Biomedical subjects

Jason Roy

Publications and source records attributed to Jason Roy.

At least 19 recordsLinked to original sources

PubMed2006

Effects of childhood primary hypertension on carotid intima media thickness: a matched controlled study.

To determine whether carotid intima media thickness is increased in children with primary hypertension, the current study compared carotid intima media thickness in hypertensive children with that of normotensive control subjects matched closely for body mass index and determined the relationship between carotid intima media thickness and hypertension severity determined by ambulatory blood pressure monitoring. Children with newly diagnosed office hypertension (n=28) had carotid intima media thickness, left ventricular mass index, and ambulatory blood pressure monitoring performed. Carotid intima media thickness was performed in normotensive control subjects (n=28) matched pairwise to hypertensive subjects for age (+/-1 year), gender, and body mass index (+/-10%). Eighty-two percent of subjects were overweight or obese (body mass index > or =85th percentile). The median carotid intima media thickness of hypertensive subjects was greater than that of matched controls (0.67 versus 0.63 mm; P=0.045). In the hypertensive subjects, carotid intima media thickness correlated strongly with several ambulatory blood pressure monitoring parameters, with the strongest correlation for daytime systolic blood pressure index (r=0.57; P=0.003). In the hypertensive group, the prevalence of left ventricular hypertrophy was 32%, but unlike carotid intima media thickness, left ventricular mass index did not correlate with ambulatory blood pressure monitoring. Together, the findings that hypertensive subjects had increased carotid intima media thickness compared with matched controls and that higher carotid intima media thickness correlated with more severe hypertension by ambulatory blood pressure monitoring provide strong evidence that carotid intima media thickness is increased in childhood primary hypertension, independent of the effects of obesity.

Adolescent↗

PubMed2006

The changing risk profile of the American adolescent smoker: implications for prevention programs and tobacco interventions.

PURPOSE: To determine how the association between cigarette smoking and other risky behaviors, such as substance use, violence, and risky sexual practices, has changed between 1991 and 2003. METHODS: Youth Risk Behavior Surveys (YRBS) from 1991 to 2003 were analyzed. For each cohort, logistic regression models controlling for gender, race/ethnicity, and school grade were used to describe the associations between smoking and other risky behaviors. Changes in the odds ratios over time were confirmed with a trend analysis. RESULTS: The strength of the relationship between smoking and other risky behaviors increased for lifetime number of sexual partners (1991 odds ratio [OR] 1.49; 2003 OR 1.61 (p < .001)), sexual partners in the past 3 months (1991 OR 1.77; 2001 OR 2.05 (p < .001)), and never wearing a bicycle helmet 1991 OR 1.40; 1997 OR 5.94 (p < .001). Increases were also seen for binge drinking, and physical fighting. The association between cigarette smoking and marijuana use decreased slightly. CONCLUSIONS: Future prevention efforts and tobacco intervention programs should recognize that current adolescent smokers are even more likely to engage in risky sexual behavior, risky alcohol-related behaviors, and to not use a seatbelt or bicycle helmet than were adolescents in the early nineties.

Adolescent↗

PubMed2006

Promoting independence for wheelchair users: the role of home accommodations.

PURPOSE: The objective of this research is to investigate whether home accommodations influence the amount of human help provided to a nationally representative sample of adults who use wheelchairs. DESIGN AND METHODS: We analyzed data from the Adult Disability Follow-back Survey (DFS), Phase II, of the Disability Supplement to the 1994-1995 National Health Interview Surveys (NHIS-D). The analytic sample consisted of 899 adults aged 18 and older who reported using wheelchairs in the previous 2 weeks. We conducted logistic regression and ordinary least squares (OLS) regression analyses to test the influence of home accommodations on the receipt of any human help, and among respondents who received help, on the hours of help received, respectively. We analyzed paid and unpaid help separately. RESULTS: Home accommodations were related to the receipt of unpaid, but not paid, help. Relative to having no home accommodations, the presence of each additional accommodation decreased the odds of having unpaid help by 14% (OR =.86; 95% CI =.76,.97). Additionally, we observed an inverse relationship between the number of accommodations in the home and hours of unpaid help (p <.01). For wheelchair users who live alone, specific types of home accommodations were also inversely related to hours of unpaid help. IMPLICATIONS: Policies that reimburse for home accommodations may be an efficient response to the growing demand for home-care support while enabling greater autonomy and independence for people who use wheelchairs.

Activities of Daily Living↗

PubMed2006

The impact of conjugate pneumococcal vaccination on routine childhood vaccination and primary care use in 2 counties.

BACKGROUND: Pneumococcal conjugate vaccine immunization recommendations were rapidly implemented by primary care providers. Before the recommendations, concern was expressed that adding pneumococcal conjugate vaccine might result in delays in other vaccinations or preventive services. OBJECTIVES: The study objectives were to measure whether incorporation of pneumococcal conjugate vaccine by primary care providers delayed other vaccinations or added primary health care visits. DESIGN AND METHODS: In 2 counties surrounding Rochester and Nashville, we reviewed a representative sample of primary care charts for children born before and after licensure of pneumococcal conjugate vaccine. Receipt of vaccinations and health care visits were compared for the 2 age-matched cohorts. RESULTS: We reviewed 1459 records from Rochester and 1857 records from Nashville. The pre-pneumococcal conjugate vaccine and post-pneumococcal conjugate vaccine cohorts had similar demographic characteristics. The median age for receipt of any vaccination was not older for the postvaccine cohort than for the prevaccine cohort in either community. The percentage of children up-to-date for vaccinations by 18 months for postvaccine versus prevaccine cohorts was similar in Rochester (72% in each cohort) and in Nashville (58% postvaccine and 65% prevaccine). The number of well-child care visits or other health care visits during the first 18 months of life was not statistically different between the 2 cohorts. CONCLUSIONS: Implementation of pneumococcal conjugate vaccine was not associated with delays in other childhood vaccinations or more primary care visits.

Child, Preschool↗

PubMed2005

The effect of provider-level ascertainment bias on profiling nursing homes.

Profiling health care providers for the purpose of public reporting and quality improvement has become commonplace. Recently, the Centers for Medicare and Medicaid Services (CMS) began publishing measures of quality for every Medicare/Medicaid-certified nursing home in the country. The facility-specific quality indicators (QIs) reported by CMS are based on quarterly measures from the minimum data set (MDS). However, some QIs from the MDS are potentially subject to ascertainment bias. Ascertainment bias would occur if there was variation in the way items that make up QIs are measured by nurses from each facility. This is potentially a problem for difficult-to-measure items such as pain and pressure ulcers. To assess the impact of ascertainment bias on profiling, we utilize data from a reliability study of nursing homes from six states. We develop methods for profiling providers in situations where the data consist of a response variable for each subject based on assessments from an internal rater, and, for a subset of subjects in each facility, a response variable based on assessments from an independent (external) rater. The internal assessments are potentially subject to provider-level ascertainment bias, whereas the independent assessments are considered the 'gold standard'. Our methods extend popular Bayesian approaches for profiling by using the paired observations from the subset of subjects with error-prone and error-free assessments to adjust for ascertainment bias. We apply the methods to MDS merged with the reliability data, and compare the bias-corrected profiles with those of standard approaches.

Aged↗

PubMed2005

Impact of cognitive function on assessments of nursing home residents' pain.

OBJECTIVES: We sought to examine the impact of residents' cognitive function on the quality of Minimum Data Set (MDS) pain data using the latent variable approach. RESEARCH DESIGN: Using the Resident Assessment Instrument (RAI) protocol, nursing home (NH) staff and well-trained study nurses independently assessed 3736 NH residents. MEASURES: Inter-rater agreement of pain ratings between NH staff and study nurses was quantified by weighted kappas and polychoric correlations and compared among groups of residents with no/mild, moderate, and severe cognitive impairment. Probit models were built to examine the effect of residents' cognitive function on thresholds raters used to rate pain. RESULTS: Of 3736 residents, 40.4% had no/mild, 35.9% moderate, and 23.7% severe cognitive impairment. Both NH staff and study nurses recorded less frequent and less severe pain for residents with more severe cognitive impairment. The inter-rater agreement on pain ratings between NH staff and study nurses was good-weighted kappas were greater than 0.5 and polychoric correlations greater than 0.7. The thresholds raters used to record pain were similar for NH staff and study nurses and progressively increased when raters recorded pain for residents with more severe cognitive impairment. CONCLUSIONS: Given the RAI protocols, the quality of MDS pain data collected by NH staff was similar to that of well-trained nurses regardless of residents' cognitive function. Our results strongly support the notion that specialized pain assessment instruments are needed to adequately detect pain for the large proportion of cognitive impaired NH residents.

Aged↗

PubMed2005

The quality of the quality indicator of pain derived from the minimum data set.

OBJECTIVE: To examine facility variation in data quality of the level of pain documented in the minimum data set (MDS) as a function of level of hospice enrollment in nursing homes (NHs). DATA SOURCE: Clinical assessments on 3,469 nonhospice residents from 178 NHs were merged with On-line Survey Certification and Reporting data of 2000, Medicare Claims data of 2000 and the MDS of 2000-2002. STUDY DESIGN: Using the same assessment protocol, NH staff and study nurses independently assessed 3,469 nonhospice residents. Study nurses' assessments being gold standard, we quantified and compared quality of NH staff's pain rating across NHs with high, medium, or low hospice use. Multilevel models were built to assess the effect of NH hospice use levels on the occurrence of false positive (FP) and false negative (FN) errors in NH-rated "severe pain." PRINCIPAL FINDINGS: Of 178 NHs, 25 had medium and 41 high hospice use. NHs with higher hospice use had lower sensitivities. In multilevel analysis, we found a significant facility-level variation in the probability of FP and FN errors in facility-rated "severe pain." Resident characteristics only explained 4 and 0 percent of the facility variation in FP and FN, respectively; characteristics and locations (state) of NHs further explained 53 and 52 percent of the variance. After controlling for resident and NH characteristics, staff in NHs with medium or high hospice use were less likely to have FP or FN errors in their MDS documentation of pain than were staff in NHs with low or no hospice use. CONCLUSIONS: The examination of data quality of pooled MDS data from multiple NHs is insufficient. Multilevel analysis is needed to elucidate sources of heterogeneity in the quality of MDS data across NHs. Facility characteristics, e.g., hospice use or NH location, are systematically associated with overrated/underrated pain and may bias pain quality indicator (QI) comparisons. To ensure the integrity of QI comparison in the NH setting, the government may need to institute regular audits of MDS data quality.

Aged↗

PubMed2005

Bereaved family member perceptions of quality of end-of-life care in U.S. regions with high and low usage of intensive care unit care.

OBJECTIVES: To compare the quality of end-of-life care of persons dying in regions of differing practice intensity. DESIGN: Mortality follow-back survey. SETTING: Geographic regions in the highest and lowest deciles of intensive care unit (ICU) use. PARTICIPANTS: Bereaved family member or other knowledgeable informants. MEASUREMENTS: Unmet needs, concerns, and rating of quality of end-of-life care in five domains (physical comfort and emotional support of the decedent, shared decision-making, treatment of the dying person with respect, providing information and emotional support to family members). RESULTS: Decedents in high- (n=365) and low-intensity (n=413) hospital service areas (HSAs) did not differ in age, sex, education, marital status, leading causes of death, or the degree to which death was expected, but those in the high-intensity ICU HSAs were more likely to be black and to live in nonrural areas. Respondents in high-intensity HSAs were more likely to report that care was of lower quality in each domain, and these differences were statistically significant in three of five domains. Respondents from high-intensity HSAs were more likely to report inadequate emotional support for the decedent (relative risk (RR)=1.2, 95% confidence interval (CI)=1.0-1.4), concerns with shared decision-making (RR=1.8, 95% CI=1.0-2.9), inadequate information about what to expect (RR=1.5, 95% CI=1.3-1.8), and failure to treat the decedent with respect (RR=1.4, 95% CI=1.0-1.9). Overall ratings of the quality of end-of-life care were also significantly lower in high-intensity HSAs. CONCLUSION: Dying in regions with a higher use of ICU care is not associated with improved perceptions of quality of end-of-life care.

Aged↗

PubMed2005

Missing covariates in longitudinal data with informative dropouts: bias analysis and inference.

We consider estimation in generalized linear mixed models (GLMM) for longitudinal data with informative dropouts. At the time a unit drops out, time-varying covariates are often unobserved in addition to the missing outcome. However, existing informative dropout models typically require covariates to be completely observed. This assumption is not realistic in the presence of time-varying covariates. In this article, we first study the asymptotic bias that would result from applying existing methods, where missing time-varying covariates are handled using naive approaches, which include: (1) using only baseline values; (2) carrying forward the last observation; and (3) assuming the missing data are ignorable. Our asymptotic bias analysis shows that these naive approaches yield inconsistent estimators of model parameters. We next propose a selection/transition model that allows covariates to be missing in addition to the outcome variable at the time of dropout. The EM algorithm is used for inference in the proposed model. Data from a longitudinal study of human immunodeficiency virus (HIV)-infected women are used to illustrate the methodology.

Algorithms↗

PubMed2005

Telemedicine reduces absence resulting from illness in urban child care: evaluation of an innovation.

BACKGROUND: Common acute illness challenges everyone involved in child care. Impoverished inner-city families, whose children are most burdened by morbidity and whose reliance on child care is most important, are those least equipped to deal with this challenge. OBJECTIVE: To assess the impact of telemedicine on absence from child care due to illness (ADI). DESIGN/METHODS: A before-and-after design with historical and concurrent controls was used to study ADI in 5 inner-city child care centers in Rochester, New York, between January 1, 2001, and June 30, 2003. Enrollment averaged 138 children per center, of whom Medicaid covered 66%. Center 5 provided only concurrent controls. Telemedicine service began in the first 4 centers in a staggered fashion starting in May 2001. Baseline data on ADI before availability of telemedicine were collected in each center for a minimum of 18 weeks. The telemedicine model for diagnosis and treatment of common acute problems involved both real-time and store-and-forward information exchange between a child and telemedicine assistant in child care and an office-based telemedicine clinician. Devices used were an all-purpose digital camera (with attachments designed to facilitate capture of ear, nose, throat, skin, and eye images) and an electronic stethoscope. ADI indexed illness that had interrupted care and education for children and burdened both parents and the community with work loss and health care-related costs. Detailed attendance records and staff and parent interviews provided data. The total number of days of attendance expected from all registered children over the course of a week (total child-days) served as the denominator in calculating rates for ADI. The center-week served as the primary unit of analysis. This study is descriptive in character; statistics are not inferential but instead serve to summarize observations. RESULTS: For the 400 weeks of valid observations contributed by the 5 centers, the mean ADI was 6.41 absences per 100 child-days per week. In bivariate analysis, predictors of ADI were children's mean age, child care center, proportion of children covered by Medicaid, season of the year, and availability of telemedicine. ADI during weeks with telemedicine (4.07 absences per 100 child-days) was less than half that during weeks without telemedicine (8.78 absences per 100 child-days). After adjusting for potentially confounding variables using the generalized estimating equations method, telemedicine remained the strongest predictor of ADI. A 63% reduction in ADI was attributable to telemedicine, an effect similar to the 59% variation in ADI with season of the year. During the 201 total weeks that telemedicine services were available, 940 telemedicine encounters occurred. Telemedicine clinicians for these 940 encounters recommended exclusion from child care for 7.0% and in-person visits for 2.8% of the children. In surveys, parents indicated that 91.2% of telemedicine contacts allowed them to stay at work and that 93.8% of problems managed by telemedicine would otherwise have led to an office or emergency department visit. CONCLUSIONS: Telemedicine holds substantial potential to reduce the impact of illness on health and education of children, on time lost from work in parents, and on absenteeism in the economy.

Absenteeism↗

PubMed2004

Handling drop-out in longitudinal studies.

Drop-out is a prevalent complication in the analysis of data from longitudinal studies, and remains an active area of research for statisticians and other quantitative methodologists. This tutorial is designed to synthesize and illustrate the broad array of techniques that are used to address outcome-related drop-out, with emphasis on regression-based methods. We begin with a review of important assumptions underlying likelihood-based and semi-parametric models, followed by an overview of models and methods used to draw inferences from incomplete longitudinal data. The majority of the tutorial is devoted to detailed analysis of two studies with substantial rates of drop-out, designed to illustrate the use of effective methods that are relatively easy to apply: in the first example, we use both semi-parametric and fully parametric models to analyse repeated binary responses from a clinical trial of smoking cessation interventions; in the second, pattern mixture models are used to analyse longitudinal CD4 counts from an observational cohort study of HIV-infected women. In each example, we describe exploratory analyses, model formulation, estimation methodology and interpretation of results. Analyses of incomplete data requires making unverifiable assumptions, and these are discussed in detail within the context of each application. Relevant SAS code is provided.

Biometry↗

PubMed2004

Daily pain that was excruciating at some time in the previous week: prevalence, characteristics, and outcomes in nursing home residents.

OBJECTIVES: To examine the prevalence, correlates, and consequences of nursing home (NH) staff reports of "excruciating" level of pain at some time in the previous week in persons with daily pain reported on the Minimum Data Set (MDS). DESIGN: Cross-sectional study. SETTING: NHs in the United States. PARTICIPANTS: A total 2,138,442 persons who resided in 15,745 nursing homes in the United States. MEASUREMENTS: Pain reported as daily and at its most excruciating in the previous week on the MDS at initial and follow-up assessment. Associations were examined with demographic characteristics, functioning, and measures of disease burden reported on the MDS. RESULTS: NH staff noted that 80,512 (3.7%) of residents had daily pain that was at one or more times excruciating in the previous week. This level of pain was more prevalent in younger residents. Nearly two-thirds (62.1%) of persons with this level of pain were no longer independent in activities of daily living, but 48.8% were rated to have normal cognitive status. In contrast, those without daily pain that was sometimes excruciating were less likely to be cognitively intact (25.7%P<.001) and less likely to have declined in their functioning (30.1%, P=.001). More than one in five with daily pain that was excruciating at times had a cancer diagnosis, and 21.5% experienced weight loss. Of the 24,300 persons with a second assessment, 10,284 (42.3%) still had excruciating pain at some time in the previous week. CONCLUSION: NH residents with daily pain that was sometimes excruciating were younger and seriously ill with functional decline and weight loss. Too often, persons remain in this level of pain.

Activities of Daily Living↗

PubMed2004

Government expenditures at the end of life for short- and long-stay nursing home residents: differences by hospice enrollment status.

OBJECTIVES: To examine end-of-life government expenditures for short- and long-stay Medicare- and Medicaid-eligible (dual-eligible) nursing home (NH) hospice and nonhospice residents. DESIGN: A retrospective cohort study. SETTING: Six hundred fifty-seven Florida NHs. PARTICIPANTS: Dual-eligible NH residents who died in Florida NHs between July and December 1999 (N=5,774). MEASUREMENTS: Nursing home stays of 90 days or less were considered short stays (n=1,739), and those over 90 days were long stays (n=4,035). Three diagnosis groups were studied: cancer without Alzheimer's disease or dementia, Alzheimer's disease or dementia, and other diagnoses. Eligibility and expenditure claims data for 1998 and 1999 were merged with vital statistics and NH resident assessment data to determine diagnoses, location of death, hospice enrollment, eligibility, and expenditures. RESULTS: Twenty percent of short-stay (n=350) and 26% of long-stay (n=958) NH decedents elected hospice; of these, 73% of short-stay and 58% of long-stay NH residents had hospice stays of 30 days or less. Overall, mean government expenditures in the last month of life were significantly less for hospice than nonhospice residents (7,365 dollars; 95% confidence interval (CI)=7,144-7586 dollars vs 8,134 dollars; 95% CI=7,896-8,372 dollars), but 1-month expenditures were only significantly lower for hospice residents with short NH stays, not for those with long NH stays. CONCLUSION: Overall, hospice care in NHs does not appear to increase government expenditures. Because significantly lower expenditures are observed for short-stay NH hospice residents, policy restricting access to Medicare hospice for Medicare skilled nursing facility residents may represent a missed opportunity for savings.

Aged↗

PubMed2003

Inter-rater reliability of nursing home quality indicators in the U.S.

BACKGROUND: In the US, Quality Indicators (QI's) profiling and comparing the performance of hospitals, health plans, nursing homes and physicians are routinely published for consumer review. We report the results of the largest study of inter-rater reliability done on nursing home assessments which generate the data used to derive publicly reported nursing home quality indicators. METHODS: We sampled nursing homes in 6 states, selecting up to 30 residents per facility who were observed and assessed by research nurses on 100 clinical assessment elements contained in the Minimum Data Set (MDS) and compared these with the most recent assessment in the record done by facility nurses. Kappa statistics were generated for all data items and derived for 22 QI's over the entire sample and for each facility. Finally, facilities with many QI's with poor Kappa levels were compared to those with many QI's with excellent Kappa levels on selected characteristics. RESULTS: A total of 462 facilities in 6 states were approached and 219 agreed to participate, yielding a response rate of 47.4%. A total of 5758 residents were included in the inter-rater reliability analyses, around 27.5 per facility. Patients resembled the traditional nursing home resident, only 43.9% were continent of urine and only 25.2% were rated as likely to be discharged within the next 30 days. Results of resident level comparative analyses reveal high inter-rater reliability levels (most items >.75). Using the research nurses as the "gold standard", we compared composite quality indicators based on their ratings with those based on facility nurses. All but two QI's have adequate Kappa levels and 4 QI's have average Kappa values in excess of.80. We found that 16% of participating facilities performed poorly (Kappa <.4) on more than 6 of the 22 QI's while 18% of facilities performed well (Kappa >.75) on 12 or more QI's. No facility characteristics were related to reliability of the data on which Qis are based. CONCLUSION: While a few QI's being used for public reporting have limited reliability as measured in US nursing homes today, the vast majority of QI's are measured reliably across the majority of nursing facilities. Although information about the average facility is reliable, how the public can identify those facilities whose data can be trusted and whose cannot remains a challenge.

Aged↗

PubMed2003

Clinical and organizational factors associated with feeding tube use among nursing home residents with advanced cognitive impairment.

CONTEXT: Empiric data and expert opinion suggest that use of feeding tubes is not beneficial for older persons with advanced dementia. Previous research has shown a 10-fold variation in this practice across the United States. OBJECTIVE: To identify the facility and resident characteristics associated with feeding tube use among US nursing homes residents with severe cognitive impairment. DESIGN, SETTING, AND PARTICIPANTS: Cross-sectional study of all residents with advanced cognitive impairment who had Minimum Data Set assessments within 60 days of April 1, 1999, (N = 186,835) and who resided in Medicare- or Medicaid-certified US nursing homes. Main Outcomes Measures Facility and resident characteristics described in the 1999 On-line Survey Certification of Automated Records and the 1999 Minimum Data Set. Multivariate analysis using generalized estimating equations determined the facility and resident factors independently associated with feeding tube use. RESULTS: Thirty-four percent of residents with advanced cognitive impairment had feeding tubes (N = 63,101). Resident characteristics associated with a greater likelihood of feeding tube use included younger age, nonwhite race, male sex, divorced marital status, lack of advance directives, a recent decline in functional status, and no diagnosis of Alzheimer disease. Controlling for these patient factors, residents living in facilities that were for profit (adjusted odds ratio [OR], 1.09; 95% confidence interval [CI], 1.06-1.12); located in an urban area (OR, 1.14; 95% CI, 1.11-1.16); having more than 100 beds (OR, 1.04; 95% CI, 1.01-1.07); and lacking a special dementia care unit (OR, 1.11; 95% CI, 1.07-1.15) had a higher likelihood of having a feeding tube. Additionally, feeding tube use was more likely among residents living in facilities that had a smaller proportion of residents with do-not-resuscitate orders, had a higher prevalence of nonwhite residents, and lacked a nurse practitioner or physician assistant on staff. CONCLUSIONS: More than one third of severely cognitively impaired residents in US nursing homes have feeding tubes. Feeding tube use is independently associated with both the residents' clinical characteristics and the nursing homes' fiscal, organizational, and demographic features.

Advance Directives↗

PubMed2003

Scaled marginal models for multiple continuous outcomes.

In studies that involve multivariate outcomes it is often of interest to test for a common exposure effect. For example, our research is motivated by a study of neurocognitive performance in a cohort of HIV-infected women. The goal is to determine whether highly active antiretroviral therapy affects different aspects of neurocognitive functioning to the same degree and if so, to test for the treatment effect using a more powerful one-degree-of-freedom global test. Since multivariate continuous outcomes are likely to be measured on different scales, such a common exposure effect has not been well defined. We propose the use of a scaled marginal model for testing and estimating this global effect when the outcomes are all continuous. A key feature of the model is that the effect of exposure is represented by a common effect size and hence has a well-understood, practical interpretation. Estimating equations are proposed to estimate the regression coefficients and the outcome-specific scale parameters, where the correct specification of the within-subject correlation is not required. These estimating equations can be solved by repeatedly calling standard generalized estimating equations software such as SAS PROC GENMOD. To test whether the assumption of a common exposure effect is reasonable, we propose the use of an estimating-equation-based score-type test. We study the asymptotic efficiency loss of the proposed estimators, and show that they generally have high efficiency compared to the maximum likelihood estimators. The proposed method is applied to the HIV data.

Antiretroviral Therapy, Highly Active↗

PubMed2003

Modeling longitudinal data with nonignorable dropouts using a latent dropout class model.

In longitudinal studies with dropout, pattern-mixture models form an attractive modeling framework to account for nonignorable missing data. However, pattern-mixture models assume that the components of the mixture distribution are entirely determined by the dropout times. That is, two subjects with the same dropout time have the same distribution for their response with probability one. As that is unlikely to be the case, this assumption made lead to classification error. In addition, if there are certain dropout patterns with very few subjects, which often occurs when the number of observation times is relatively large, pattern-specific parameters may be weakly identified or require identifying restrictions. We propose an alternative approach, which is a latent-class model. The dropout time is assumed to be related to the unobserved (latent) class membership, where the number of classes is less than the number of observed patterns; a regression model for the response is specified conditional on the latent variable. This is a type of shared-parameter model, where the shared "parameter" is discrete. Parameter estimates are obtained using the method of maximum likelihood. Averaging the estimates of the conditional parameters over the distribution of the latent variable yields estimates of the marginal regression parameters. The methodology is illustrated using longitudinal data on depression from a study of HIV in women.

Analysis of Variance↗

PubMed2003

Classification and regression tree analysis in public health: methodological review and comparison with logistic regression.

BACKGROUND: Audience segmentation strategies are of increasing interest to public health professionals who wish to identify easily defined, mutually exclusive population subgroups whose members share similar characteristics that help determine participation in a health-related behavior as a basis for targeted interventions. Classification and regression tree (C&RT) analysis is a nonparametric decision tree methodology that has the ability to efficiently segment populations into meaningful subgroups. However, it is not commonly used in public health. PURPOSE: This study provides a methodological overview of C&RT analysis for persons unfamiliar with the procedure. METHODS AND RESULTS: An example of a C&RT analysis is provided and interpretation of results is discussed. Results are validated with those obtained from a logistic regression model that was created to replicate the C&RT findings. Results obtained from the example C&RT analysis are also compared to those obtained from a common approach to logistic regression, the stepwise selection procedure. Issues to consider when deciding whether to use C&RT are discussed, and situations in which C&RT may and may not be beneficial are described. CONCLUSIONS: C&RT is a promising research tool for the identification of at-risk populations in public health research and outreach.

Decision Trees↗