Bias adjustment in analysing longitudinal data with informative missingness

Total observed caries experience: assessing the effectiveness of community-based caries prevention

OBJECTIVES

To compare common measures for dental caries in longitudinal caries research in support of a new measure, the Total Observed Caries Experience (TOCE).

METHODS

TOCE was calculated as the total number of decayed or filled teeth ever observed. TOCE was compared to dmft, DMFT, and the prevalence of untreated decay using data from a 4-year, cluster randomized community-based caries prevention program. Outcomes were analyzed using generalized estimating equations and assessed change over time and effects across treatment groups.

RESULTS

For TOCE and DMFT, the risk of caries over time increased. In contrast, risks decreased for dmft and the prevalence of untreated decay, driven largely by decayed primary teeth being exfoliated. Effects across treatment groups showed increased risk for some outcomes and attenuated risks in others; however, treatment effects for TOCE were more in line with expected results when compared with controls.

CONCLUSIONS

TOCE accounts for caries on primary and permanent dentition and is not subject to bias due to exfoliation. It can be used as a robust measure to evaluate community-based caries prevention programs.

Design and baseline data from a PCORI-funded randomized controlled trial of family-centered tailoring of diabetes self-management resources

This article describes the methodology, recruitment, participant characteristics, and sustained, intensive stakeholder engagement for Project ACE (Achieving control, Connecting resources, Empowering families). Project ACE is a randomized controlled trial of children and youth ages 8-16 with type 1 diabetes evaluating the impact of tailored self-management resources on hemoglobin A1c (A1c) and quality of life (QOL). Despite strong evidence that controlling A1c reduces long-term complications, <25% of US youth with type 1 diabetes meet A1c targets. Many interventions are efficacious in improving A1c and QOL for these youth, whose families often struggle with the substantial demands of the treatment regimen. However, most such interventions are ineffective in the real world due to lack of uptake by families and limited healthcare system resources. Project ACE is a multi-site trial designed to improve diabetes outcomes by tailoring existing, evidence-based interventions to meet families' needs and preferences. We hypothesize that this family-centered approach will result in better A1c and QOL than usual care. Project ACE has recruited and randomized 214 eligible 8-16year old youth and their parents. The 9-month intervention consisted of 4 group sessions tailored to families' self-management barriers as identified by a validated instrument. Outcomes including A1c and QOL for parents and youth will be assessed for 1year after the intervention. Stakeholder engagement was used to enhance this trial's recruitment, retention and integration into routine clinical care. Findings will inform implementation and dissemination of family-centered approaches to address self-management barriers.

TRIAL REGISTRATION NUMBER

NCT02024750 Trial Registrar: Clinicaltrials.gov, https://clinicaltrials.gov/ct2/show/NCT02024750.

Ten-year effects of the advanced cognitive training for independent and vital elderly cognitive training trial on cognition and everyday functioning in older adults

OBJECTIVES

To determine the effects of cognitive training on cognitive abilities and everyday function over 10 years.

DESIGN

Ten-year follow-up of a randomized, controlled single-blind trial (Advanced Cognitive Training for Independent and Vital Elderly (ACTIVE)) with three intervention groups and a no-contact control group.

SETTING

Six U.S. cities.

PARTICIPANTS

A volunteer sample of 2,832 persons (mean baseline age 73.6; 26% African American) living independently.

INTERVENTION

Ten training sessions for memory, reasoning, or speed of processing; four sessions of booster training 11 and 35 months after initial training.

MEASUREMENTS

Objectively measured cognitive abilities and self-reported and performance-based measures of everyday function.

RESULTS

Participants in each intervention group reported less difficulty with instrumental activities of daily living (IADLs) (memory: effect size = 0.48, 99% confidence interval (CI) = 0.12-0.84; reasoning: effect size = 0.38, 99% CI = 0.02-0.74; speed of processing: effect size = 0.36, 99% CI = 0.01-0.72). At a mean age of 82, approximately 60% of trained participants, versus 50% of controls (P < .05), were at or above their baseline level of self-reported IADL function at 10 years. The reasoning and speed-of-processing interventions maintained their effects on their targeted cognitive abilities at 10 years (reasoning: effect size = 0.23, 99% CI = 0.09-0.38; speed of processing: effect size = 0.66, 99% CI = 0.43-0.88). Memory training effects were no longer maintained for memory performance. Booster training produced additional and durable improvement for the reasoning intervention for reasoning performance (effect size = 0.21, 99% CI = 0.01-0.41) and the speed-of-processing intervention for speed-of-processing performance (effect size = 0.62, 99% CI = 0.31-0.93).

CONCLUSION

Each Advanced Cognitive Training for Independent and Vital Elderly cognitive intervention resulted in less decline in self-reported IADL compared with the control group. Reasoning and speed, but not memory, training resulted in improved targeted cognitive abilities for 10 years.

Treatment of death in the analysis of longitudinal studies of gerontological outcomes

BACKGROUND

Longitudinal studies in gerontology are characterized by termination of measurement from death. Death is related to many important gerontological outcomes, such as functional disability, and may, over time, change the composition of an older study population. For these reasons, treating death as noninformative censoring of a longitudinal outcome may result in biased estimates of regression coefficients related to that outcome.

METHODS

In a longitudinal study of community-living older persons, we analytically and graphically illustrate the dependence between death and functional disability. Relative to survivors, decedents display a rapid decline of functional ability in the months preceding death. Death's strong relationship with functional disability demonstrates that death is not independent of this outcome and, hence, leads to informative censoring. We also demonstrate the "healthy survivor effect" that results from death's selection effect, with respect to functional disability, on the longitudinal makeup of an older study population.

RESULTS

We briefly survey commonly used approaches for longitudinal modeling of gerontological outcomes, with special emphasis on their treatment of death. Most common methods treat death as noninformative censoring. However, joint modeling methods are described that take into account any dependency between death and a longitudinal outcome.

CONCLUSIONS

In longitudinal studies of older persons, death is often related to gerontological outcomes and, therefore, cannot be safely assumed to represent noninformative censoring. Such analyzes must account for the dependence between outcomes and death as well as the changing nature of the cohort.

Early blood gas abnormalities and the preterm brain

The authors explored associations between blood gas abnormalities in more than 1,000 preterm infants during the first postnatal days and indicators of neonatal brain damage. During 2002-2004, women delivering infants before 28 weeks' gestation at one of 14 participating institutions in 5 US states were asked to enroll in the study. The authors compared infants with blood gas values in the highest or lowest quintile for gestational age and postnatal day (extreme value) on at least 1 of the first 3 postnatal days with the remainder of the subjects, with separate analyses for blood gas abnormalities on multiple days and for partial pressure of oxygen in the alveolar gas of <35. Outcomes analyzed were ventriculomegaly and an echolucent lesion on an ultrasound scan in the neonatal intensive care unit, and cerebral palsy, microcephaly, and a low score on a Bayley Scale of Infant Development at 24 months. Every blood gas derangement (hypoxemia, hyperoxemia, hypocapnia, hypercapnia, and acidosis) was associated with multiple indicators of brain damage. However, for some, the associations were seen with only 1 day of exposure; others were evident with 2 or more days' exposure. Findings suggest that individual blood gas derangements do not increase brain damage risk. Rather, the multiple derangements associated with indicators of brain damage might be indicators of immaturity/vulnerability and illness severity.

Estimation in longitudinal or panel data models with random-effect-based missing responses

In studies with longitudinal or panel data, missing responses often depend on values of responses through a subject-level unobserved random effect. Besides the likelihood approach based on parametric models, there exists a semiparametric method, the approximate conditional model (ACM) approach, which relies on the availability of a summary statistic and a linear or polynomial approximation to some random effects. However, two important issues must be addressed in applying ACM. The first is how to find a summary statistic and the second is how to estimate the parameters in the original model using estimates of parameters in ACM. Our study is to address these two issues. For the first issue, we derive summary statistics under various situations. For the second issue, we propose to use a grouping method, instead of linear or polynomial approximation to random effects. Because the grouping method is a moment-based approach, the conditions we assumed in deriving summary statistics are weaker than the existing ones in the literature. When the derived summary statistic is continuous, we propose to use a classification tree method to obtain an approximate summary statistic for grouping. Some simulation results are presented to study the finite sample performance of the proposed method. An application is illustrated using data from the study of Modification of Diet in Renal Disease.

Race and ethnicity as factors in mental health service use among veterans with PTSD

Many veterans with posttraumatic stress disorder (PTSD) either do not seek treatment or participate in treatment only episodically. The authors conducted a retrospective cohort study using Department of Veterans Affairs (VA) administrative data to determine rates of mental health care use and to examine whether the odds of service use varied by race or ethnicity. They examined all veterans with a new diagnosis of PTSD during a one-year period. Analyses used logistic or negative binomial regression with generalized estimating equations to measure associations of race and ethnicity with mental health service use after controlling for demographic, disability, and access factors. After adjustment, veteran race, but not Hispanic ethnicity, was associated with decreases in some pharmacotherapy measures and increases in some counseling measures.

Lower-than-expected prevalence and severity of retinopathy in an incident cohort followed during the first 4-14 years of type 1 diabetes: the Wisconsin Diabetes Registry Study

The authors examined the development of diabetic retinopathy in a population-based cohort of persons with incident type 1 diabetes to investigate the possibility of lowered retinopathy prevalence and severity compared with previous US studies. A total of 474 diabetic persons from Wisconsin were followed from diagnosis through 4-14 years' duration during 1990-2002. Retinopathy was determined by fundus photography at 4, 7, 9, and 14 years' duration. Risk of developing retinopathy was modeled on demographic and diabetes-care characteristics by means of a generalized linear model using the complementary log-log link for interval-censored data. Prevalence of retinopathy increased with duration of diabetes, from 6% at 4 years to 73% at 14 years, and was highest among adults (> or =20 years of age). Risk of developing retinopathy increased with increasing duration, worse glycemic control, and age up to 20 years. Indicators of diabetes care were related to retinopathy through their effect on glycemic control. Improvements in diabetes care leading to better glycemic control may have contributed to the much lower prevalence and less severe retinopathy observed than expected on the basis of a previous report from the same region of Wisconsin. The observed decreased prevalence has important implications for persons with type 1 diabetes, since retinopathy is a serious microvascular complication.

Analysis of missing data in palliative care studies

This report discusses the general problem of the analysis of data that could include missing values. In the palliative care setting, the data may not be missing at random, but instead be related to the outcome of interest, and therefore the use of standard statistical procedures may be problematic. This study summarizes differing results that were found when using three simple methods for estimating missing data in an example data set testing for differences in the use of morphine or methadone for relief of pain. Differences in the conclusions are discussed and recommendations are made to improve the reporting of studies with missing data.