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 397 records · Page 22Linked to original sources
PubMed2001

Application of robust estimating equations to the analysis of quantitative longitudinal data.

A model fit by general estimating equations (GEE) has been used extensively for the analysis of longitudinal data in medical studies. To some extent, GEE tries to minimize a quadratic form of the residuals, and therefore is not robust in the sense that it, like least squares estimates, is sensitive to heavy-tailed distributions, contaminated distributions and extreme values. This paper describes the family of truncated robust estimating equations and its properties for the analysis of quantitative longitudinal data. Like GEE, the robust estimating equations aim to assess the covariate effects in the generalized linear model in the complete population of observations, but in a manner that is more robust to the influence of aberrant observations. A simulation study has been conducted to compare the finite-sample performance of GEE and the robust estimating equations under a variety of error distributions and data structures. It shows that the parameter estimates based on GEE and the robust estimating equations are approximately unbiased and the type I errors of Wald tests do not tend to be inflated. GEE is slightly more efficient with pure normal data, but the efficiency of GEE declines much more quickly than the robust estimating equations when the data become contaminated or have heavy tails, which makes the robust estimating equations advantageous with non-normal data. Both GEE and the robust estimating equations are applied to a longitudinal analysis of renal function in the Diabetes Control and Complications Trial (DCCT). For this application, GEE seems to be sensitive to the working correlation specification in that different working correlation structures may lead to different conclusions about the effect of intensive diabetes treatment. On the other hand, the robust estimating equations consistently conclude that the treatment effect is highly significant no matter which working correlation structure is used. The DCCT Research Group also demonstrated a significant effect using a mixed-effects longitudinal model.

Albuminuria↗
PubMed1992

Likelihood-based analyses of longitudinal twin and family data: experiences with pedigree-based approaches.

Substantial progress has been made recently in analyses of longitudinal twin and family data, principally for two reasons. The first is the continuing development of more refined models for describing longitudinal data; the second is the widespread availability of analytic methods (e.g. LISREL) with which to implement these models. Computational restrictions have limited likelihood-based analyses of longitudinal genetic data to analyses of covariance matrices or mean squares; however, advances in computer technology now make it feasible to conduct likelihood analyses of longitudinal pedigree data. We consider potential advantages of using pedigree-based methods. Our initial experiences with the application of these methods to simulated twin data, using the FISHER (Lange, K.L., et al., Genet. Epidemiol. 5:471, 1988) quantitative genetics package, are discussed, with particular attention to practical details such as running times on several computers. Preliminary results of the pedigree-based analyses, including robust estimation methods, convincingly demonstrate the failure of methods assuming the multivariate normal distribution for simulated twin data provided by Carey (this issue).

Adolescent↗
PubMed1986

Longitudinal study of bone loss in the second metacarpal.

This longitudinal study was undertaken to ascertain the rate of bone loss and to identify aging, cohort and/or time effects on bone loss in male participants of the Baltimore Longitudinal Study of Aging. Hand-wrist radiographs were obtained from 1958-1981 and were evaluated for total width, medullary width, and length of the second metacarpal. Data were analyzed using an age-time matrix with 8-year intervals for three epochs and nine age groups. The bone measurements were analyzed in three perspectives (cross-sectional, longitudinal and time-series). The results demonstrate that there is both a cross-sectional and longitudinal loss of cortical bone with age in the second metacarpal. Furthermore, the results show that males lose approximately 14% of their cortical bone, at a rate of about 2% per decade, over the adult lifespan. The majority of this loss occurs between the ages of 45 and 69 and is due primarily to aging and is not an artifact of cohort differences or secular change.

Adult↗
PubMed2002

Dietary calcium intake and serum vitamin D are major determinants of bone mass variations in women. A longitudinal study.

BACKGROUND AND AIMS: Bone mineral density (BMD) is one of the main determinants in the pathogenesis of fractures. However, data on factors predicting longitudinal variations in BMD are still limited and incomplete. Such data would be of great importance in order to better focus prevention strategies in both the clinical setting and at the population level. The aim of the study was to investigate the predictive value of both serological and questionnaire variables for bone mass variations in healthy women participating in a population-based longitudinal study carried out in Napoli, Italy. METHODS: High completion rate (85.2%) and adequate sample size were obtained: 139 women (45 to 79 years of age) were examined at study entry and then again after two years (24 +/- 2 months) following the same protocol. They underwent medical examination, questionnaire, anthropometric measurements, blood sampling and urine collection. BMD was measured by dual energy X-ray absorptiometry (DEXA) at the lumbar spine (L1-L4) and femoral neck. Data analysis included calculation of the percent variation in BMD in the 2-year period. Longitudinal data underwent stepwise analysis for a global evaluation of mutual interactions between independent variables. RESULTS AND CONCLUSIONS: Our findings indicate that dietary and serum calcium, and serum 25(OH)vitamin D are the only independent determinants of BMD variations at the lumbar and femoral level, respectively. While the pharmacological significance of calcium and vitamin D in the therapy of established osteoporosis is still controversial, the present longitudinal data evidence their role as essential nutrients in determining the natural history of BMD variations.

Aged↗
PubMed2001

Correction of kyphotic deformity before and after transection of the anterior longitudinal ligament--a cadaver study.

With a custom-made measuring unit, two separate experiments, involving six and five cadaveric torsos with intact rib cages and sternums, respectively, were carried out to determine the effect of the transection of the anterior longitudinal ligament with and without osteodiscectomy and its influence on the thoracic kyphosis. The open or thoracoscopically assisted anterior release, as part of the operative treatment of scoliosis or kyphosis, usually consists of a transection of the anterior longitudinal ligament (ALL) and an additional discectomy. A complete osteodiscectomy, however, is not always possible with a minimally invasive approach. As part of our biomechanical research, we attempted to quantify the amount of correction achievable with a defined force prior to and following the isolated transection of the anterior longitudinal ligament. The aim of the study was to clarify whether or not an isolated transection of the anterior longitudinal ligament is sufficient to obtain an adequate anterior release of the spine. In the surgical treatment of kyphotic deformities, anterior release of the spine is performed in the form of a transection of the ALL and discectomy. Recently, video-assisted thoracic surgery has become increasingly popular in spine surgery. As part of this change in surgical technique, the question has arisen as to what extent an isolated transection of the ALL provides an adequate release of the thoracic spine. Eleven human spines were retrieved from fresh cadavers, dissected, and attached to a specially constructed apparatus. The spine was attached to the construct at the twelfth vertebral body. C6 and C7 were fixed in synthetic resin. We installed the instruments in such a manner as to reproducibly apply a torsional moment of 10 Nm to the spine. Motion was only permitted in the sagittal plane. Segmental transections of the ALL were carried out from T3 to T7. For comparison, the sagittal Cobb angle was also documented following an anterior release combined with an osteodiscectomy. With the isolated transection of the ALL, an average correction of the sagittal Cobb angle of 4 degrees in each functional spinal motion segment was recorded. In comparison, the additional osteodiscectomy led to a further average increase of only 2 degrees per level. The measurements performed on human cadavers showed that the isolated transection of the ALL leads to a sufficient anterior release of the thoracic spine, allowing a correction of the kyphotic deformity. The release with a concomitant osteodiscectomy represents a more time-consuming and more invasive procedure resulting in only a slightly greater amelioration of the sagittal Cobb angle, while being associated with a greater patient morbidity.

Aged↗
PubMed1982

Longitudinal studies of fetal growth with the use of dynamic image ultrasonography.

To characterize the growth of individual fetuses, serial measurements of the crown-rump length (CRL), biparietal diameter (BPD), head circumference (HC), and abdominal circumference (AC) were made at specified intervals of 2 to 3 weeks (between 6 to 8 weeks [conceptual age] and delivery) in 20 normal fetuses with known dates of conception. Values for the HC-AC ratio and the estimated fetal weight were also obtained from these data. Measurements of CRL were similar to those reported previously, except in those cases in which the follicular phase was prolonged or the fetal growth potential was reduced. Mathematical modeling demonstrated that the linear-cubic model was optimal for the BPD and HC growth curves, the linear model for the AC, and the linear-quadratic model for fetal weight. The HC-AC ratio could not be modeled adequately because of significant individual variability. Optimal models for average longitudinal growth curves were determined for the BPD, HC, AC, and fetal weight from individual growth curves. These average longitudinal curves were similar in shape, mean values, and variability to those obtained in previous cross-sectional studies. Statistical analysis indicated that the coefficients of the average longitudinal curves are not likely to change by more than 5% to 15% if the size of the sample is increased very significantly. These results indicate that the average longitudinal growth curves are good estimators of the true population growth curves.

Adult↗
PubMed1986

Explaining discrepancies between longitudinal and cross-sectional models.

Data from longitudinal studies may be analyzed both cross-sectionally and longitudinally. Discrepancies between estimates obtained from these analyses pose questions about the validity of cross-sectional estimates of change. In some cases these discrepancies are the result of period effects, cohort effects, or selective dropout. In others, they are the result of incomplete modeling of the process and are spurious rather than substantive. In this report, we show that when the true relation between a dependent variable and age is non-linear (e.g. quadratic), but is modeled as linear, the estimated age effect will be a function of the age distribution. In a continuous-time idealization, if the age distribution is Gaussian, the estimated age effects agree. If the age distribution is symmetric and the non-linearity is quadratic, cross-sectional and longitudinal results agree. Otherwise they do not. We illustrate these points by analysis of the relation between aging and pulmonary function in middle and old age using data from a large, prospective, longitudinal study.

Adult↗
PubMed2006

Early age of first sexual intercourse and affiliation with deviant peers predict development of SUD: a prospective longitudinal study.

OBJECTIVE: A recent study has reported that early sexual behavior predicts the development of substance use disorders (SUD). However, that relationship was considered by its authors to be only tentative, because it was based on cross-sectional data rather than longitudinal data. Another recent study reported that deviant activities of peers predict development of SUD, although that relationship has not yet been confirmed. The objective of this study was to assess whether early onset of sexual intercourse and affiliation with deviant peers serve as predictors of the development of SUD, using a prospective longitudinal study design. METHOD: We conducted a prospective longitudinal study of adolescents. These subjects were recruited at the age of 10-12years, with follow-up evaluations at ages 14, 16, 19, 22, and 25. The sample included 136 male subjects. Cox regression analyses were performed, with age of first intercourse, neurobehavioral disinhibition, exposure to drugs in the neighborhood, and deviant activities of peers as factors in the analyses. RESULTS: Earlier age at first intercourse and deviant activities of peers each predicted a significantly higher risk of subsequently developing a SUD (Wald=8.3, df=1, p=0.004; Wald=7.5, df=1, p=0.006, respectively). CONCLUSIONS: The results of this study confirm that early onset of sexual intercourse and affiliation with deviant peers predict the early development of substance use disorders, using a prospective longitudinal study design.

Adolescent↗
PubMed2004

Longitudinal myocardial velocity gradient derived from pulsed Doppler tissue imaging in AL amyloidosis: a sensitive indicator of systolic and diastolic dysfunction.

OBJECTIVE: This study was performed to clarify whether the longitudinal myocardial velocity gradient could give new insights into the features of congestive heart failure in patients with primary amyloidosis. METHODS: A total of 33 consecutive patients with biopsy specimen-proven primary amyloidosis were examined by pulsed Doppler tissue imaging. RESULTS: In all, 22 had evidence of heart involvement, of whom 11 had clinical congestive heart failure. Sample volumes were placed on basal and midventricle in the apical 2- and 4-chamber views. Peak systolic, and peak early and late diastolic wall-motion velocities were measured at each site. Longitudinal myocardial velocity gradients were calculated as the difference between basal and midmyocardial velocities. Single-point analysis of pulsed Doppler tissue imaging could not distinguish any difference among groups, whereas the longitudinal myocardial velocity gradients in systole and early diastole were significantly impaired in the patients with congestive heart failure compared with both the patients without cardiac involvement and those with cardiac amyloidosis without congestive heart failure. CONCLUSION: Longitudinal myocardial velocity gradient is a sensitive method for detecting systolic and diastolic dysfunction in cardiac amyloidosis and is superior to single-point analysis Doppler tissue imaging.

Adult↗
PubMed1998

The difference in anatomy of the lumbar anterior longitudinal ligament in young African-Americans and Scandinavians.

OBJECTIVE: To examine possible structural differences in the lumbar anterior longitudinal ligament between young African-American and Scandinavian people. DESIGN: The lumbar anterior longitudinal ligament was examined in 48 African-American and Scandinavian young men and women (17 to 30 years old). Apart from the selection of gender, age, height, and weight, all subjects were recruited consecutively during routine forensic autopsies. Sections of the ligament were obtained from the L1 and the L5 vertebra for thickness and width measurements and histologic preparations. The measurements were performed with a digital vernier caliper with an accuracy of 0.1 mm. RESULTS: The width of the L1 part of the lumbar anterior longitudinal ligament in the African-American group (27.5+/-1.8mm) was greater than that in the Scandinavian group (17.7+/-1.5mm) (p < .001), and the thickness of the L1 part in the African-American group (3.8+/-0.4mm) was greater than that in the Scandinavian group (3.3+/-0.3mm) (p < .001). The width of the L5 part in the African-American group (34.1+/-2.0mm) was greater than that in the Scandinavian group (25.0+/-1.5mm) (p < .001), and the thickness of the L5 part in the African-American group (3.3+/-0.3mm) was greater than that in the Scandinavian group (2.2+/-0.3mm) (p < .001). The cross-sectional area was greater in the African-American group in both sections (L1, African-American 106.4+/-13.7mm2 vs Scandinavian 52.8+/-7.9mm2; L5, African-American 112.2+/-14.6mm2 vs Scandinavian 55.6+/-6.7mm2) (p < .001). No histologic differences between the two groups were observed. CONCLUSIONS: The lumbar anterior longitudinal ligament was considerably wider and thicker, which yielded a markedly greater cross-sectional area, in African-American than in Scandinavian subjects. In contrast, the ligament appears to be histologically similar in the two groups. These are previously unrecognized differences in anatomy between young African-American and Scandinavian people.

Adolescent↗
PubMed1999

Longitudinal peri-implant clinical responses in anterior mandibles of female patients: a preliminary report.

STATEMENT OF PROBLEM: There is a need for specific documentation that successful osseointegration related to Brânemark implants can be maintained, despite an apparent clinically vulnerable peri-implant soft tissue status. PURPOSE: This study monitored longitudinal peri-implant clinical responses in the mandible and sought to question whether a relatively deep mucogingival pocket will give rise to a greater loss of peri-implant bone than a shallow pocket. METHODS AND MATERIAL: Subjects, 8 women patients (mean age 62 years), were treated with Brânemark mandibular osseointegrated implant-supported prostheses. Four of these subjects had limited (average <3.5 mm) and 4 had unlimited (average > or =3.5 mm) peri-implant pocket probing depths and constituted the control and test subgroups, respectively. Longitudinal changes in peri-implant PAL and radiographic bone support were assessed. RESULTS: Overall probing attachment levels (PALs) of the peri-implant mucogingival complex showed little change. The PAL loss was only minimally significant within the control subjects, and not significant for test subjects. The difference in these mucogingival responses between test and control subjects was significant (P = .04). There was no significant overall longitudinal change in peri-implant bone levels. The longitudinal change of peri-implant bone level was not significant within or between the control and test subjects.

Aged↗
PubMed2006

The association between lower urinary tract symptoms and renal function in men: a cross-sectional and 5-year longitudinal analysis.

PURPOSE: We assessed the association between LUTS and renal function in men in a cross-sectional and a longitudinal study. MATERIALS AND METHODS: Men participating in health investigation in Vienna entered this study. In the cross-sectional analysis a consecutive series of men were studied and in the longitudinal analysis men were reevaluated after 5 years. LUTS were assessed by I-PSS and renal function was evaluated by GFR, as calculated by the Cockcroft-Gault equation. RESULTS: A total of 2.469 men with a mean age of 47.1 years (range 30 to 80) entered the cross-sectional study and 439 with a mean age of 51.7 years (range 45 to 79) could be assessed in longitudinal analysis. In the cross-sectional study there was no association between the degree of LUTS and GFR during 5 life decades (p = 0.55). An identical pattern was observed for irritative and obstructive scores. In the longitudinal cohort mean GFR +/- SD decreased from 84.3 +/- 20.2 ml per minute at baseline to 79.2 +/- 18.7 ml per minute after 5 years (-6.0%, p <0.0001). The mean decrease in GFR after 5 years was 4.5 ml per minute (-5.4%) in men without/mild LUTS (I-PSS 7 or less at ages 55.4 +/- 11.0 years), 3.9 ml per minute (-4.9%) in those with moderate LUTS (I-PSS 8 to 20 at ages 61.3 +/- 11.6 years) and 4.2 ml per minute (-5.2%) in those with severe LUTS (I-PSS greater than 20 at ages 64.3 +/- 7.4 years). On linear regression analysis in the 2 study cohorts neither I-PSS, nor obstructive or irritative score affected GFR (p >0.05). The only determinants for GFR less than 90 ml per minute were age (p <0.0001) and hypertension (p <0.0001). CONCLUSIONS: LUTS do not represent an independent risk factor for impaired renal function in men.

Adult↗
PubMed2003

Neuroanatomical abnormalities before and after onset of psychosis: a cross-sectional and longitudinal MRI comparison.

BACKGROUND: Psychotic disorders, such as schizophrenia, are associated with neuroanatomical abnormalities, but whether these predate the onset of symptoms or develop progressively over the course of illness is unclear. We investigated this issue with MRI to study people with prodromal symptoms who are at ultra high-risk for the development of psychosis. METHODS: We did two comparisons, cross-sectional and longitudinal. For the cross-sectional comparison, 75 people with prodromal signs of psychosis were scanned with MRI. After at least 12 months of follow-up, 23 (31%) had developed psychosis and 52 (69%) had not. Baseline MRI data from these two subgroups were compared. For the longitudinal comparison, 21 of the ultra high-risk individuals were scanned again with MRI after at least 12 months. Ten of these had developed psychosis and 11 had not. MRI data from baseline and follow-up were compared within each group of people. FINDINGS: In the cross-sectional comparison, compared with people who did not develop psychosis, those who did develop the disorder had less grey matter in the right medial temporal, lateral temporal, and inferior frontal cortex, and in the cingulate cortex bilaterally. In the longitudinal comparison, when re-scanned, individuals who had developed psychosis showed a reduction in grey matter in the left parahippocampal, fusiform, orbitofrontal and cerebellar cortices, and the cingulate gyri. In those who had not become psychotic, longitudinal changes were restricted to the cerebellum. INTERPRETATION: Some of the grey-matter abnormalities associated with psychotic disorders predate the onset of frank symptoms, whereas others appear in association with their first expression.

Adult↗
PubMed1996

Incidence and progression of nuclear opacities in the Longitudinal Study of Cataract.

PURPOSE: To estimate incidence and progression rates of nuclear opacities in the Longitudinal Study of Cataract, an epidemiologic study of the natural history of all types of lens opacities. METHODS: The Lens Opacities Classification System III was used to assess longitudinal changes between baseline and follow-up lens photographs for the 764 Longitudinal Study of Cataract participants. Baseline data, collected until December 1988 as part of a case-control study, included color slit, retroillumination, and Scheimpflug photographs. The same data were collected by the longitudinal Study of Cataract at four subsequent visits at yearly intervals. RESULTS: Among patients free of nuclear opacities at baseline, the incidence of new opacities was 6% after 2 years and 8% after 5 years of follow-up. The progression of pre-existing nuclear opacities was much higher. After 2 years, nuclear opacities had progressed in more than one third of the patients with pre-existing opacities; after 5 years, almost half had progressed. Older age was significantly related to higher incidence of new nuclear opacities, but not to progression of pre-existing opacities. Patients with other opacity types had higher nuclear incidence and progression rates. CONCLUSIONS: In this clinic-based, older-patient population, new nuclear opacities developed in less than one tenth of the patients after 5 years of follow-up. In contrast, almost one half of the patients with pre-existing opacities had worsened after 5 years. These estimated rates can be used to plan intervention or other studies of nuclear changes in similar populations.

Aged↗
PubMed2002

Longitudinal cerebrospinal fluid tau load increases in mild cognitive impairment.

Cross-sectional cerebrospinal fluid (CSF) levels of tau and amyloid (A) beta (beta) are of diagnostic importance for Alzheimer's disease (AD) and mild cognitive impairment (MCI). However, most longitudinal studies of tau fail to demonstrate progression. Because predominantly brain-derived proteins such as tau, have higher ventricle to lumbar ratios, we hypothesized that adjusting for the ventricular enlargement of AD would correct for the dilution of tau, and improve detection of longitudinal change. Abeta which is not exclusively brain derived, shows a ratio <1, and no benefit was expected from adjustment. In a 1 year longitudinal study of eight MCI and ten controls, we examined CSF levels of hyperphosphorylated (P) tau231, Abeta40, and Abeta42. In cross-section, MCI patients showed elevated Ptau231 and Abeta40 levels, and greater ventricular volumes. Longitudinally, only after adjusting for the ventricular volume and only for Ptau231, were increases seen in MCI. Further studies are warranted on mechanisms of tau clearance and on using imaging to interpret CSF studies.

Aged↗
PubMed2002

Late urologic morbidity in 177 consecutive patients after radiotherapy for cervical carcinoma: a longitudinal study.

PURPOSE: To provide longitudinal data on urologic morbidity after radiotherapy and brachytherapy for cervical carcinoma. METHODS AND MATERIALS: Five-year longitudinal urologic morbidity data were recorded from 177 consecutive patients of median age 59 years (range: 22-86 years) with cervical carcinoma receiving radiotherapy with curative intent at the Copenhagen University Hospital, Denmark. FIGO stages (%) were as follows: Stage I (15), Stage II (30), Stage III (54), and Stage IV (1). Late morbidity was calculated as cumulative incidence based on actuarial estimates. RESULTS: The 5-year cumulative incidence based on actuarial estimates of urologic morbidity Grades 1 + 2 + 3, Grades 2 + 3, and Grade 3 were 62%, 32%, and 5%, respectively. Frequencies of urologic morbidity in the 54 recurrence-free survivors at the end of follow-up indicated some reversibility in the case of Grades 1 and 2 morbidity. CONCLUSION: With the longitudinal design used in the present study, a rate of mild and moderate morbidity higher than that found in most of the previously reported literature was observed, giving cause for concern and underlining the importance of further longitudinal studies on this subject, specifically studies that relate to the background urologic morbidity in the female population, as well as to the fact that urologic morbidity might regress.

Adult↗
PubMed2001

Clustering of risk factors for coronary heart disease. the longitudinal relationship with lifestyle.

PURPOSE: The purpose of this study was to investigate whether or not clustering of biological coronary heart disease (CHD) risk factors exists and to investigate the longitudinal relationship between lifestyle parameters (dietary intake, daily physical activity, smoking behaviour, alcohol consumption) and a biological CHD risk factor clustering score. This was defined as belonging to one or more gender specific 'high risk' quartiles for the following CHD risk factors: ratio between total serum cholesterol and high density lipoprotein cholesterol (TC:HDL), mean arterial blood pressure (MABP), body fatness [sum of skinfolds (SSF)], and cardiopulmonary fitness (VO2-max). METHODS: The data were derived from the Amsterdam Growth and Health Study, an observational longitudinal study in which six repeated measurements were carried out over a period of 15 years covering adolescence and young adulthood. The longitudinal relationships were analysed with generalized estimating equations. RESULTS: The results showed significant clustering for the TC:HDL ratio, SSF, and VO(2)-max. MABP was not significantly associated with the other CHD risk factors. Daily physical activity and alcohol consumption (only for males) were both inversely related to the clustering score. None of the other lifestyle parameters showed significant relationships with the clustering score. CONCLUSIONS: Based on this small longitudinal study, it can be stated that during adolescence and young adulthood both daily physical activity and alcohol consumption were related to a healthy CHD risk profile.

Adolescent↗
PubMed2004

Longitudinal data analysis. A comparison between generalized estimating equations and random coefficient analysis.

The analysis of data from longitudinal studies requires special techniques, which take into account the fact that the repeated measurements within one individual are correlated. In this paper, the two most commonly used techniques to analyze longitudinal data are compared: generalized estimating equations (GEE) and random coefficient analysis. Both techniques were used to analyze a longitudinal dataset with six measurements on 147 subjects. The purpose of the example was to analyze the relationship between serum cholesterol and four predictor variables, i.e., physical fitness at baseline, body fatness (measured by sum of the thickness of four skinfolds), smoking and gender. The results showed that for a continuous outcome variable, GEE and random coefficient analysis gave comparable results, i.e., GEE-analysis with an exchangeable correlation structure and random coefficient analysis with only a random intercept were the same. There was also no difference between both techniques in the analysis of a dataset with missing data, even when the missing data was highly selective on earlier observed data. For a dichotomous outcome variable, the magnitude of the regression coefficients and standard errors was higher when calculated with random coefficient analysis then when calculated with GEE-analysis. Analysis of a dataset with missing data with a dichotomous outcome variable showed unpredictable results for both GEE and random coefficient analysis. It can be concluded that for a continuous outcome variable, GEE and random coefficient analysis are comparable. Longitudinal data-analysis with dichotomous outcome variables should, however, be interpreted with caution, especially when there are missing data.

Adipose Tissue↗