1-1-2.
40.
Robert Koch Institute. German Health Update 2012 (GEDA 2012). Public Use File first Version, 2014. DOI: 10.7797/29-201213-1-1-1.
41.
RahmouniKCorreiaMHaynesW, et al.Obesity-associated hypertension. Hypertension2005;
45: 9–14.
LinXBreslowN.Bias correction in generalized linear mixed models with multiple components of dispersion. J Am Stat Assoc1996;
91: 1007–1016.
44.
González-ManteigaWLombardíaMMolinaI, et al.Estimation of the mean squared error of predictors of small area linear parameters under a logistic mixed model. Comput Stat Data Anal2007;
51: 2720–2733.
45.
González-ManteigaWLombardíaMMolinaI, et al.Bootstrap mean squared error of small-area EBLUP. J Stat Comput Simul2008;
78: 443–462.
46.
DemidenkoE.Mixed models. Theory and applications with R.
Hoboken:
John Wiley & Sons, 2013.
47.
SchwarzG.Estimating the dimension of a model. Ann Stat1978;
6: 461–464.
48.
CravenPWahbaG.Smoothing noisy data with spline functions: estimating the correct degree of smoothing by the method of generalized cross-validation. Numer Math1979;
31: 377–403.
49.
AkaikeH.A new look at the statistical model identification. IEEE Trans Autom Control1974;
19: 716–723.
50.
JiangJ.Linear and generalized linear mixed models and their applications.
New York, NY:
Springer, 2007.
51.
TheobaldC.Generalizations of mean squared error applied to ridge regression. J R Stat Soc Ser B (Methodol)1974;
36: 103–106.
52.
YamashitaTYamashitaKKamimuraR.A stepwise AIC method for variable selection in linear regression. Commun Stat – Theory Methods2007;
36: 2395–2403.
53.
FriedmanJHastieTHöflingH, et al.Regularization paths for generalized linear models via coordinate descent. J Stat Softw2010;
33: 1–22.
54.
ZouHHastieT.Regularization and variable selection via the elastic net. J R Stat Soc Ser B (Methodol)2005;
67: 301–320.
55.
TibshiraniR.Regression shrinkage and selection via the lasso. J R Stat Soc Ser B (Methodol)1996;
58: 267–288.
56.
FriedmanJHastieTHöflingH, et al.Pathwise coordinate optimization. Ann Appl Stat2007;
1: 302–332.
57.
DevauxMSassiFChurchJ, et al.Exploring the relationship between education and obesity. OECD J Econom Stud2011;
1: 121–159.
58.
TraversyGChaputJP.Alcohol consumption and obesity: an update. Curr Obes Rep2015;
4: 122–130.
59.
MolinaIRaoJ.Small area estimation of poverty indicators. Canad J Stat2010;
38: 369–385.
60.
MolinaINandramBRaoJ.Small area estimation of general parameters with application to poverty indicators: a hierarchical Bayes approach. Ann Appl Stat2014;
8: 852–885.
VidraNTrias-LlimósSJanssenF.Impact of obesity on life expectancy among different European countries: secondary analysis of population-level data over the 1975-2012 period. BMJ Open2019;
9: 1–8.
63.
BurgardJKrauseJKreberD, et al. The generalized equivalence of regularization and min-max robustification in linear mixed models. Stat Pap. DOI: 10.1007/s00362-020-01214-z.
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
