Proceedings. American Statistical Association. Annual Meeting最新文献

A change-point model is essential in longitudinal data to infer an individual specific time to an event that induces a change of trend. However, in general, change points are not known for population-based data. We present an unknown change-point model that fits the linear and non-linear mixed effects for pre- and post-change points. We address the left-censored observations. Through stochastic approximation expectation maximization (SAEM) with the Metropolis Hasting sampler, we fit a random change-point non-linear mixed effects model. We apply our method on the longitudinal viral load (VL) data reported to the HIV surveillance registry from New York City.

Among many approaches for selecting match control cases, few methods exist for natural experiments (Li, Zaslavsky & Landrum, 2007), especially when studying clustered or hierarchical data. The lack of randomization of treatment exposure gives importance to using proper statistical procedures that control for individual differences. In this natural experimental study, which has a hierarchical structure, we plan to evaluate the efforts of 455 counties across the United States to make targeted efforts to improve mental health services and reduce jail utilization over time. Nested within states, counties are clustered on health and social indicators, which affect the likelihood of making improvements in these areas. Similar to a randomized trial, prior to collecting survey data, it is necessary to identify matched control counties as study sites based on an array of state and county covariates. Accounting for the hierarchal structure of data, a blend of various probability-based models are presented to achieve this goal. Methods include multivariable models that control for observed differences among treatment and control groups, shrinkage based LASSO as a variable selection technique, and logistic models.

Adaptive design principles are applied to the National Immunization Survey-Teen (NIS-Teen), sponsored by Centers for Disease Control and Prevention, which monitors vaccination coverage of U.S. adolescents age 13-17 years. Data collection is ongoing in two phases: (1) a random-digit-dial telephone survey to interview parents/guardians with age-eligible adolescents, followed by (2) a mail survey to vaccination providers, called the provider record check (PRC), to obtain vaccination histories for the adolescents. A logistic regression model relating the probability that an Immunization History Questionnaire (IHQ) is returned for a teen-provider pair to characteristics of the adolescent, mother, household, and providers was fit. R-indicators and partial R-indicators for the PRC phase of the 2015 NIS-Teen are presented to evaluate the representativeness of response in the PRC. The indicators are visualized using interactive graphics embodied in an R Shiny application to track the real time changes. Programmatic interventions to improve representativeness are discussed, which include strategies for prompting providers and special treatment of certain subgroups.

In on-road driving behavior studies, vehicle acceleration is sampled at high frequencies and then reduced to meaningful metrics over short driving segments. We examined road test data from 65 subjects driving over a common route, as well as driving in naturalistic situations using their own vehicle. We isolated 24-second segments, then reduced the accelerometer data via two methods: 1) standard deviation (SD) within a segment, and 2) re-centering parameter from a time series model previously developed for driving simulator data. We analyzed the data via random effects models to ascertain the intraclass correlations (ICC's) of the metrics. With and without adjusting for speed, the ICC of SD within a segment tended to be much greater than the ICC of the re-centering parameter for the segment (range: 0-30% vs. 0-1%). Also, ICC's from the naturalistic driving data tended to be greater than the fixed-route data (range: 0-27% vs. 0-9%), which could reflect individuals exhibiting their more usual driving behavior in naturalistic environments. Findings illustrate the challenges of identifying meaningful driving metrics and comparing these across different epochs, road segments and research platforms.

The National Immunization Surveys (NIS) include dual frame random-digit-dial telephone surveys used to monitor vaccination coverage in the United States among children age 19-35 months (NIS-Child) and adolescents age 13-17 years (NIS-Teen), and to assess influenza vaccination for children age 6 months-17 years (NIS-Flu). The surveys collect household-reported demographic and access-to-care data during telephone interviews with the survey-eligible child's parent or guardian. The parent or guardian is then asked for consent to contact the child's vaccination provider(s) to obtain a provider-reported immunization history using a mailed questionnaire. The success of the NIS relies heavily on getting a respondent to answer the telephone, and the caller ID display is the earliest opportunity to convey information to a respondent about the identity of the caller. An evaluation was conducted in Quarter 4 of 2017 to determine the impact on contact rates of using an alternate caller ID display. The caller ID for the NIS surveys was previously set to display "NORC UCHICAGO", identifying the contractor administering the surveys, with a Chicago-based telephone number. It was hypothesized that having the caller ID display the name of the more recognizable survey sponsor instead of the contractor would increase contact rates. Half of the sample was randomly flagged to display the "NORC UCHICAGO" caller ID text as a control, and the other half was flagged to display "CDC NATL IMMUN" as a treatment. This paper presents the study design, results, conclusions, limitations, and recommendations for future research.

Linking the National Hospital Care Survey (NHCS) with the National Death Index (NDI) provides information on the outcomes of hospitalizations and allows for analysis of individual and provider characteristics associated with in-hospital and post-discharge mortality. We test the viability of confirming hospital mortality through the linkage of preliminary 2011 NHCS data for "known dead" inpatient discharges (i.e., patients that died during a hospitalization) with the NDI, assessing the true match rate and the quality of the match. We then expand the analysis to identify patients with a 30-, 60-, and 90-day post-discharge mortality. The true match rate for the "known dead" is 94 percent.

The Fligner and Verducci (1988) multistage model for rankings is modified to create the moving average maximum likelihood estimator (MAMLE), a locally smooth estimator that measures stage-wise agreement between two long ranked lists, and provides a stopping rule for the detection of the endpoint of agreement. An application of this MAMLE stopping rule to bivariate data set in tau-path order (Yu, Verducci and Blower (2011)) is discussed. Data from the National Cancer Institute measuring associations between gene expression and compound potency are studied using this application, providing insights into the length of the relationship between the variables.

In random digit dial (RDD) telephone surveys, advance letters mailed prior to dialing sampled telephone numbers may increase survey response rates (de Leeuw et al. 2007). The ability to mail advance letters to RDD samples relies on the availability of addresses that matched to the sampled telephone numbers. Traditionally, address matching was possible only for landline telephone samples with directory listings, which are not generally available for cell telephone numbers. It is now possible to obtain mailing addresses for a sizeable proportion of cell telephone numbers. Since cell telephone samples are now an increasingly large part of RDD telephone surveys, the use of advance letters mailed prior to dialing cell telephone numbers may result in an increase in response rates similar to those seen for landline telephone numbers. To test this possibility, mailing addresses were obtained for samples of landline and cell telephone numbers in the 2013 National Immunization Survey, a large, national, dual-frame RDD survey sponsored by the Centers for Disease Control and Prevention and fielded by NORC at the University of Chicago. Prior to dialing, advance letters were mailed to half of the cases in the landline and cell telephone samples with available addresses. In this study, we compared address match rates and address accuracy rates between the landline and cell telephone samples and measured the effect of the advance letter on survey response rates in the landline and cell telephone samples. We found that while advance letters had a positive effect on screener completion in the landline sample, they did not impact screener completion in the cell telephone sample. The lack of effect in the cell telephone sample may be due to a higher rate of inaccurate address matching than in the landline telephone sample: in the cell telephone sample, recently-updated addresses were found to be more accurate, and when the analysis was restricted to advance letters mailed to recently-updated addresses, the impact on screener completion in the cell telephone sample was similar to that in the landline sample. We also found that advance letters had a larger positive effect on interview completion in the landline sample, but sample sizes in the cell telephone sample for the experiment were too small to evaluate the impact on interview completion. Implications of these results for dual-frame RDD telephone surveys will be discussed.

Wait time is the differences between the time a patient arrives in the emergency department (ED) and the time an ED provider examines that patient. This study focuses on the development of a negative binomial model to examine factors associated with ED wait time using the National Hospital Ambulatory Medical Care Survey (NHAMCS). Conducted by National Center for Health Statistics (NCHS), NHAMCS has been gathering, analyzing, and disseminating information annually about visits made for medical care to hospital outpatient department and EDs since 1992. To analyze ED wait times, a negative binomial model was fit to the ED visit data using publically released micro data from the 2009 NHAMCS. In this model, the wait time is the dependent variable while hospital, patient, and visit characteristics are the independent variables. Wait time was collapsed into discrete values representing 15 minutes intervals. The findings are presented.

For the problem of assessing initial agreement between two rankings of long lists, inference in the Fligner and Verducci (1988) multistage model for rankings is modified to provide a locally smooth estimator of stage-wise agreement. An extension to the case of overlapping but different sets of items in the two lists, and a stopping rule to identify the endpoint of agreement, are also provided. Simulations show that this approach performs very well under several conditions. The methodology is applied to a database of popular names for newborns in the United States and provides insights into trends as well as differences in naming conventions between the two sexes.