SEARCH · PubMed Health

Results for “Logistic Models”

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 703 records · Page 39Linked to original sources

PubMed2003

Fixing the nonconvergence bug in logistic regression with SPLUS and SAS.

When analyzing clinical data with binary outcomes, the parameter estimates and consequently the odds ratio estimates of a logistic model sometimes do not converge to finite values. This phenomenon is due to special conditions in a data set and known as 'separation'. Statistical software packages for logistic regression using the maximum likelihood method cannot appropriately deal with this problem. A new procedure to solve the problem has been proposed by Heinze and Schemper (Stat. Med. 21 (2002) pp. 2409-3419). It has been shown that unlike the standard maximum likelihood method, this method always leads to finite parameter estimates. We developed a SAS macro and an SPLUS library to make this method available from within one of these widely used statistical software packages. Our programs are also capable of performing interval estimation based on profile penalized log likelihood (PPL) and of plotting the PPL function as was suggested by Heinze and Schemper (Stat. Med. 21 (2002) pp. 2409-3419).

Likelihood Functions↗

PubMed2002

Use of naloxone challenge to predict sexual performance of rams before the fall breeding season.

Serving capacity tests (SCT) identify rams as sexually active, sexually inactive, or low sexual performers. Labor and time requirements make SCT impractical and expensive. Therefore, a hormone-based test for libido was developed using the responses of LH and testosterone (T) after naloxone. This test effectively identified sexually active and inactive rams during the breeding season (November and December). The current study evaluated the effectiveness of this method at a different time of the year. Objectives were to determine if the blood test can detect differences in libido during late July and August, and to characterize LH and T profiles up to 120 min postnaloxone treatments at this time of year. Two experiments were conducted using i.v. naloxone at 0.75 mg/kg BW. Sexually active (SA) and sexually inactive (SIA) rams were identified using SCT. In Experiment 1, 16 SA and 18 SIA rams of several breeds were identified based on SCT conducted before naloxone tests. In Experiment 2, 72 Suffolk x white face rams were given SCT after naloxone tests and were of unknown sexual activity at the time of the naloxone tests. Mixed model analyses for repeated measures with repeated factors (month and sample time) and a covariate for mean LH or T before naloxone treatment were used for LH and T data. Proc Logistics modeled probabilities that rams were sexually active. The LH response to naloxone increased but did not differ (P > 0.09) by ram class (SA or SIA) or month. The T response peaked 75-90 min after naloxone treatment and was still elevated at 120 min. Testosterone differed (P < 0.03) for ram class by month by time in Experiment I and differed (P < 0.01) between months in Experiment 2. Proc Logistics incorrectly indicated that SA and SIA rams were all sexually active. This study indicates that the naloxone challenge test cannot discriminate between SA and SIA rams during July and August. Further research is required to determine the time of year when this sire identification test for libido has significant accuracy.

Animals↗

PubMed2001

Within-subject exposure dependency in case-crossover studies.

In the case-crossover design, only cases are sampled and the hazard ratio is estimated from within-subject comparisons of exposures at the event time and in M control periods prior to the event. We consider the effect of within-subject dependence of exposures in successive time periods. We show that estimates obtained from the conditional logistic model are biased. This bias disappears if the distribution of exposures in the M+1 successive time intervals is exchangeable. In contrast, the Mantel-Haenszel estimator for the odds ratio is approximately unbiased provided that exposures are stationary. Suitable methods of analysis of case-crossover designs using maximum likelihood may be derived from cohort rather than case-control models.

Bias↗

PubMed2004

Substance use and sexual risk: a participant- and episode-level analysis among a cohort of men who have sex with men.

Prior reports associating substance use with sexual risk behavior have generally used summary measures and have not adjusted for participants' background levels of substance use. In this 1999-2001 US study (the EXPLORE study), the authors determined whether substance use during sex was independently associated with sexual risk during recent sexual episodes, as reported by 4,295 human immunodeficiency virus-negative men who have sex with men. The main outcome measure was serodiscordant unprotected anal sex (SDUA). The influence of participant-level characteristics was examined by using repeated-measures logistic models. In assessing the influence of episode-level predictors on SDUA, the influence of participant-level characteristics, including 6-month substance use, was removed by using conditional logistic regression, in effect making each participant his own control. The authors also adjusted for partner characteristics. Eleven percent of participants reported heavy alcohol use, 37% used poppers, 19% sniffed cocaine, and 13% used amphetamines. In the participant-level analysis, use of poppers, amphetamines, and sniffed cocaine as well as heavy alcohol use in the prior 6 months were independently associated with SDUA. In the conditional analysis, consumption of > or = 6 alcoholic drinks or use of poppers, amphetamines, or sniffed cocaine just before or during sex was independently associated with SDUA. The authors concluded that programs aimed at preventing human immunodeficiency virus transmission should emphasize the influence of substance use during sex on increased risk behavior.

Adolescent↗

PubMed2003

A regressive model analysis of congenital sensorineural deafness in German Dalmatian dogs.

The objective of the present study was to analyze the mode of inheritance for congenital sensorineural deafness (CSD) in German Dalmatian dogs by consideration of association between phenotypic breed characteristics and CSD. Segregation analysis with regressive logistic models was employed to test for different mechanisms of genetic transmission. Data were obtained from all three Dalmatian kennel clubs associated with the German Association for Dog Breeding and Husbandry (VDH). CSD was tested by veterinary practitioners using standardized protocols for Brainstem Auditory-Evoked Response (BAER). The sample included 1899 Dalmatian dogs from 354 litters in 169 different kennels. BAER testing results were from the years 1986 to 1999. Pedigree information was available for up to seven generations. The segregation analysis showed that a mixed monogenic-polygenic model including eye color as covariate among all other tested models best explained the segregation of affected animals in the pedigrees. The recessive major gene segregated in dogs with blue and brown eye color as well as in dogs with and without pigmented coat patches. Models which took into account the occurrence of patches, percentage of puppies tested per litter, or inbreeding coefficient gave no better adjustment to the most general (saturated) model. A procedure for the simultaneous prediction of breeding values and the estimation of genotype probabilities for CSD is expected to improve breeding programs significantly.

Animals↗

PubMed1993

Clinical index to predict bacteraemia caused by staphylococci.

OBJECTIVES: To define risk factors associated with bacteraemia caused by Staphylococcus aureus or coagulase-negative staphylococci; and to use them to define patients in need of empiric anti-staphylococcal antibiotic treatment. DESIGN: Derivation set: observational, prospective study; validation set: retrospective analysis of a prospectively collected database. SETTING: Derivation set: Beilinson Medical Centre, Petah Tiqva, Israel--a 900-bed university hospital. Validation set: St Thomas's Hospital, London, UK--an 800-bed teaching hospital. SUBJECTS: All episodes of bacteraemia detected at Beilinson Medical Centre between March 1988 and September 1990 (derivation set, n = 1410), and at St Thomas's Hospital during 1987-1990 (validation set, n = 1040). INTERVENTIONS: None. MAIN OUTCOME MEASURES: Percentage of staphylococcal bacteraemia in groups of patients defined by the models. RESULTS: The following factors were associated with Staphylococcus aureus bacteraemia: focus of infection (whether high or low risk), haemodialysis, intravenous drug abuse and infection acquired in the orthopaedic ward. A logistic model was used to divide the derivation set into three groups with percentages of Staphylococcus aureus bacteraemia of 1.8%, 13.2% and 33.7% (P < 0.0001); and the validation group 2.5%, 18.2% and 53.2% (P < 0.0001). Factors associated with coagulase-negative staphylococcal bacteraemia were: central or peripheral intravenous catheter as the focus of infection, a preterm neonate, the presence of a central intravenous catheter, low temperature, and a low white blood cell count. A second model including those factors was used to divide the derivation set into three groups with percentages of coagulase-negative staphylococcal bacteraemia of 1.9%, 22.8%, and 43% (P < 0.0001). In the validation set, the percentages were 2.9%, 22.4% and 31.0% (P < 0.001). CONCLUSIONS: The present study defines groups at high risk for staphylococcal bloodstream infection, in which empiric treatment should include an anti-staphylococcal drug.

Adolescent↗

PubMed1998

Multi-level modelling of conception in artificial insemination by donor.

Data on insemination with donor's sperm have a crossed hierarchical structure due to the coexistence of female factors (ovulatory cycles within pregnancy attempt within women) and male factors (inseminations within donations within donors). A crossed random multi-level logistic model, taking account of this structure, permits an improved estimation of the fixed effects and provides insights into their influence at each level in the hierarchy. We present an efficient algorithm for fitting such models using alternating EM steps. We further discuss the inclusion of compositional covariates to determine what information the quality of a donation conveys regarding the donor basal fecundability and on the specific sperm donation.

Adult↗

PubMed2001

Logistic discrimination of mixtures of M. tuberculosis and non-specific tuberculin reactions.

Interpretation of the Mantoux test for tuberculous infection can be complicated by cross-reactions caused by infection with non-specific mycobacteria. Thus, the distribution of positive indurations is a mixture of two distributions. To estimate tuberculous infection prevalence, the marginal distribution of indurations needs to be separated into its component distributions. Observations from several populations with different mixes of the two types of infection are required. Homogeneity across populations of distributions of indurations for each type of infection is assumed. A logistic model is specified for the probability of having tuberculous infection conditional on the observed induration size. No other assumptions about the two distributions are made. Maximum likelihood is used to estimate the logistic function. Goodness-of-fit criteria are discussed. The method is applied to a series of tuberculin surveys carried out in (South) Korea. Estimated infection prevalence agrees reasonably well with several ad hoc criteria. The goodness-of-fit test rejects underlying assumptions of homogeneity. One reason appears to be a decline over time in induration sizes caused by tuberculous infection. However, not all reasons for this rejection are obvious. The proposed method of mixture analysis provides an additional tool for the interpretation of prevalence survey data where the diagnostic test lacks specificity as a result of cross-reactions.

Adolescent↗

PubMed2006

[Simplified prognostic model of overall intrahospital mortality of children in central Africa].

OBJECTIVES: To find a simple mortality prediction model based on nutritional and infection indicators for the assessment of the care of children admitted to hospital in central Africa. METHOD: Cohort study of 414 children admitted at Goma Hospital between 1.4.2003 and 31.3.2004. We conducted univariate analysis and logistic regression, computed adjusted odds ratios and constructed a prognostic score from the coefficients of logistic regression. The performance of logistic model and score were evaluated by the calculation of areas under the ROC curves. RESULTS: The intrahospital mortality rate reached 15.9%. In univariate analysis, age, WAZ, arm circumference, neurological status (Blantyre coma score), stiff neck, subcostal indrawning, and infection were significantly associated with mortality. Logistic regression model analysis and adjusted odds ratios (AOR) confirmed higher risks of death for young (AOR 3.4 (1.4-8.8) and underweight children (WAZ -2->-3 and WAZ < or = -3, AOR 3.2 (1.4-7.6) and AOR 4.4 (1.7-11.2)), for children with arm circumference under 115 mm (AOR 3.4 (1.5-7.3)), impaired consciousness (AOR 9.6 (3.1-29.9)) and bloodstream infections (AOR 6.6 (2.1-21.1)). The area under the ROC curve of the prognostic model is 0.83 (0.78-0.88), that of the prognostic score, 0.80 (0.75-0.86). CONCLUSION: This study provides a simple mortality prediction model for hospitalised children in central Africa, based on age, weight for age or arm circumference, neurological status (Blantyre coma score), and infection. This model and scoring system can be used to evaluate programs set up to reduce intrahospital mortality in this region.

Age Factors↗

PubMed2001

Pair approximation for lattice models with multiple interaction scales.

Pair approximation has frequently proved effective for deriving qualitative information about lattice-based stochastic spatial models for population, epidemic and evolutionary dynamics. Pair approximation is a moment closure method in which the mean-field description is supplemented by approximate equations for the frequencies of neighbor-site pairs of each possible type. A limitation of pair approximation relative to moment closure for continuous space models is that all modes of interaction between individuals (e.g., dispersal of offspring, competition, or disease transmission) are assumed to operate over a single spatial scale determined by the size of the interaction neighborhood. In this paper I present a multiscale pair approximation which allows different sized neighborhoods for each type of interaction. To illustrate and test the approximation I consider a spatial single-species logistic model in which offspring are dispersed across a birth neighborhood and established individuals have a death rate depending on the population density in a competition neighborhood, with one of these neighborhoods nested inside the other. Analysis of the steady-state equations yields several qualitative predictions that are confirmed by simulations of the model, and numerical solutions of the dynamic equations provide a close approximation to the transient behavior of the stochastic model on a large lattice. The multiscale pair approximation thus provides a useful intermediate between the standard pair approximation for a single interaction neighborhood, and a complete set of moment equations for more spatially detailed models.

Animals↗

PubMedDate not supplied

A prospective, randomized trial of early versus late administration of a single dose of surfactant-TA.

Thirty-two neonates weighing 500-1500 g with documented surfactant deficiency and without evidence of severe birth asphyxia, infection, prolonged rupture of membranes greater than or equal to 72 h, or oligohydramnios were randomly assigned to receive a single intratracheal dose of surfactant-TA (100 mg/kg) either within 30 min of birth (n = 16, early group) or at 6 h of age (n = 16, late group). By 6 h of age, all neonates of the late group had moderate/severe RDS, while none of the neonates of the early group had either clinical or radiological respiratory distress syndrome. The incidence of bronchopulmonary dysplasia was significantly lower in survivors of the early group than those of the late group (1/15 versus 7/14, a 43% reduction with a 95% confidence interval of 14-72%, P = 0.025). These beneficial effects of early surfactant treatment remained after controlling for the various confounding factors in the logistic models.

Bronchopulmonary Dysplasia↗

PubMed2004

[Modification and its application of generalized Schumacher model].

Based on the concrete analysis on growth equations presented by others, a modified Schumacher growth equation was proposed as, which included Gompertz function, Schumacher equation and generalized Schumacher equation, and had stronger self-adaptability and practicality. The analytic character and adaptability of the modified Schumacher equation were analyzed. According to the genetic algorithms method, this model was used to fit the growth data of endangered plant of Tsuga longibracteata and Platycladus orientalis. The results showed that the modified Schumacher equation was not only better than Schumacher equation and generalized Schumacher equation significantly, but also better than classical Logistic model and Li's self-adaptive model. So it could be used to study the dynamics simulation for tree growth and dynamics law for population growth.

Logistic Models↗

PubMed1996

A statistical model predicting the seizure threshold for right unilateral ECT in 106 patients.

Titration of the electroconvulsive therapy (ECT) stimulus to the patient's convulsive threshold is the only way to directly assess the patient's seizure threshold. This technique is presently practiced by 39% of ECT providers, according to a recent survey. Because multiple variables influence the seizure threshold in patients, multivariate statistical methods may provide a useful strategy to determine which variables exert the most influence on convulsive threshold. A multivariate ordinal logistic model of seizure threshold was developed on an experimental group of 66 consecutive patients undergoing titrated right unilateral (RUL) ECT for major depression. The accuracy of the model was cross-validated on a second group of 40 patients undergoing similar RUL ECT procedures. The final multivariate ordinal logistic regression model for the seizure threshold level (STL) was significant (Likelihood ratio chi 2 = 54.115; p < 0.0001:R2 = 0.313). Increasing age, African-American race, and longer inion-nasion distances (p < 0.06) predicted higher STL. Female gender was associated with a lower STL. The ability of the final model to accurately predict STL for the validation group was fair (pairwise correlation was 0.576; p < 0.001). The model did well for predicting lower STL, but fared poorly for higher STL. In conclusion, modeling STL may help establish the relative contribution of variables thought to be important to seizure threshold. However, STL models remain impractical for clinical applications in estimating seizure threshold at this time, and empirical stimulus titration should be used.

Adult↗

PubMed1994

Estimating the effect of regression toward the mean under stochastic censoring.

This investigation estimated regression toward the mean in the evaluation of a blood cholesterol educational program in which the probability of return for a follow-up measure was positively related to the initial measure--a situation the authors term "stochastic censoring." The situation in which selection is by a fixed cutpoint is well known. The authors estimate the extent of regression toward the mean when selection is probabilistically related to an initially higher blood cholesterol level. A Monte Carlo method is proposed, conditional on the observed data, for estimating the regression toward the mean and its precision. In 106 simulations, regression toward the mean was estimated to be 0.012 mmol/liter (standard deviation (SD), 0.032 mmol/liter). Similar estimates were obtained using Empirical Bayes shrinkage estimates of baseline cholesterol (regression toward the mean = 0.0106 mmol/liter) and numeric integration (0.0109 mmol/liter). The observed reduction in blood cholesterol was 0.271 mmol/liter (SD, 0.061 mmol/liter); after correction for regression toward the mean, the estimate of the true educational effect was 0.259 mmol/liter (SD, 0.069 mmol/liter). Functions are presented that will enable investigators to predict regression toward the mean and its standard deviation under the conditions that errors are gaussian and stochastic censoring can be approximated by a logistic model.

Cholesterol↗

PubMed2004

Lifestyle behaviors associated with lower risk of having the metabolic syndrome.

The metabolic syndrome is a cluster of risk factors that predisposes individuals to cardiovascular disease (CVD) and diabetes and is present in almost one fourth of adult Americans. Risk factors involved with the metabolic syndrome can be altered via modifiable lifestyle factors, such as diet, physical activity, and smoking and drinking habits. The objective of this study was to examine the extent to which these modifiable lifestyle behaviors are associated with the risk of having the metabolic syndrome. Data from the Third National Health and Nutrition Examination Survey (NHANES III), conducted between 1988 and 1994, were used to measure the risk of having the metabolic syndrome in healthy adult Americans who follow certain lifestyle behaviors, such as dietary practices, levels of physical activity, smoking and drinking habits. Low physical activity level, high carbohydrate (CHO) intake, and current smoking habits were all significantly associated with an increased risk of having the metabolic syndrome, even after adjusting for other related covariates. Relative to physically inactive subjects, being physically active was associated with lower odds ratio (OR) (0.36, confidence interval [CI] 0.21 to 0.68, P < .01) in overweight men and in normal weight (0.36, CI 0.18 to 0.70, P < .01) and overweight (0.61, CI 0.38 to 0.97, P < .05) women. Although the type of CHO could not be distinguished, relative to a high CHO diet, men having a low or moderate CHO intake had a lower risk of having the metabolic syndrome with respective ORs of 0.41 (CI 0.24 to 0.67, P < .01) and 0.44 (CI 0.25 to 0.77, P < .01); no effect of dietary CHO was observed in women. Moderate alcohol consumption was not significantly related to the risk of having the metabolic syndrome in men, but was associated with a lower OR in women (0.76, CI 0.61 to 0.95, P < .05). Regression models indicate a reduced risk of having the metabolic syndrome when selected low-risk lifestyle factors are present in combination, particularly in subjects with body mass index (BMI) < 30 kg/m(2). According to our cross-sectional logistic models, the risk of having the metabolic syndrome is substantially lower in individuals who are physically active, nonsmoking, have a relatively low CHO intake and moderate alcohol consumption, and who maintain a BMI in the non-obese range. These observations have potentially important value for public health recommendations.

Adult↗

PubMedDate not supplied

Applying sample survey methods to clinical trials data.

This paper outlines the utility of statistical methods for sample surveys in analysing clinical trials data. Sample survey statisticians face a variety of complex data analysis issues deriving from the use of multi-stage probability sampling from finite populations. One such issue is that of clustering of observations at the various stages of sampling. Survey data analysis approaches developed to accommodate clustering in the sample design have more general application to clinical studies in which repeated measures structures are encountered. Situations where these methods are of interest include multi-visit studies where responses are observed at two or more time points for each patient, multi-period cross-over studies, and epidemiological studies for repeated occurrences of adverse events or illnesses. We describe statistical procedures for fitting multiple regression models to sample survey data that are more effective for repeated measures studies with complicated data structures than the more traditional approaches of multivariate repeated measures analysis. In this setting, one can specify a primary sampling unit within which repeated measures have intraclass correlation. This intraclass correlation is taken into account by sample survey regression methods through robust estimates of the standard errors of the regression coefficients. Regression estimates are obtained from model fitting estimation equations which ignore the correlation structure of the data (that is, computing procedures which assume that all observational units are independent or are from simple random samples). The analytic approach is straightforward to apply with logistic models for dichotomous data, proportional odds models for ordinal data, and linear models for continuously scaled data, and results are interpretable in terms of population average parameters. Through the features summarized here, the sample survey regression methods have many similarities to the broader family of methods based on generalized estimating equations (GEE). Sample survey methods for the analysis of time-to-event data have more recently been developed and implemented in the context of finite probability sampling. Given the importance of survival endpoints in late phase studies for drug development, these methods have clear utility in the area of clinical trials data analysis. A brief overview of methods for sample survey data analysis is first provided, followed by motivation for applying these methods to clinical trials data. Examples drawn from three clinical studies are provided to illustrate survey methods for logistic regression, proportional odds regression and proportional hazards regression. Potential problems with the proposed methods and ways of addressing them are discussed.

Clinical Trials as Topic↗

PubMed2005

Predicting severity of cochlear hair cell damage in adult chickens using DPOAE input-output functions.

Distortion product otoacoustic emissions (DPOAE) were recorded from the ear canal of aged broiler chickens which have been shown to present with age-related cochlear degeneration [Hear. Res. 166 (2002) 82]. We describe the relationship between the shape of the DPOAE input-output (I/O) function and the type of hair cell damage present at and between the cochlear frequency places of the DPOAE primary tones (f1 and f2). The mid stimulus level compressive growth of the mean DPOAE I/O functions is reduced in a graded fashion relative to the severity of hair cell damage. However, individual DPOAE I/O functions within most hair cell damage groups show large variability from this characteristic. Various least squares regression models were used to predict hair cell density from indices derived from the DPOAE I/O function (area, threshold and slope). The results showed that no simple linear relationship exists between hair cell density and the DPOAE I/O function indices. Multivariate binary logistic regression used DPOAE I/O function indices to predict membership in hair cell damage groups. The logistic model revealed that DPOAE threshold can be used to predict the occurrence of severe/total hair cell damage with good specificity though poor sensitivity.

Aging↗

PubMed1994

The prediction of coronary heart disease mortality as a function of major risk factors in over 30 000 men in the Italian RIFLE pooling Project. A comparison with the MRFIT primary screenees. The RIFLE research group.

BACKGROUND: Few risk functions for the prediction of coronary heart disease mortality have been produced in Italy. This study used a large population sample to evaluate the effect of major risk factors on coronary mortality. METHODS: Coronary deaths in 45 cohorts of men (n = 31317, aged 30-69 years) were studied and related to selected cardiovascular risk factors. RESULTS: After 6 years, 1089 men had died, of whom 239 were coronary fatalities. Univariate and multivariate (Cox model) analyses conducted on each age group (30-39, 40-49, 50-59, and 60-69 years) showed a positive association between coronary deaths and systolic blood pressure, serum cholesterol level and cigarette smoking, with few exceptions. A multiple logistic model was produced for men aged 35-57 years, assessing the role of age, serum cholesterol, cigarettes smoked per day and diastolic instead of systolic blood pressure, using the same endpoint as that employed in a similar model published from the analysis of MRFIT primary screenees in the USA to facilitate valid comparison. The coefficients in the present study were similar to those in the US cohort: no statistically significant differences could be detected when comparing the pairs of coefficients. CONCLUSION: Coefficients relating cholesterol, blood pressure and cigarette smoking to coronary mortality in Italian men are similar to those in American men from the same age groups.

Adult↗