Background: Data from the memberships of large, integrated health care systems can be valuable for clinical, epidemiologic, and health services research, but a potential selection bias may threaten the inference to the population of interest.Methods: We reviewed administrative records of members of Kaiser Permanente Southern California (KPSC) in 2000 and 2010, and we compared their sociodemographic characteristics with those of the underlying population in the coverage area on the basis of US Census Bureau data.Results: We identified 3,328,579 KPSC members in 2000 and 3,357,959 KPSC members in 2010, representing approximately 16% of the population in the coverage area. The distribution of sex and age of KPSC members appeared to be similar to the census reference population in 2000 and 2010 except with a slightly higher proportion of 40 to 64 year olds. The proportion of Hispanics/Latinos was comparable between KPSC and the census reference population (37.5% vs 38.2%, respectively, in 2000 and 45.2% vs 43.3% in 2010). However, KPSC members included more blacks (14.9% vs 7.0% in 2000 and 10.8% vs 6.5% in 2010). Neighborhood educational levels and neighborhood household incomes were generally similar between KPSC members and the census reference population, but with a marginal underrepresentation of individuals with extremely low income and high education.Conclusions: The membership of KPSC reflects the socioeconomic diversity of the Southern California census population, suggesting that findings from this setting may provide valid inference for clinical, epidemiologic, and health services research.
OBJECTIVE -The purpose of this study was to assess changes in the prevalence of preexisting diabetes (diabetes antedating pregnancy) and gestational diabetes mellitus (GDM) from 1999 through 2005.RESEARCH DESIGN AND METHODS -In this retrospective study of 175,249 women aged 13-58 years with 209,287 singleton deliveries of Ն20 weeks' gestation from 1999 through 2005 in all Kaiser Permanente hospitals in southern California, information from clinical databases and birth certificates was used to estimate the prevalence of preexisting diabetes and GDM.RESULTS -Preexisting diabetes was identified in 2,784 (1.3%) of all pregnancies, rising from an age-and race/ethnicity-adjusted prevalence of 0.81 per 100 in 1999 to 1.82 per 100 in 2005 (P trend Ͻ 0.001). Significant increases were observed in all age-groups and all racial/ethnic groups. After women with preexisting diabetes were excluded, GDM was identified in 15,121 (7.6%) of 199,298 screened pregnancies. The age-and race/ethnicity-adjusted GDM prevalence remained constant at 7.5 per 100 in 1999 to 7.4 per 100 in 2005 (P trend ϭ 0.07). Among all deliveries to women with either form of diabetes, 10% were due to preexisting diabetes in 1999, rising to 21% in 2005, with GDM accounting for the remainder.CONCLUSIONS -The stable prevalence of GDM and increase in the prevalence of preexisting diabetes were independent of changes in the age and race/ethnicity of the population. The increase in preexisting diabetes, particularly among younger women early in their reproductive years, is of concern.
BackgroundLog-binomial and robust (modified) Poisson regression models are popular approaches to estimate risk ratios for binary response variables. Previous studies have shown that comparatively they produce similar point estimates and standard errors. However, their performance under model misspecification is poorly understood.MethodsIn this simulation study, the statistical performance of the two models was compared when the log link function was misspecified or the response depended on predictors through a non-linear relationship (i.e. truncated response).ResultsPoint estimates from log-binomial models were biased when the link function was misspecified or when the probability distribution of the response variable was truncated at the right tail. The percentage of truncated observations was positively associated with the presence of bias, and the bias was larger if the observations came from a population with a lower response rate given that the other parameters being examined were fixed. In contrast, point estimates from the robust Poisson models were unbiased.ConclusionUnder model misspecification, the robust Poisson model was generally preferable because it provided unbiased estimates of risk ratios.Electronic supplementary materialThe online version of this article (10.1186/s12874-018-0519-5) contains supplementary material, which is available to authorized users.
OBJECTIVETo estimate the prevalence of postpartum glucose testing within 6 months of pregnancies complicated by gestational diabetes mellitus (GDM), assess factors associated with testing and timing of testing after delivery, and report the test results among tested women.RESEARCH DESIGN AND METHODSThis was a retrospective study of 11,825 women who were identified as having GDM using the 100-g oral glucose tolerance test (OGTT) from 1999 to 2006. Postpartum testing (75-g 2-h OGTT or fasting plasma glucose [FPG]) within 6 months of delivery and test results from laboratory databases are reported. Postpartum test results are categorized as normal, impaired fasting glucose (IFG) and/or impaired glucose tolerance (IGT), and provisionally diabetic.RESULTSAbout half (n = 5,939) the women were tested with either a FPG or 75-g OGTT from 7 days to 6 months postpartum. Of these women, 46% were tested during the 6- to 12-week postpartum period. Odds of testing were independently associated with age, race/ethnicity, household income, education, foreign-born status, parity, mode of delivery, having a postpartum visit, having GDM coded at discharge, and pharmacotherapy for GDM. Of the 5,857 women with test results, 16.3% (n = 956) had IFG/IGT and 1.1% (n = 66) had provisional diabetes. After adjustment for demographic and clinical factors, abnormal postpartum test results was associated with having required insulin, glyburide, or metformin during pregnancy and with longer period from delivery to postpartum testing.CONCLUSIONSAfter a pregnancy complicated by GDM, automated orders for postpartum testing with notification to physicians and electronically generated telephone and e-mail reminder messages to patients may improve the rates of postpartum testing for persistence of glucose intolerance.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.