SEARCH · PubMed Health

Results for “missing 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 1,405 records · Page 78Linked to original sources

PubMed2007

A mixed-effects model for cognitive decline with non-monotone non-response from a two-phase longitudinal study of dementia.

Most longitudinal studies of elderly are characterized by substantial drop-out due to death and many other factors beyond the control of the investigators. In a two-phase longitudinal study of dementia, subjects with cognitive impairment skip the first phase survey in the next follow-up, leading to intermittent missing variables measured in that phase. In the context of analysing pre-dementia cognitive decline in an elderly population, both of the two causes of non-response can potentially be informative in the sense that the missingness is dependent on the unobserved outcome. To take these factors into account, mixed-effects models are constructed to allow the outcome and the multiple causes of missing values to share the same 'random parameter' or random effect. The crucial assumption of our model is that the random effects of the model for the outcome and that of the model for the missing-data indicators are linked in a deterministic manner. It can be thought of as an approximation of a more general and realistic situation, in which the two models have distinct, yet dependent, random effects. We conduct a simulation study to investigate possible deviations of the estimates under such a scenario. A second simulation illustrates the magnitude of the bias in estimating the difference of decline rate between two groups when the random effects are linked in different manners for the two groups.

Aged↗

PubMed2007

A random-effects Markov transition model for Poisson-distributed repeated measures with non-ignorable missing values.

In biomedical research with longitudinal designs, missing values due to intermittent non-response or premature withdrawal are usually 'non-ignorable' in the sense that unobserved values are related to the patterns of missingness. By drawing the framework of a shared-parameter mechanism, the process yielding the repeated count measures and the process yielding missing values can be modelled separately, conditionally on a group of shared parameters. For chronic diseases, Markov transition models can be used to study the transitional features of the pathologic processes. In this paper, Markov Chain Monte Carlo algorithms are developed to fit a random-effects Markov transition model for incomplete count repeated measures, within which random effects are shared by the counting process and the missing-data mechanism. Assuming a Poisson distribution for the count measures, the transition probabilities are estimated using a Poisson regression model. The missingness mechanism is modelled with a multinomial-logit regression to calculate the transition probabilities of the missingness indicators. The method is demonstrated using both simulated data sets and a practical data set from a smoking cessation clinical trial.

Algorithms↗

PubMed1990

A Markovian model for comparing incidences of side effects.

For clinical trials that entail observations at successive visits for the occurrence of a side effect, this paper considers a likelihood-based method to compare side effect incidence rates. The method, which employs the assumption of a Markov chain of order one for the vectors of binary responses, handles missing data due to premature withdrawals. An actual numerical example and a simulated example illustrate the technique.

Clinical Trials as Topic↗

PubMed1990

The measurement of change of quality of life in clinical trials.

A model is presented for the measurement of change of quality of life in clinical trials with time under the influence of one or more treatments. Quality of life is regarded as a multidimensional latent variable, and is measured through dichotomous item responses on a number of points in time. Change of quality of life is 'explained' with a latent logistic regression model which may include parameters for the time process, the effects of clinical treatments, and interaction parameters. By assuming the absence of patient/time interaction within treatment groups, the parameters of the time process and the treatment effects can be estimated independently of the latent quality of life parameters at the start of the treatment. Consequently, differential mortality, censoring mechanisms, and other mechanisms causing missing data can be ignored.

Breast Neoplasms↗

PubMed1991

A bibliography and comments on the use of statistical models in epidemiology in the 1980s.

This paper reviews developments in statistical modelling in epidemiology in the 1980's, with emphasis on cohort and case-control studies. The central roles of the logistic and proportional hazard models are highlighted, and it is shown how these models lead to a deeper understanding of classical designs and methods of analysis as well as to efficient new designs and analytical procedures. The important area of model misspecification is discussed, including the problems of omitted latent structure, mis-modelling of available measurements, missing data and errors in measurements. Various designs motivated by the logistic model are illustrated numerically, and designs based on the proportional hazards model are discussed, as are papers on sample size determination. There are brief introductions to the literature on other topics, including attributable risk, disease clustering, family studies and genetics, analysis of disease incidence data, infectious disease, longitudinal data, screening and miscellaneous related topics in statistics. An extensive bibliography is indexed according to the outline of the paper.

Data Interpretation, Statistical↗

PubMed1993

Time-period effects in longitudinal studies measuring average rates of change.

Random time-period effects are unexplained increases or decreases in the observed value for all individuals measured at a particular time point in a longitudinal study. They can be caused by learning effects, changes in equipment, personnel and overall subject co-operation. We investigate the consequences of time-period effects in random coefficient regression models, where interest is in the average rate of change (slope) of a continuous outcome. In a study with a single group of subjects, they can lead to conditionally biased estimates of the mean slope and its variance (conditional on the time-period effects). Calculations suggest that the increase in sample size required to maintain a specified precision of the mean slope estimate over repeated studies may be substantial. In a study with a concurrent control group, however, time-period effects do not distort the expectation, estimated variance or the distribution of the difference between the mean slopes. With missing data, in addition to time-period effects, an unbiased estimate of a single mean slope remains problematic, but one can use standard maximum likelihood techniques to obtain consistent estimators of the difference in mean slopes and its variance. This suggests the importance of a concurrent control group when potential time-period effects are of concern.

Bias↗

PubMedDate not supplied

Multiple imputation in public health research.

Missing data in public health research is a major problem. Mean or median imputation is frequently used because it is easy to implement. Although multiple imputation has good statistical properties, it is not yet used extensively. For two real studies and a real study-based simulation, we compared the results after using multiple imputation against several simpler imputation methods. All imputation methods showed similar results for both real studies, but somewhat different results were obtained when only complete cases were used. The simulation showed large differences among various multiple imputation methods with a different number of variables for creating the matching metric for multiple imputation. Multiple imputation using only a few covariates in the matching model produced more biased coefficient estimates than using all available covariates in the matching model. The simulation also showed better standard deviation estimates for multiple imputation than for single mean imputation.

Asthma↗

PubMed2001

A multiple imputation method for missing covariates in non-linear mixed-effects models with application to HIV dynamics.

We propose a three-step multiple imputation method, implemented by Gibbs sampler, for estimating parameters in non-linear mixed-effects models with missing covariates. Estimates obtained by the proposed multiple imputation method are compared to those obtained by the mean-value imputation method and the complete-case method through simulations. We find that the proposed multiple imputation method offers smaller biases and smaller mean-squared errors for the estimates of covariate coefficients compared to other two methods. We apply the three missing data methods to modelling HIV viral dynamics from an AIDS clinical trial. We believe that the results from the proposed multiple imputation method are more reliable than that from the other two commonly used methods.

Antiretroviral Therapy, Highly Active↗

PubMed1997

A reappraisal of early hominid phylogeny.

We report here on the results of a new cladistic analysis of early hominid relationships. Ingroup taxa included Australopithecus afarensis, Australopithecus africanus, Australopithecus aethiopicus, Australopithecus robustus, Australopithecus boisei, Homo habilis, Homo rudolfensis, Homo ergaster and Homo sapiens. Outgroup taxa included Pan troglodytes and Gorilla gorilla. Sixty craniodental characters were selected for analysis. These were drawn from the trait lists of other studies and our own observations. Eight parsimony analyses were performed that differed with respect to the number of characters examined and the manner in which the characters were treated. Seven employed ordered characters, and included analyses in which (1) taxa that were variable with respect to a character were coded as having an intermediate state, (2) characters with variable states in any taxon were excluded; (3) a variable taxon was coded as having the state exhibited by the majority of its hypodigm, (4) variable taxa were coded as missing data for that character, (5) some characters were considered irreversible, (6) masticatory characters were excluded, and (7) characters whose states were unknown in some taxa were excluded. In the final analysis, (8) all characters were unordered. All analyses were performed using PAUP 3.0s. Despite the fact that the eight analyses differed with respect to methodology, they produced several consistent results. All agreed that the "robust" australopithecines form a clade, A. afarensis is the sister taxon of all other hominids, and the genus Australopithecus, as conventionally defined, is paraphyletic. All eight also supported trees in which A. africanus is the sister taxon of a joint Homo+ "robust" clade, although in one analysis an equally parsimonious topology found A. africanus to be the sister of the "robust" species. In most analyses, the relationships of A. africanus and H. habilis were unstable, in the sense that their positions vary in trees that are marginally less parsimonious than the favored one. Trees in which "robust" australopithecines are paraphyletic were found to be extremely unparsimonious.

Animals↗

PubMed1998

What does palaeontology contribute to systematics in a molecular world?

Palaeontology provides the only direct record for morphological and genetic change through time and uniquely contributes to systematics in two ways: by providing access to denser taxon sampling than is otherwise possible and by dating divergence times. Claims that ancient DNA has survived millions of years in certain fossils suggested the possibility that palaeontology could contribute directly to molecular systematic studies. Unfortunately, none of the supposed geologically ancient DNA records stands up to detailed scrutiny and fossils therefore contribute primarily through the morphological information they preserve. Denser taxon sampling can improve the accuracy of phylogenetic estimates primarily through allowing better discrimination of homoplasy from homology. This in turn leads to more accurate hypotheses of character transformation. Denser taxon sampling also offers the opportunity for more accurate rooting, since more characters can be polarized by reference to a stem-group taxon than to an extant sister-group taxon. Missing data can be a problem for fossils, but is not crippling. Finally the temporal order of clade appearances in the fossil record can provide ancillary evidence for selecting a working phylogeny from among a number of equally most parsimonious cladograms.

Animals↗

PubMed1988

Finding motion parameters from spherical motion fields (or the advantages of having eyes in the back of your head).

A theory is developed for determining the motion of an observer given the motion field over a full 360 degree image sphere. The method is based on the fact that for an observer translating without rotation, the projected circular motion field about any equator can be divided into disjoint semicircles of clockwise and counterclockwise flow, and on the observation that the effects of rotation decouple around the three equators defining the three principal axes of rotation. Since the effect of rotation is geometrical, the three rotational parameters can be determined independently by searching, in each case, for a rotational value for which the derotated equatorial motion field can be partitioned into 180 degree arcs of clockwise and counterclockwise flow. The direction of translation is also obtained from this analysis. This search is two dimensional in the motion parameters, and can be performed relatively efficiently. Because information is correlated over large distances, the method can be considered a pattern recognition rather than a numerical algorithm. The algorithm is shown to be robust and relatively insensitive to noise and to missing data. Both theoretical and empirical studies of the error sensitivity are presented. The theoretical analysis shows that for white noise of bounded magnitude M, the expected errors is at worst linearly proportional to M. Empirical tests demonstrate negligible error for perturbations of up to 20% in the input, and errors of less than 20% for perturbations of up to 200%.

Humans↗

PubMed1995

Psychometric properties of a specific quality of life questionnaire in angina pectoris patients.

This paper describes the psychometric properties of the French version of the APQLQ (Angina Pectoris Quality of Life Questionnaire), that is to say, scaling assumptions, reliability and validity. This disease-specific Swedish questionnaire provides a global score as well as four subscores measuring physical activities, somatic symptoms, emotional distress, and life satisfaction. The French version was produced according to the forward-backward translation methodology. In a cross-sectional study, 183 coronary patients filled out the APQLQ and the MOS (Medical Outcomes Study) SF-36; 170 sent it back; the rate of missing data was low (1.3%). The factorial structure and the high level of inter-scale correlations (> 0.60) suggested that the APQLQ measured one global concept rather than separate domains. The multitrait analysis identified one problematic item correlated strongly with all domains. The internal consistency was good (alpha Cronbach > 0.70). The correlations with the SF-36 scales were consistent with what was expected. The distribution of the scores of the APQLQ according to the clinical severity of Angina Pectoris (AP) was as hypothesized: the more severe the AP, the more impaired the Quality of Life. The score significantly discriminated between symptomatic (n = 110) and asymptomatic patients (n = 60) except for the emotional distress scale (p = 0.14). We recommend to analyze the French APQLQ as an index rather than as a profile. Its reliability, concurrent and clinical validity allowed its use in clinical trials.

Aged↗

PubMed2001

[Does the structure of medical services affect the proportion of death in hospitals? An ecological analysis in the mobility regions of Switzerland?].

OBJECTIVES: With the discussion about measures to reduce over-capacities in the health services in mind, we examined the influence of medical supply structures and services on hospital utilisation directly before death. METHODS: Based on the specification of the place of dying from the death certificates, we determined the proportions of deaths in hospitals in an ecological study. We analysed deaths of persons above age 65 in each of the 106 mobility regions of Switzerland in the years 1979 to 1980. RESULTS: The proportion of deaths occurring in hospitals varied between 27% and 81%. Despite missing data regarding stationary and ambulatory care services, more than half of the variance could be explained by means of a multivariate regression analysis. Our results imply an inverse relation between the proportion of deaths in the hospital and the number of consultations provided by primary care physicians, as well as the number of nursing home beds. Further, we observe a direct relation to the number of hospital beds in a region. All indicators are calculated per inhabitant. CONCLUSIONS: In health care supply planning, such systematic associations should be taken into account. We recommend to analyse regularly interregional differences in supply and outcome of medical performances with the best available data.

Adolescent↗

PubMed1989

[General insecurity, "false" and "real" security: arguments for the routine study of hospital patients for anti-HIV].

HIV infection has become an important risk for medical personal. Use of sufficient preventive measures with all patients and patient materials is time consuming, expensive and impracticable. In the present epidemiological situation anti-HIV testing of all hospital patients will give correct information on the presence or absence of HIV infection for 999 of 1000 patients. In addition to the increased safety of medical personal, testing will have several other advantages, some for the HIV infected persons: (1) In patients with HIV-related diseases time consuming delays in diagnosing the disease can be avoided. An earlier start of therapy will improve prognosis. (2) In HIV-infected persons a therapy can be chosen which exerts the least stress to the immune system. This may delay manifestation or progression of HIV-related diseases. (3) By medical surveillance of HIV-infected persons prophylaxis and immediate therapy of opportunistic infections is possible, which may prolong survival time. (4) Counselling of persons previously unaware of their HIV infection will slow down spread of HIV in the population. (5) Missing data on prevalence and incidence of HIV infection in different geographical areas will be available. Apart from the considerable cost reduction by reducing the use of extensive preventive measures to less than 10% of patients, routine screening of all patients is already economic if testing of 10,000 patients will prevent one single new infection by counselling of people with previously unknown HIV infection.

AIDS Serodiagnosis↗

PubMed1987

Oestrogen receptor activity in breast cancer detected at a prevalence screening examination.

In view of the possible introduction of screening programmes, this study compares oestrogen receptor (ER) levels in a series of women whose primary tumour was detected by screening and an age-matched consecutive series of women whose tumours were diagnosed after symptomatic presentation. Because of missing data and other statistical considerations, the comparison was made using T1 and T2 categories of tumour only. Some differences were found: the distribution of ER levels was significantly different in the two groups, with more extreme values in the symptomatic series; the screening series, however, had more moderate/rich ER levels than the symptomatic group. Tumours of special pathological type (for example, tubular, cribriform, lobular, medullary, and mucoid) were more likely to be ER-moderate or -rich, and there were more of these tumours in the screening series. The relationship of these findings to tumour growth rate is discussed. The study highlights the difficulty of obtaining sufficient tissue for conventional DCC biochemical assays from the small non-invasive tumours found by screening, and suggests that newer alternative methods employing monoclonal antibodies may be required for such types of tumour.

Aged↗

PubMedDate not supplied

[Results of physiological and psychological studies of different body types. A study of 210 patients with cardiovascular diseases and 100 students (author's transl)].

210 male patients with myocardial infarction, functional cardiovascular diseases, and angina pectoris hospitalized in the course of a cardiac rehabilitation program had been studied. Missing data in some anthropometric variables reduced the sample to 199 patients. The body of data included various anthropometric variables and indices, parameters of the circulatory and respiratory systems, biochemical measures, personality questionnaires, and information about the psychosocial background. In order to replicate the findings a sample of 100 students was studied, too. The measures and indices of body type are influenced by age, which leads to age-dependent spurious correlations between body type and other variables if this dependency is not corrected. Different statistical approaches proved the superiority of a factor-analytic typisation based on a "body-breadth" and "body-size" factor. There is a low correlation between body type and disposition for certain diseases showing that patients with "athletic" or "pycnic" body build suffer preferably from coronary infarction. Significant differences between body types exist for blood pressure, physical fitness, several psychosocial data, and the number of previous courses of treatment. Correlations of body type with personality dimensions and bodily complaints did not occur.

Age Factors↗

PubMed1988

Factors relating to age of onset in autism.

We examined the distribution of ages of onset of autism and related communication handicaps and assessed factors related to age of onset. Subjects were approximately 1,800 children seen at Division TEACCH (Treatment and Education of Autistic and related Communication handicapped CHildren) since 1970. Exact numbers of subjects varied with different analyses due to missing data. Data were gathered through direct assessment, interview, and questionnaire. Seventy-six percent of autistic children were identified by parents by 24 months of age, and 94% by 36 months. Families reporting early onset tended to seek help sooner and to be seen by TEACCH sooner. Early onset was most clearly related to severity as measured by IQ and ratings on the Childhood Autism Rating Scale (Schopler, Reichler, & Renner, 1986). The findings support the treatment of age of onset of autism by DSM-III-R (American Psychiatric Association, 1987).

Age Factors↗

PubMed1994

Beta blocker use in the treatment of community hospital patients discharged after myocardial infarction.

OBJECTIVE: To explore the reasons for underutilization of beta blocker treatment after acute myocardial infarction. DESIGN: A retrospective chart review. SETTING: Two large community hospitals in Milwaukee, Wisconsin. PATIENTS/PARTICIPANTS: All subjects (n = 694) discharged alive from July 1, 1990, to June 30, 1991, who had a diagnosis of acute myocardial infarction were eligible. Of these, 250 had missing data, resulting in a final sample of 444. RESULTS: Twenty-nine percent of the 444 patients were prescribed beta blocker therapy on discharge. Characteristics of the patients and their treatment associated with receipt of beta blocker therapy were identified with a logistic regression model. The adjusted odds ratios were 0.52 for female gender, 0.34 for no health insurance, 0.21 for chronic obstructive pulmonary disease, 0.46 for congestive heart failure, 0.28 for atrioventricular block, 1.86 for hypertension, 1.93 for chest pain during acute myocardial infarction, and 4.65 for prehospital beta blocker use. Prescription of beta blocker therapy was also influenced by receipt of other treatment modalities. The adjusted odds ratios were 0.23 for receipt of beta blocker therapy associated with myocardial revascularization, 0.18 for prescription on discharge of calcium channel blockers, and 0.22 for receipt of angiotensin-converting enzyme inhibitors. CONCLUSION: A minority of patients discharged after acute myocardial infarction receive beta blocker therapy, and women are only half as likely as men to receive it, after controlling for other factors. Though there are no data relating to whether calcium channel blockers or angiotensin-converting enzyme inhibitors lessen the protective effect of beta blocker therapy post-acute myocardial infarction, it would appear that these agents are frequently being used in lieu of beta blocker therapy for post-acute myocardial infarction patients.

Adrenergic beta-Antagonists↗