SEARCH · PubMed Health

Results for “longitudinal data”

Explore indexed PubMed citations for clinical trials, systematic reviews and public health research. Read source abstracts and follow each citation to its original PubMed record.

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

At least 163 records · Page 9Linked to original sources

PubMed2001

Longitudinal data analysis: an application to construction of a natural history profile of Duchenne muscular dystrophy.

A 30-month prospective study of 27 Scandinavian boys with confirmed diagnosis of Duchenne muscular dystrophy was carried out to construct profiles of the natural history of the disease. Assessments which included measures of voluntary muscle strength and function were done at 3 monthly intervals except for the first and second which were separated by 1 month. Recently developed statistical methods for analysis of longitudinal data with repeated observations on the same individual were used avoiding the problem of induced serial correlations. This allowed for the construction of both reference and prediction profiles for the variables %MRC, motor ability, walking time for 10 m and the sum of myometry of seven muscle groups.

Child↗

PubMed1998

Temporal relations between obesity and insulin: longitudinal data from the Normative Aging Study.

Although obesity and insulin levels are generally associated in cross-sectional data, the temporal and causal nature of their association is not yet clear. Increased obesity may have preceded increased insulin levels or vice versa. The authors examined the temporal relations between fasting insulin blood levels and weight in longitudinal data from the ongoing Normative Aging Study. Two insulin measurements from which a rate of change (delta Insulin) could be calculated were available from 376 non-diabetic male subjects (mean age = 62.1 years). Rate of change in weight could be calculated for the previous inter-examination period (delta Weight1), the contemporaneous period (delta Weight2), and the inter-examination period following the second insulin measurement (delta Weight3). delta Weight2 was a significant predictor (p = 0.0005) of delta insulin in multiple linear regression models that included control for potential confounders (body mass index, waist-to-hip ratio, antihypertensive and diuretic medication use, and age) and for correlation between the initial level and change in insulin (mean fasting insulin). delta Weight1 was added to the model and was found not to be statistically significant (p = 0.15). When the model was stratified by age tertile, the regression coefficient on delta Weight1 was -0.44 (p = 0.018) for the youngest stratum, -0.06 (p = 0.72) for the middle stratum, and 0.21 (p = 0.19) for the oldest men. Similarly, delta Insulin was a significant predictor of delta Weight3 (p = 0.026) in a separate regression model. These findings are consistent with both possible temporal sequences of association between changes in insulin and obesity. The intricate homeostatic mechanisms that regulate changes in insulin and obesity may not be readily amenable to description in terms of cause and effect.

Aged↗

PubMed2001

Linear mixed models with flexible distributions of random effects for longitudinal data.

Normality of random effects is a routine assumption for the linear mixed model, but it may be unrealistic, obscuring important features of among-individual variation. We relax this assumption by approximating the random effects density by the seminonparameteric (SNP) representation of Gallant and Nychka (1987, Econometrics 55, 363-390), which includes normality as a special case and provides flexibility in capturing a broad range of nonnormal behavior, controlled by a user-chosen tuning parameter. An advantage is that the marginal likelihood may be expressed in closed form, so inference may be carried out using standard optimization techniques. We demonstrate that standard information criteria may be used to choose the tuning parameter and detect departures from normality, and we illustrate the approach via simulation and using longitudinal data from the Framingham study.

Biometry↗

PubMed2005

The effect of dislike of school on risk of teenage pregnancy: testing of hypotheses using longitudinal data from a randomised trial of sex education.

STUDY OBJECTIVE: To examine whether attitude to school is associated with subsequent risk of teenage pregnancy. To test two hypotheses that attitude to school is linked to pregnancy via pathways involving young people having "alternative" expectations or deficits in sexual health knowledge and confidence. DESIGN: Analysis of longitudinal data arising from a trial of sex education. Examination of associations between attitude to school and protected first sex, unprotected first sex, unprotected and protected last sex, and pregnancy, both crude and adjusting in turn for expectation of parenting by age 20, lack of expectation of education/training at age 20, and sexual health knowledge and confidence. SETTING: Schools in central and southern England. PARTICIPANTS: Girls of median age 13.7 years at baseline, 14.7 years at follow up 1, and 16.0 years at follow up 2. MAIN RESULTS: In unadjusted analysis, attitude to school was significantly associated with protected and unprotected first sex by follow up 1, protected first sex between follow up 1 and 2, unprotected last sex, and pregnancy. Dislike of school was more strongly associated with increased risk of these outcomes than was ambivalence to school. These associations remained after adjusting for socioeconomic status and for expectation of parenting, lack of expectation of education/training, and various indicators of knowledge and confidence about sexual health. CONCLUSIONS: Dislike of school is associated with subsequent increased risk of teenage pregnancy but the mechanism underlying any possible causal link is unlikely to involve "alternative" expectations or deficits in sexual health knowledge or confidence.

Adolescent↗

PubMed2000

Application of a generalized random effects regression model for cluster-correlated longitudinal data to a school-based smoking prevention trial.

In cluster-randomized trials, groups of subjects (clusters) are assigned to treatments, whereas observations are taken on the individual subjects. Since observations on subjects in the same cluster are typically more similar than observations from different clusters, analyses of such data must take intracluster correlation into account rather than assuming independence among all observations. Random effects models are useful for this purpose. The problem becomes more complicated if, in addition, repeated observations are taken on subjects over time. This introduces intraindividual correlation, which is typical for longitudinal studies. The Waterloo Smoking Prevention Project, study 3 (WSPP3), 1989-1996, is a study giving rise to cluster-correlated longitudinal data, where schools were randomized to either a smoking intervention program or to a control condition. Smoking status was assessed on grade 6 students in these schools, with annual follow-up observations throughout elementary and high school years. The authors illustrate the use of a generalized random effects model for analyzing this type of data. This model obtains appropriate estimates and standard errors for both individual-level covariates and those at the level of the cluster.

Adolescent↗

PubMed2005

Use of total lymphocyte count for informing when to start antiretroviral therapy in HIV-infected children: a meta-analysis of longitudinal data.

BACKGROUND: Total lymphocyte count has been proposed as an alternative to the percentage of CD4+ T-cells to indicate when antiretroviral therapy should be started in children with HIV in resource-poor settings. We aimed to assess thresholds of total lymphocyte count at which antiretroviral therapy should be considered, and compared monitoring of total lymphocyte count with monitoring of CD4-cell percentage. METHODS: Longitudinal data on 3917 children with HIV infection were pooled from observational and randomised studies in Europe and the USA. The 12-month risks of death and AIDS by most recent total lymphocyte count and age were estimated by parametric survival models, based on measurements before antiretroviral therapy or during zidovudine monotherapy. Risks were derived and compared at thresholds of total lymphocyte count and CD4-cell percentage for starting antiretroviral therapy recommended in WHO 2003 guidelines. FINDINGS: Total lymphocyte count was a powerful predictor of the risk of disease progression despite a weak correlation with CD4-cell percentage (r=0.08-0.19 dependent on age). For children older than 2 years, the 12-month risk of death and AIDS increased sharply at values less than 1500-2000 cells per muL, with little trend at higher values. Younger children had higher risks and total lymphocyte count was less prognostic. Mortality risk was substantially higher at thresholds of total lymphocyte count recommended by WHO than at corresponding thresholds of CD4-cell percentage. When the markers were compared at the threshold values at which mortality risks were about equal, total lymphocyte count was as effective as CD4-cell percentage for identifying children before death, but resulted in an earlier start of antiretroviral therapy. INTERPRETATION: In this population, total lymphocyte count was a strong predictor of short-term disease progression, being only marginally less predictive than CD4-cell percentage. Confirmatory studies in resource-poor settings are needed to identify the most cost-effective markers to guide initiation of antiretroviral therapy.

Aging↗

PubMed2003

Cost of dementia: impact of disease progression estimated in longitudinal data.

AIMS: Several studies have shown that health care costs are higher for demented than for non-demented persons and that health care costs are higher for more severe demented persons than less severe demented persons. However, most studies report on cross-sectional study designs, and thus fail to examine the influence of disease progression on changes in health care costs to individual persons. The objective of this study was, using longitudinal data, to examine changes in total health care costs with disease progression in demented persons. METHODS: We assumed that disease progression could be characterised by transitions between different states of dementia which reflected the degree to which the disease progressed over time. Then, changes in health care costs were regressed on a set of explanatory variables including disease progression. A total of 465 demented and non-demented persons were interviewed twice. The time between interviews was about three years. Before each interview, the participant was examined for dementia and classified by type of dementia (Alzheimer's disease, vascular or other types of dementia) and degree of dementia (very mild, mild, moderate, severe). RESULTS: The results of this longitudinal study confirmed that health care costs increased over time for non-demented as well as for demented persons and that health care costs increased with disease progression. In particular, the health care costs increased when the disease had progressed into the severe state of the dementia. Also, decline in functional abilities was an important factor for explaining changes in health care costs.

Activities of Daily Living↗

PubMed2000

A mixture model for longitudinal data with application to assessment of noncompliance.

In clinical trials of a self-administered drug, repeated measures of a laboratory marker, which is affected by study medication and collected in all treatment arms, can provide valuable information on population and individual summaries of compliance. In this paper, we introduce a general finite mixture of nonlinear hierarchical models that allows estimates of component membership probabilities and random effect distributions for longitudinal data arising from multiple subpopulations, such as from noncomplying and complying subgroups in clinical trials. We outline a sampling strategy for fitting these models, which consists of a sequence of Gibbs, Metropolis-Hastings, and reversible jump steps, where the latter is required for switching between component models of different dimensions. Our model is applied to identify noncomplying subjects in the placebo arm of a clinical trial assessing the effectiveness of zidovudine (AZT) in the treatment of patients with HIV, where noncompliance was defined as initiation of AZT during the trial without the investigators' knowledge. We fit a hierarchical nonlinear change-point model for increases in the marker MCV (mean corpuscular volume of erythrocytes) for subjects who noncomply and a constant mean random effects model for those who comply. As part of our fully Bayesian analysis, we assess the sensitivity of conclusions to prior and modeling assumptions and demonstrate how external information and covariates can be incorporated to distinguish subgroups.

Anti-HIV Agents↗

PubMedDate not supplied

Methods for the analysis of informatively censored longitudinal data.

This paper describes the problem of informative censoring in longitudinal studies where the primary outcome is rate of change in a continuous variable. Standard approaches based on the linear random effects model are valid only when the data are missing in a non-ignorable fashion. Informative censoring, which is a special type of non-ignorably missing data, occurs when the probability of early termination is related to an individual subject's true rate of change. When present, informative censoring causes bias in standard likelihood-based analyses, as well as in weighted averages of individual least-squares slopes. This paper reviews several methods proposed by others for analysis of informatively censored longitudinal data, and outlines a new approach based on a log-normal survival model. Maximum likelihood estimates may be obtained via the EM algorithm. Advantages of this approach are that it allows general unbalanced data caused by staggered entry and unequally-timed visits, it utilizes all available data, including data from patients with only a single measurement, and it provides a unified method for estimating all model parameters. Issues related to study design when informative censoring may occur are also discussed.

Linear Models↗

PubMed2006

Age trajectories of grip strength: cross-sectional and longitudinal data among 8,342 Danes aged 46 to 102.

PURPOSE: The purpose is to study the age trajectory of hand-grip strength after the age of 45 years. METHODS: In this study, we use data from three large nationwide population-based surveys of Danes aged 45 to 102 years with a total of 8342 participants with grip-strength measurements and up to 4 years of follow-up. Grip strength was measured by using a portable hand dynamometer. RESULTS: Grip strength declines throughout life for both males and females, but among the oldest women, the longitudinal curve reaches a horizontal plateau. The course of the decline is estimated by using full information in the longitudinal data and is found to be almost linear in the age span of 50 to 85 years. In this age span, mean annual grip-strength loss is estimated to be 0.59 (0.02) (SE) kg for men and 0.31 (0.01) kg for women. CONCLUSION: This study confirms the previously reported grip-strength decline with increasing age. Estimates were obtained by using full-information methods from large population-representative studies. Equations of expected grip strength, as well as tables with sex-, age-, and height-stratified reference data, provide an opportunity to include grip-strength measurement in clinical care in similar populations.

Aged↗

PubMed2001

Bayesian discrimination with longitudinal data.

The motivation for the methodological development is a double-blind clinical trial designed to estimate the effect of regular injection of growth hormone, with the purpose of identifying growth hormone abusers in sport. The data formed part of a multicentre investigation jointly sponsored by the European Union and the International Olympic Committee. The data are such that for each individual there is a matrix of marker variables by time point (nominally 8 markers at each of 7 time points). Data arise out of a double-blind trial in which individuals are given growth hormone at one of two dose levels or placebo daily for 28 days. Monitoring by means of blood samples is at 0, 21, 28, 30, 33, 42 and 84 days. We give a new method of Bayesian discrimination for multivariate longitudinal data. This involves a Kronecker product covariance structure for the time by measurements (markers) data on each individual. This structure is estimated by an empirical Bayes approach, using an ECM algorithm, within a Bayesian Gaussian discrimination model. In future one may have markers for an individual at one or more time points. The method gives probabilities that an individual is on placebo or on one of the two dose regimes.

Journal Article↗

PubMed2003

Psychiatric disorders in adolescents with developmental disabilities: longitudinal data on diagnostic disagreement in 150 clients.

The current paper describes the prevalence of psychiatric diagnoses in a large sample (n = 150) of adolescents with developmental disabilities who were hospitalized for inpatient psychiatric treatment. Differential diagnoses made during their inpatient stay in a specialty unit for the assessment and treatment of dually diagnosed adolescents are presented and contrasted with longitudinal/historical data on these same patients' diagnoses prior to admission. Results indicate that these individuals received a wide spectrum of diagnoses during their adolescent years. The paper offers indirect support that correctly diagnosing psychiatric conditions is often challenging in adolescents with developmental disabilities. Factors related to diagnostic complexity and misdiagnoses (false positives and false negatives) are discussed. Longitudinal data on psychotropic medication usage for these individuals are also presented.

Adolescent↗

PubMed2003

Comparison of the linkage results of two phenotypic constructs from longitudinal data in the Framingham Heart Study: analyses on data measured at three time points and on the average of three measurements.

BACKGROUND: Family studies are often conducted in a cross-sectional manner without long-term follow-up data. The relative contribution of a gene to a specific trait could change over the lifetime. The Framingham Heart Study offers a unique opportunity to investigate potential gene x time interaction. We performed linkage analysis on the body mass index (BMI) measured in 1970, 1978, and 1986 for this project. RESULTS: We analyzed the data in two different ways: three genome-wide linkage analyses on each exam, and one genome-wide linkage analysis on the mean of the three measurements. Variance-component linkage analyses were performed by the SOLAR program. Genome-wide scans show consistent evidence of linkage of quantitative trait loci (QTLs) on chromosomes 3, 6, 9, and 16 in three measurements with a maximum multipoint LOD score > 2.2. However, only chromosome 9 has a LOD score = 2.14 when the mean values were analyzed. More interestingly, we found potential gene x environment interactions: increasing LOD scores with age on chromosomes 3, 9, and 16 and decreasing LOD scores on chromosome 6 in the three exams. CONCLUSION: The results indicate two points: 1) it is possible that a gene (or genes) influencing BMI is (are) up- or down-regulated as people aged due to aging process or changes in lifestyle, environments, or genetic epistasis; 2) using mean values from longitudinal data may reduce the power to detect linkage and may have no power to detect gene x time, and/or gene x gene interactions.

Adult Children↗

PubMed2000

Analysis of proximal femur DXA scans in growing children: comparisons of different protocols for cross-sectional 8-month and 7-year longitudinal data.

Dual-energy X-ray absorptiometry (DXA) is a widely used method for measuring bone mineral in the growing skeleton. Because scan analysis in children offers a number of challenges, we compared DXA results using six analysis methods at the total proximal femur (PF) and five methods at the femoral neck (FN). In total we assessed 50 scans (25 boys, 25 girls) from two separate studies for cross-sectional differences in bone area, bone mineral content (BMC), and areal bone mineral density (aBMD) and for percentage change over the short term (8 months) and long term (7 years). At the proximal femur for the short-term longitudinal analysis, there was an approximate 3.5% greater change in bone area and BMC when the global region of interest (ROI) was allowed to increase in size between years as compared with when the global ROI was held constant. Trend analysis showed a significant (p < 0.05) difference between scan analysis methods for bone area and BMC across 7 years. At the femoral neck, cross-sectional analysis using a narrower (from default) ROI, without change in location, resulted in a 12.9 and 12.6% smaller bone area and BMC, respectively (both p < 0.001). Changes in FN area and BMC over 8 months were significantly greater (2.3%, p < 0.05) using a narrower FN rather than the default ROI. Similarly, the 7-year longitudinal data revealed that differences between scan analysis methods were greatest when the narrower FN ROI was maintained across all years (p < 0.001). For aBMD there were no significant differences in group means between analysis methods at either the PF or FN. Our findings show the need to standardize the analysis of proximal femur DXA scans in growing children.

Absorptiometry, Photon↗

PubMed2006

Flexible modeling via a hybrid estimation scheme in generalized mixed models for longitudinal data.

To circumvent the computational complexity of likelihood inference in generalized mixed models that assume linear or more general additive regression models of covariate effects, Laplace's approximations to multiple integrals in the likelihood have been commonly used without addressing the issue of adequacy of the approximations for individuals with sparse observations. In this article, we propose a hybrid estimation scheme to address this issue. The likelihoods for subjects with sparse observations use Monte Carlo approximations involving importance sampling, while Laplace's approximation is used for the likelihoods of other subjects that satisfy a certain diagnostic check on the adequacy of Laplace's approximation. Because of its computational tractability, the proposed approach allows flexible modeling of covariate effects by using regression splines and model selection procedures for knot and variable selection. Its computational and statistical advantages are illustrated by simulation and by application to longitudinal data from a fecundity study of fruit flies, for which overdispersion is modeled via a double exponential family.

Animals↗

PubMed1999

Comparing three preprocessing strategies for longitudinal data. An example in functional outcomes research.

Longitudinal monitoring of individual patient data is becoming routine in physician office practice. This study compares three different methods for evaluating clinical outcomes for individual patients: raw change score analysis versus normative and ipsative statistical analyses. Two discrete samples of intermittent claudication patients making vascular surgery office visits--drawn from interventional management versus stable, routinely followed control groups--were tested four times using both generic and disease-specific functional status measures. Results indicated that the ipsative method was most consistent with several different types of a priori hypotheses that are often evaluated in analysis of repeated measures data.

Evaluation Studies as Topic↗

PubMed2000

Analysis of messy longitudinal data from a randomized clinical trial. MRC Lung Cancer Working Party.

The randomized clinical trial, LU19, conducted by the Medical Research Council Lung Cancer Working Party, was designed to compare ACE (doxorubicin, cyclophosphamide and etoposide) chemotherapy plus G-CSF (granulocyte colony-stimulating factor) at 2-week intervals versus ACE chemotherapy alone at standard 3-week intervals in patients with small-cell lung cancer. This trial investigated whether more intensive administration of ACE would improve overall survival and affect the quality of life of patients. The report on overall survival and other outcome measures will be published in the Journal of Clinical Oncology. In this paper we focus on methods of analysing aspects of data reflecting quality of life. Twelve symptoms of lung cancer and its treatment - cough, haemoptysis, pain, nausea, vomiting, hoarse voice, sore mouth, rash, lethargy, lack of appetite, alopecia, and dysphagia - were scheduled to be assessed on seven occasions for the ACE arm and on eight occasions for the ACE+G-CSF arm by clinicians during the first 18 weeks of the treatment period. However, in practice the number of assessment forms completed per patient ranged from 1 to 9, and assessment time-points were very different from those planned. These 'messy' longitudinal data are explored by both a summary measure approach, in which experience of a symptom is summarized by a single value, and an extensive model-based statistical approach, which explicitly takes into account correlation within repeated measures. These analyses provide a clear picture of symptom comparisons between the two treatments. The application of various methods offers not only an approach to assessing the robustness of the results but also a basis for investigating reasons for inconsistency of results across methods. We conclude that except lethargy, which is worse in the ACE+G-CSF arm, all symptoms are similar across the two arms during the treatment period.

Amsacrine↗

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↗