Search PubMed⌕ Search

SEARCH · Search PubMed

Results for “longitudinal”

Search indexed PubMed citations on genomics, clinical trials, systematic reviews and public health. Explore titles, authors and supplied subject terms, then open the PubMed record.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 415 records · Page 23Linked to original sources
PubMed1986

Stability and change in adult intelligence: 1. Analysis of longitudinal covariance structures.

We address two questions of central interest in adult intellectual development: the equivalence of psychometric tests' measurement properties at different ages, and the stability of individual differences in intelligence over time. We performed a series of longitudinal factor analyses using the LISREL program to model longitudinal data from Schaie's Seattle Longitudinal Study. The results indicate complete invariance in the loadings of five subtests of Thurstone's Primary Mental Abilities battery on a general intelligence factor. Individual differences in general intelligence were highly stable over 14-year epochs, with standardized factor correlations averaging about .9 between adjacent 7-year testing intervals. These results indicate that most individuals in this relatively select longitudinal sample maintained their relative ordering in intelligence.

Adult↗
PubMed1997

The association between dietary patterns and cardio vascular disease risk indicators in healthy youngsters: results covering fifteen years of longitudinal development.

OBJECTIVE: To examine longitudinal relationships between nutrition and risk indicators for cardio vascular diseases (CVD) during adolescence and young adulthood. DESIGN: A longitudinal study over fifteen years. SUBJECTS: 98 females and 84 males, from 13 to 27 years. METHODS: By means of six interviews dietary patterns were determined. Blood samples were analyzed for serum concentration of total cholesterol (TC), and high-density-lipoprotein (HDL), blood pressure, body fat and maximal aerobic power (VO2max) were determined. The longitudinal relations were analyzed with generalized estimation equations (GEE), a statistical technique in which relations at different time-points are tested simultaneously. RESULTS: Compared to Dutch recommendations six out of seven macro nutrients appear to be unfavorable with respect to CVD. Borderline or high CVD risk values are apparent at 27 y in more than 25% of the subjects with respect to percentage body fat and serum total cholesterol in both sexes. In males 40% or more show borderline hypertension. The 'univariate' longitudinal analyses showed significantly positive relations: (1) between the intake of animal protein, saturated fat (SFA), cholesterol (Cho1) and TC, and HDL; (2) between total energy intake (EN) and systolic blood pressure, and VO2max. Significantly negative associations were found: (1) between EN, poly-unsaturated fat (PUFA) and TC concentrations; (2) between EN and sum of four skinfolds (SSF). CONCLUSIONS: With increasing age, over a period of 15 y in both sexes the SFA and Cho1 intake relate significantly to the development of a negative CVD risk profile. The intake of PUFA relates positive to a CVD risk profile. The significantly negative relation between EN intake and body fat (SSF) is partly explained by the relation between EN and VO2max.

Adolescent↗
PubMed2004

Predictors of children's body mass index: a longitudinal study of diet and growth in children aged 2-8 y.

OBJECTIVE: To identify longitudinal variables related to children's body mass index (BMI) (kg/m(2)) at age 8 y. DESIGN: A longitudinal design, with nine interviews per child from ages 2 to 8 y. SUBJECTS: In all, 70 white children (37 males, 33 females) who were continuous participants since infancy in the longitudinal study. Families were primarily middle and upper socioeconomic status. MEASUREMENTS: At each interview, children's height and weight were measured, and mothers provided 3 days of the child's intake data (a 24-h recall and 2 days of food records). ANALYSES: Analyses used were means+/-s.d., correlations, repeated measures analysis of variance, and forward stepwise regression. BMI at each interview was calculated and age of adiposity rebound was determined. RESULTS: Children's BMI at 8 y was negatively predicted by age of adiposity rebound and positively predicted by their BMI at 2 y. Additionally, each model included one longitudinal dietary variable; mean protein and fat intakes recorded between 2 and 8 y were positive predictors of BMI at 8 y; mean carbohydrate intake over the same time period was negatively related to BMI at 8 y. R(2) values indicated that these three-variable models predicted 41-43% of the variability in BMI among children. BMI of 23% of the children exceeded the 85th CDC percentile. CONCLUSIONS: The results of this study show that factors in early life are associated with children's BMI at age 8 y.

Body Height↗
PubMed2001

Ageing with spinal cord injury: cross-sectional and longitudinal effects.

STUDY DESIGN: Longitudinal and cross-sectional. OBJECTIVE: To determine whether, for studies of ageing with a spinal cord injury, the cross-sectional differences in outcomes across both age and years post injury (YPI) differ from the longitudinal change. SETTING: Two SCI centres in England: the National Spinal Injuries Centre in Aylesbury, and the Regional Spinal Injuries Centre in Southport. METHODS: A total of 315 people who sustained spinal cord injuries prior to 1971 underwent comprehensive health and psychosocial status interviews at one or more of the study assessments (1990, 1993, 1996, and 1999). A range of continuous and dichotomous outcomes was analyzed to detect both cross-sectional differences by age and average individual changes over multiple measurements. RESULTS: Frequently, outcomes changed longitudinally without showing any cross-sectional differences. Cross-sectional age was more commonly associated with the worsening of a condition while cross-sectional YPI was commonly associated with improvement. After controlling for cross-sectional effects, psychological measures generally showed minor deterioration, measures of community integration both improved and deteriorated, upper extremity pain increased, lower enxtremity pain decreased, and participants tended to quit smoking. CONCLUSION: Using longitudinal findings that control for cross-sectional differences produces a more complete description of ageing with a spinal cord injury.

Activities of Daily Living↗
PubMed2001

Changes of leptin and metabolic hormones in preterm infants: a longitudinal study in early postnatal life.

OBJECTIVES: Very little is known concerning the physiological role of leptin and growth in the early postnatal period and the association of leptin with other metabolic hormones in preterm infants. This study aims to investigate these relationships, and to explore the longitudinal and dynamic profile of leptin and metabolic hormones including insulin, ACTH, cortisol and FT4 in this category of patient. We also postulate that a rapid increase in body weight and body mass index in the first few weeks of life may be associated with a corresponding increase in serum leptin if the 'adipoinsular axis' is active at this stage. DESIGN: A longitudinal study in a cohort of preterm infants < 34 weeks gestation for the first five weeks of postnatal life. PATIENTS AND METHODS: Sixty-one preterm newborns < 34 weeks gestation were prospectively enrolled. Blood samples were collected in the morning when the newborns were 24 h of age (day 1), and on days 5, 14 and 35 of life. Serum leptin, insulin, cortisol, FT4, glucose and plasma ACTH concentrations were analysed using standard biochemical methods. Spearman's correlation coefficient was used to assess the inter-relationship of different metabolic hormones on the first day of life, and the relationship between metabolic hormones on day 1 and anthropometric or clinical parameters. The mixed-effects models were further used for analysing the multiple longitudinal measurements, and also for comparing the hormone concentrations between day 1 (baseline) and their corresponding levels on days 5, 14 and 35. RESULTS: Serum leptin on day 1 (baseline) was significantly associated with serum insulin (r = 0.30, P < 0.05) and insulin:glucose ratio (r = 0.29, P < 0.05) in infants < 34 weeks gestation. Female preterm infants had significantly higher serum leptin (P < 0.05) and insulin (P < 0.05) levels than male infants. In addition, the duration between the last dose of antenatal dexamethasone and delivery significantly influenced the serum concentrations of leptin (r = - 0.27, P < 0.05), cortisol (r = 0.52, P < 0.001), plasma ACTH (r = 0.28, P < 0.05) and insulin:glucose ratio (r = - 0.27, P < 0.05) on the first day of life. Despite significant increase in body weight (P < 0.00001) and body mass index (P < 0.00001) by day 35 when compared to birth weight, no corresponding significant changes were observed for serum leptin, insulin and FT4. However, there was an increasing though statistically nonsignificant trend in serum leptin after day 14 of life. CONCLUSION: This study characterized the longitudinal profile of leptin and metabolic hormones in preterm infants < 34 weeks gestation in the first 5 weeks of life. Serum leptin was significantly associated with serum insulin and insulin:glucose ratio supporting the hypothesis that an 'adipoinsular axis' exists and is likely to be functional before 34 weeks of gestation. Although a significant increase in body weight was demonstrated by day 35, no significant corresponding changes occurred with regard to serum leptin. We postulate that the limited quantity of adipose tissue at this stage of development might have contributed to this observation. Moreover, our results also showed that the duration between the last dose of antenatal corticosteroid and delivery could influence the postnatal concentrations of adipoinsular and pituitary-adrenal hormones. Thus, it is possible that antenatal dexamethasone might affect fetal growth and development via these neuroendocrine pathways in early intrauterine life.

Adrenocorticotropic Hormone↗
PubMed1999

Total body water data for white adults 18 to 64 years of age: the Fels Longitudinal Study.

BACKGROUND: Total body water (TBW) volume is reported to decrease with age, but much of the published data are 20 to almost 50 years old and are cross-sectional. Proper interpretation of clinical levels of TBW and trends with age necessitates the availability of current longitudinal data from healthy individuals. METHODS: Mixed longitudinal data for TBW of 274 white men and 292 white women (18 to 64 years of age) in the Fels Longitudinal Study were collected on a regular schedule over a recent eight-year period. The concentration of deuterium was measured by deuterium nuclear magnetic resonance spectroscopy. Body composition estimates were made with dual-energy x-ray absorptiometry, and random effect models were used to determine the patterns of change over time with and without covariates. RESULTS: The mean TBW data for the Fels men are either similar to or approximately 2 to as much as 6 liters greater than that reported by most other investigators 20 to 50 years ago. For Fels women, the mean TBW ranges from approximately 2 to as much as 5 liters less than that reported previously. These comparisons with much earlier studies reflect cohort effects and the secular changes in overall body size that have occurred during the past 60 to 70 years. These findings are reinforced by the fact that some early data sets included individuals born almost 140 years ago. After adjusting for the covariate effects of total body fat (TBF) and fat-free mass (FFM) with age, there were no significant age or age-squared effects on TBW in the men. In the women, after adjusting for the covariate associations of TBF and FFM with age, there was a small, but significant, negative linear association of TBW with age. In the men and women, the mean ratio of TBW to weight declined with age as a function of an increase in body fatness and more so for the men than the women. CONCLUSION: The findings from these mixed longitudinal data indicate that TBW volume, on average, maintains a reasonable degree of stability in men and women through a large portion of adulthood. These TBW data are recommended as current reference data for healthy adults.

Adipose Tissue↗
PubMed1999

Relating speech and swallow function to dropout in a longitudinal study of head and neck cancer.

The relation between functional outcome and dropout from a 12-month follow-up period was examined in a longitudinal study whose objective was to define and quantify the functional effects of oral surgical resection and reconstruction on speech and swallowing in patients with head and neck cancer. In a group of 150 patients recruited to a surgical study in the Cancer Control Science Program in Head and Neck Cancer Rehabilitation, dropout from all causes and dropout from specific causes (medical, patient, and administrative specific) were assessed in relation to longitudinal speech and swallow function. In univariate analysis, better speech articulation was associated with decreased risk of dropout from all causes and from medical-specific causes. Better swallow performance was associated with decreased risk of medical-specific dropout. Multivariate analysis revealed the following: (1) only articulation function was associated with dropout from all causes; (2) the association of speech articulation function with medical dropout was diminished after adjusting for advanced age and surgical resection variables; (3) the association of speech articulation function became significant for patient-specific dropout after adjusting for advanced age and surgical resection variables and indicated that better function decreased the risk of this type of dropout; and (4) swallowing function was not related to dropout. Patients treated for oral or oropharyngeal cancer who have poorer speech outcomes are more likely to drop out from a longitudinal study. Basing study results on only patients who complete a longitudinal study will understate the level of dysfunction experienced.

Adult↗
PubMedDate not supplied

[Professional status and development of morbidity of health insurance patients in the period 1990 to 2003. A longitudinal Analysis of Routine Data from the Gmünder Ersatzkasse].

In Germany there are only insufficient data both with regard to general morbidity events as well as with regard to the effects of specific professions on the morbidity process. For this reason the data of the Gmuender Ersatzkasse (GEK) have been analysed with a longitudinal design for the period 1990 to 2003. Our sample includes all employed members of the GEK who had been insured as of January 1, 1990 and were between 30 and 59 years old at this baseline. The total sample comprises 129,173 men and 13,567 women. The statistical analysis was performed with the statistical package "Transitional Data Analysis" (TDA) which was developed especially for longitudinal data. The analysis includes both cumulated morbidity rates as well as transition rates (Cox regression). The analysis shows that the cumulated morbidity rates of the seven diseases are higher for women than for men in the age category 30-39 years. In the age categories 40-49 and 50-59 years the cumulated morbidity rates are higher for men than for women. With regard to the occupations, the highest morbidity rates are found for manual occupations and services without special skills. The lowest rates are found for professions with high skills, engineers and managers. The results of the longitudinal analysis show distinctive social gradients. For occupations with lower skills the morbidity rates are about 100 percent higher than those of occupations with higher skills. Longitudinal analyses on the basis of health insurance fund data can make an important contribution to the monitoring of health and morbidity and should therefore be conducted also by other health insurance funds.

Adult↗
PubMed1999

Neuropsychometric tests in cross sectional and longitudinal studies - a regression analysis of ADAS - cog, SKT and MMSE.

INTRODUCTION: In clinical and drug studies, different neuropsychometric tests are used. So far, no empirical data have been published to compare studies using different tests. The purpose of this study was to calculate a regression formula allowing a comparison of cross-sectional and longitudinal data from three neuropsychometric tests that are frequently used in drug studies (Alzheimer's Disease Assessment Scale, ADAS-cog; Syndrom Kurz Test, SKT; Mini Mental State Examination, MMSE). METHOD: 177 patients with dementia according to ICD10 criteria were studied for the cross sectional and 61 for the longitudinal analysis. Correlations and linear regressions were calculated between tests. Significance was proven with ANOVA and t-tests using the SPSS statistical package. RESULTS: Significant Spearman correlations and slopes in the regression occurred in the cross sectional analysis (ADAS-cog-SKT r(s) = 0.77, slope = 0.45, SKT-ADAS-cog slope = 1.3, r2 = 0.59; ADAS-cog-MMSE r2 = 0.76, slope = -0.42, MMSE-ADAS-cog slope = -1.5, r2 = 0.64; MMSE-SKT r(s) = -0.79, slope = -0.87, SKT-MMSE slope = -0.71, r2 = 0.62; p<0.001 after Bonferroni correction; N = 177) and in the longitudinal analysis (SKT-ADAS-cog, r(s) = 0.48, slope = 0.69, ADAS-cog-SKT slope = 0.69, p<0.001, r2 = 0.32, MMSE-SKT, r(s) = 0.44, slope = -0.41, SKT-MMSE, slope = -0.55, p<0.001, r2 = 0.21). CONCLUSIONS: The results allow calculation of ADAS-scores when SKT scores are given, and vice versa. In longitudinal studies or in the course of the disease, scores assessed with the ADAS-cog and the SKT may now be statistically compared. In all comparisons, bottom and ceiling effects of the tests have to be taken into account.

Aged↗
PubMed2002

Anterior open bite in the deciduous dentition: longitudinal follow-up and craniofacial growth considerations.

The aim of this longitudinal cephalometric study was to evaluate craniofacial growth changes in subjects with an anterior open bite in the deciduous dentition. From longitudinal records of untreated subjects, an open bite group (n = 14) was selected at the age of 5 years based on the presence of a negative overbite and compared with a control group (n = 14) with a regular overbite at this age. Cephalometric measurements were analyzed at ages 5, 9, and 12 years. Although only 1 subject in the deciduous dentition open bite group had an open bite at 12 years of age, the overbite remained lower during the longitudinal follow-up. Early cephalometric characteristics of the open bite group included a reduced overbite depth indicator and a lower ANB angle. At ages 9 and 12 years, the open bite sample was also characterized by shorter ramus height. An underlying skeletal pattern seems to be present in the deciduous dentition open bite sample that persists during the longitudinal follow-up. The overbite depth indicator might help to identify patients with anterior open bite tendencies.

Case-Control Studies↗
PubMed2003

Dental age and dental health determined longitudinally from patient records in three towns in Finland.

The aims of this study were to test the suitability of normal dental records in determining the eruption times of teeth and to compare the retrospective longitudinal DMF values with the statistical cross-sectional means. Patient records for two cohorts (1970 and 1980) were collected from three Finnish towns. The study population comprised 1910 children. The eruption times of the permanent teeth, DMFT and DMFS curves (excluding F due to fractures and M due to orthodontic reasons) were investigated for the year of examination in different study locations. The cross-sectional DMF values were compared to longitudinally calculated data. Tooth eruption was earlier in girls than in boys. However, girls did not have significantly higher DMF values than boys. In all towns, dental health was significantly better in the 1980 cohort than in the 1970 cohort. Comparison of the cross-sectional DMF values in the municipal health center statistics showed that the mean statistics gave considerably higher values than the measured values did after 15 years of age. The inconsistency between measured longitudinal DMF values and the cross-sectional statistical DMF values indicates the importance of creating computer programs for analyzing data longitudinally from normal dental records.

Adolescent↗
PubMedDate not supplied

Mixed longitudinal studies: their efficiency for the estimation of trends over time.

Mixed longitudinal studies may be employed for the estimation of trends over time because they give more comparable data than a series of cross-sectional studies, or because of multiple objectives. The estimation of trends from such studies is an example of their use for a cross-sectional rather than longitudinal objective, for which the appropriate analysis is different from that usually described for the analysis of longitudinal data. The efficiency of mixed longitudinal studies for measuring trends in measurements, such as height and weight relative to a comparable series of cross-sectional studies, is calculated for various designs, percentage follow-up and correlation between measurements on the same subjects. The relative efficiency was over 0.8 for most designs and sometimes greater than 1.0. It generally decreased as the number of surveys increased, but increased as the difference between age groups studied increased.

Age Factors↗
PubMedDate not supplied

Age-related slowing on the Digit Symbol task: longitudinal and cross-sectional analyses.

It is generally known that slowing on tasks like the Digit Symbol Substitution task occurs with age. However, simultaneous cross-sectional and longitudinal data are rare. The results from the cross-sectional analysis of Digit Symbol performance with age are compared to results from two occasion longitudinal analyses on the same task. The longitudinal data were obtained from 79 participants who were retested after 7 years and were between the ages of 48 and 73 years old at Time 1. Results indicated that the pattern of actual change observed in the longitudinal analyses replicated the simulated change indicated in the cross-sectional analysis. Antecedents were examined using the original cross-sectional sample of 150 participants. Significant relationships with Digit Symbol performance were found for age, education, and gender, but not for ratings of physical health.

Aged↗
PubMed2005

Discordance between cross-sectional and longitudinal studies for the effect of dust on COPD: why?

Regression analyses for the effect of an environmental agent on lung function often give discordant results when derived from cross-sectional compared with longitudinal studies. To evaluate why this occurs, a normal population was created by computer, and modeled to simulate functional change during life. Thus, factors known to influence lung function measurement (including those that may cause COPD) were manipulated experimentally so that their contributions to any discordance could be assessed. Regression analyses showed that significant discordance could be induced if the oldest birth cohort failed to reach the same maximal level of function as the youngest (a "cohort effect"). This distorted the cross-sectional (but not longitudinal) estimate for the dominating effect of age and additionally influenced cross-sectional estimates for the effects of partially collinear variables such as cumulative exposure to hazardous environmental dust. Discordance also occurred if regression coefficients became imprecise through random measurement/reporting error, between-subject variability, and differing susceptibility, but then the differences (sometimes marked) between cross-sectional and longitudinal estimates were not significant. We conclude that modeling a population with known characteristics can provide a useful means of demonstrating that cross-sectional versus longitudinal discordance may be fundamental and unavoidable (though explicable), or merely a consequence of imprecision.

Aging↗
PubMed2002

Individual longitudinal patterns in biochemical and hematological markers for the early prediction of pre-eclampsia.

OBJECTIVE: To analyze the individual longitudinal patterns of maternal biochemical and hematological tests performed throughout gestation in order to predict at the 20th week of pregnancy the later development of pre-eclampsia. STUDY DESIGN: A longitudinal study was conducted on 187 white normotensive pregnant women all with a history of pre-eclampsia. Blood samples were performed at the 8th week of gestation and then every 4 weeks until the 36th week. The longitudinal patterns of urea, creatinine, uric acid, total proteins, hematocrit, red blood cells, hemoglobin, mean red cell volume, ferritin and iron were derived. By means of regression analysis, for each woman and each significant marker, a 'theoretical physiological pattern', from the 8th to the 20th week, was constructed. By comparing the observed values of each marker for each woman with her 'theoretical physiological pattern', variables indicating the match or mismatch to it were derived. Such variables were used, together with other maternal characteristics, in a logit regression for the probability of developing pre-eclampsia later in pregnancy. RESULTS: In 140 cases, pregnancies followed a physiological course, while 47 women developed pre-eclampsia during the third trimester. In the physiological gestations, the weekly mean values of creatinine, hematocrit, total proteins, uric acid and urea showed patterns that were significantly different from those of the pathological group. The logit model was able to classify correctly 96% of the physiological and 87% of the pathological pregnancies, with a negative predictive value of 96% and a positive predictive value of 89% (area under the receiver operator characteristics (ROC) curve 0.98). The ability of the model to predict later complications at the 20th week was confirmed by a validation procedure. CONCLUSION: The simultaneous use of individual longitudinal patterns of parameters, achieved non-invasively as part of the standard methods of antenatal care that provide a global evaluation of plasma volume expansion, showed a high ability to predict, early in pregnancy, the later development of pre-eclampsia.

Adult↗
PubMed2000

Dropouts in longitudinal studies: definitions and models.

The widely used distinction of Little and Rubin (1) about types of randomness for missing data presents difficulties in its application to dropouts in longitudinal repeated measurement studies. In its place, a new typology of randomness for dropouts is proposed that relies on using a survival model for the dropout process. In terms of a stochastic process, dropping out is a change of state. Then, the longitudinal measures and dropout processes can be modeled simultaneously, each conditional on the complete previous history of both repeated measures and states. In this context, Poisson regression is used to fit various proportional hazards models, some of which are new, to the dropout process using the longitudinal measurements responses as time-varying covariates. As examples of longitudinal measurement studies displaying nonrandom dropout processes, a dental study of testosterone production in rats and clinical trials for treatment of gallstones and of depression are analyzed.

Algorithms↗
PubMed2006

A comparative longitudinal quality of life study using the Spitzer quality of life index in a randomized multicenter phase III trial (FFCD 9102): chemoradiation followed by surgery compared with chemoradiation alone in locally advanced squamous resectable thoracic esophageal cancer.

BACKGROUND: The aim of the study was to compare the longitudinal quality of life (QoL) between chemoradiation with or without surgery in patients with locally advanced squamous resectable esophageal cancer included in a randomized multicenter phase III trial (FFCD 9102). MATERIALS AND METHODS: All patients with locally advanced resectable (T3-4 N0-1 M0) epidermoid or glandular esophageal cancer (n = 451) received induction chemoradiation. Responders (n = 259) were randomized between surgery (arm A) and continuation of chemoradiation (arm B). The Spitzer QoL Index was scored (0-10) at inclusion and at each follow-up, every 3 months during 2 years. QoL at baseline and longitudinal changes were respectively compared with univariate ANOVA and mixed-model analysis of variance for repeated measurements. The time interval between the follow-up was assessed and the same analyses were performed among survivors with 2 years of follow-up. RESULTS: The squamous histology was predominant in both arms. The mean QoL score decreased between baseline and the first follow-up and between the first and the second follow-ups. QoL scores at the first follow-up were comparatively worse in arm A than in arm B (7.52 versus 8.45, P < 0.01), whereas the longitudinal QoL study showed no difference between treatments (adjusted P = 0.26). Furthermore, the longitudinal QoL was not different (adjusted P = 0.23) among survivors with 2 years of follow-up. CONCLUSIONS: Among patients responding to induction chemoradiation, surgery and continuation of chemoradiation had the same impact on QoL in patients with locally advanced, resectable esophageal cancer although a significantly greater decrease in the Spitzer Index was observed in the postoperative period.

Adenocarcinoma↗
PubMed2005

Directly parameterized regression conditioning on being alive: analysis of longitudinal data truncated by deaths.

For observational longitudinal studies of geriatric populations, outcomes such as disability or cognitive functioning are often censored by death. Statistical analysis of such data may explicitly condition on either vital status or survival time when summarizing the longitudinal response. For example a pattern-mixture model characterizes the mean response at time t conditional on death at time S = s (for s > t), and thus uses future status as a predictor for the time t response. As an alternative, we define regression conditioning on being alive as a regression model that conditions on survival status, rather than a specific survival time. Such models may be referred to as partly conditional since the mean at time t is specified conditional on being alive (S > t), rather than using finer stratification (S = s for s > t). We show that naive use of standard likelihood-based longitudinal methods and generalized estimating equations with non-independence weights may lead to biased estimation of the partly conditional mean model. We develop a taxonomy for accommodation of both dropout and death, and describe estimation for binary longitudinal data that applies selection weights to estimating equations with independence working correlation. Simulation studies and an analysis of monthly disability status illustrate potential bias in regression methods that do not explicitly condition on survival.

Activities of Daily Living↗