Biomedical subjects

K Y Liang

Publications and source records attributed to K Y Liang.

At least 19 recordsLinked to original sources

PubMed1999

Body weight patterns from 20 to 49 years of age and subsequent risk for diabetes mellitus: the Johns Hopkins Precursors Study.

BACKGROUND: Obesity in middle age is a well-known risk factor for the development of type 2 diabetes mellitus. However, the importance of weight and weight gain at younger ages is less certain. OBJECTIVE: To determine the relationship of body weight patterns from 20 to 49 years of age with the subsequent risk for type 2 diabetes mellitus. SETTING: An ongoing longitudinal study of former medical students. PARTICIPANTS: Nine hundred sixteen white men without diabetes at 50 years of age. MEASUREMENTS: Weight and height measured in medical school, then assessed by mailed questionnaire to 49 years of age. MAIN OUTCOME: Incident type 2 diabetes mellitus based on physician self-report. RESULTS: During 14 255 person-years of follow-up, there were 35 incident cases of type 2 diabetes mellitus (2.5 per 1000 person-years). After simultaneous adjustment for age, physical activity, lifetime maternal history of diabetes, and smoking, body mass indexes (BMIs; calculated as weight in kilograms divided by the square of height in meters) at 25, 35, and 45 years of age were all strongly associated with diabetes risk (relative risks for overweight [BMI> or =25.0] vs. not overweight, >3.0; all Ps<.05), as were maximum and average BMI to 49 years of age. The relationship of BMI at 25 years of age to diabetes risk was substantially attenuated by adjustment for BMI at 45 years of age and average BMI, but was independent of weight change, weight variability, or maximum BMI. CONCLUSION: In men, overweight at 25 years of age strongly predicts diabetes risk in middle age, largely through its association with overweight at 45 years of age and high average BMI to 49 years of age.

Adult↗

PubMed1999

Cannabis use and cognitive decline in persons under 65 years of age.

The purpose of this study was to investigate possible adverse effects of cannabis use on cognitive decline after 12 years in persons under age 65 years. This was a follow-up study of a probability sample of the adult household residents of East Baltimore. The analyses included 1,318 participants in the Baltimore, Maryland, portion of the Epidemiologic Catchment Area study who completed the Mini-Mental State Examination (MMSE) during three study waves in 1981, 1982, and 1993-1996. Individual MMSE score differences between waves 2 and 3 were calculated for each study participant. After 12 years, study participants' scores declined a mean of 1.20 points on the MMSE (standard deviation 1.90), with 66% having scores that declined by at least one point. Significant numbers of scores declined by three points or more (15% of participants in the 18-29 age group). There were no significant differences in cognitive decline between heavy users, light users, and nonusers of cannabis. There were also no male-female differences in cognitive decline in relation to cannabis use. The authors conclude that over long time periods, in persons under age 65 years, cognitive decline occurs in all age groups. This decline is closely associated with aging and educational level but does not appear to be associated with cannabis use.

Adult↗

PubMed1999

Back injury in municipal workers: a case-control study.

OBJECTIVES: The purpose of this study was to identify factors associated with acute low back injury among municipal employees of a large city. METHODS: For each of 200 injured case patients, 2 coworker controls were randomly selected, the first matched on gender, job, and department and the second matched on gender and job classification. In-person interviews were conducted to collect data on demographics, work history, work characteristics, work injuries, back pain, psychosocial and work organization, health behaviors, and anthropometric and ergonomic factors related to the job. Psychosocial work organization variables were examined with factor analysis techniques; an aggregate value for job strain was entered into the final model. Risk factors were examined via multivariate logistic regression techniques. RESULTS: High job strain was the most important factor affecting back injury (odds ratio [OR] = 2.12, 95% confidence interval [CI] = 1.28, 3.52), and it showed a significant dose-response effect. Body mass index (OR = 1.54, 95% CI = 1.08, 2.18) and a work movement index (twisting, extended reaching, and stooping) (OR = 1.42, 95% CI = 0.97, 2.08) were also significant factors. CONCLUSIONS: Results suggest that increasing workers' control over their jobs reduces levels of job strain. Ergonomic strategies and worksite health promotion may help reduce other risk factors.

Accidents, Occupational↗

PubMed1998

Inference for odds ratio regression models with sparse dependent data.

Suppose the number of 2 x 2 tables is large relative to the average table size, and the observations within a given table are dependent, as occurs in longitudinal or family-based case-control studies. We consider fitting regression models to the odds ratios using table-level covariates. The focus is on methods to obtain valid inferences for the regression parameters beta when the dependence structure is unknown. In this setting, Liang (1985, Biometrika 72, 678-682) has shown that inference based on the noncentral hypergeometric likelihood is sensitive to misspecification of the dependence structure. In contrast, estimating functions based on the Mantel-Haenszel method yield consistent estimators of beta. We show here that, under the estimating function approach, Wald's confidence interval for beta performs well in multiplicative regression models but unfortunately has poor coverage probabilities when an additive regression model is adopted. As an alternative to Wald inference, we present a Mantel-Haenszel quasi-likelihood function based on integrating the Mantel-Haenszel estimating function. A simulation study demonstrates that, in medium-sized samples, the Mantel-Haenszel quasi-likelihood approach yields better inferences than other methods under an additive regression model and inferences comparable to Wald's method under a multiplicative model. We illustrate the use of this quasi-likelihood method in a study of the familial risk of schizophrenia.

Biometry↗

PubMed1997

Application of transmission disequilibrium tests to nonsyndromic oral clefts: including candidate genes and environmental exposures in the models.

Extensive epidemiological and genetic studies of the cause of oral clefts have demonstrated strong familial aggregation but have failed to yield definitive evidence of any single genetic mechanism. We used the transmission/disequilibrium test (TDT) to investigate the relationship between oral clefts and markers associated with five candidate genes by utilizing 160 parent-offspring trios. Conditional logistic regression models extended the TDT to include covariates as effect modifiers, thus permitting tests for gene-environment interactions. For four of these candidates [transforming growth factor alpha (TGFA), transforming growth factor beta 3 (TGFB3), retinoic acid receptor (RARA), and the proto-oncogene BCL3], we detected modestly elevated odds ratios for the transmission of one marker allele to cleft probands when all the trios were analyzed together. These odds ratios increased when information on type of cleft, race, family history, or maternal smoking were incorporated as effect modifiers. We detected significant interaction between maternal smoking and the transmission of alleles for markers near TGFA and TGFB3; excess transmission of allele 3 at BCL3 was most significant among cleft lip probands; and the odds ratios for transmission of alleles at D19S178 and THRA1 were significant when ethnic group was included in the model. We suggest that utilizing an analytical strategy that allows for stratification of data and incorporating environmental effects into a single analysis may be more effective for detecting genes of small effect.

B-Cell Lymphoma 3 Protein↗

PubMed1997

Analyzing sibship correlations in birth weight using large sibships from Norway.

Data from the Medical Birth Registry of Norway were used to estimate sibship correlations in large sibships (each with > or = 5 infants among singleton live births surviving the first year of life), while adjusting for covariates such as infant gender, gestational age, maternal age, parity, and time since last pregnancy. This sample of 12,356 full sibs in 2,462 sibships born in Norway between 1968 and 1989 was selected to maximize the information on parity, and a robust approach to estimating both regression coefficients and the sibship correlation using generalized estimating equations (GEE) was employed. In concordance with previous studies, these data showed a high overall correlation in birth weight among full sibs (0.48 +/- 0.01), but this sibship correlation was influenced by parity. In particular, the correlation between the firstborn infant and a subsequent infant was slightly lower than between two subsequent sibs (0.44 +/- 0.01 vs. 0.50 +/- 0.01, respectively). The effect of time between pregnancies was statistically significant, but its predicted impact was modest over the period in which most of these large families were completed. While these data cannot discriminate whether factors influencing birth weight are maternal or fetal in nature, this analysis does illustrate how robust statistical models can be used to estimate sibship correlations while adjusting for covariates in family studies.

Birth Certificates↗

PubMed1997

Sample size calculations for studies with correlated observations.

Correlated data occur frequently in biomedical research. Examples include longitudinal studies, family studies, and ophthalmologic studies. In this paper, we present a method to compute sample sizes and statistical powers for studies involving correlated observations. This is a multivariate extension of the work by Self and Mauritsen (1988, Biometrics 44, 79-86), who derived a sample size and power formula for generalized linear models based on the score statistic. For correlated data, we appeal to a statistic based on the generalized estimating equation method (Liang and Zeger, 1986, Biometrika 73, 13-22). We highlight the additional assumptions needed to deal with correlated data. Some special cases that are commonly seen in practice are discussed, followed by simulation studies.

Biometry↗

PubMed1996

Comparison of methods for survival analysis of dependent data.

Analysis of dependent survival data by conventional partial likelihood methods produces unbiased estimates of the regression coefficients but incorrectly estimates their variance. Here we compared the conventional partial likelihood methods with two alternative methods for analyzing dependent survival data. The first alternative method estimated the regression coefficient by the partial likelihood approach but adjusted the variance to account for clustering. The second alternative method used marginal likelihoods to estimate both the regression coefficient and its variance. We evaluated the performance of the three methods using simulated and actual data. Simulated data were used to examine bias, efficiency, type I errors, and power. An Old Order Amish genealogy was analyzed under these models to illustrate their performance on real data. The simulation study showed that all three methods provided unbiased estimates of the regression coefficient, but the efficiency of the estimated regression coefficient varied according to the simulation conditions. The standard partial likelihood method showed increasing type I error as the dependence increased within clusters. Both alternative methods had acceptable levels of type I errors at all dependence levels. In the analysis of genealogic data, the regression coefficient was similar in the three methods showing stable estimates of the regression coefficients. The variance estimates from the alternative methods were slightly different from the conventional method, suggesting a flow level of dependence. This study displays the effect of violating the independence assumption and provides guidelines for using alternative statistical methods.

Analysis of Variance↗

PubMed1996

Analysis of case-control/family sampling design.

A commonly adopted design in genetic epidemiologic studies is the so-called case-control/family sampling design. Here, cases and controls are sampled and response variables, either quantitative or qualitative, for relatives of cases are contrasted with those of control relatives. This design can be used to examine familial aggregation, contribute to identification of genetic subtypes, and test the discrete versus continuous spectrum hypothesis for disorders of unknown etiology. However, the statistical independence assumption required by conventional case-control studies is violated for observations from related individuals who share the same genetic/environmental conditions. Consequently, ignoring dependence among related subjects will lead to incorrect sample size calculations and potentially erratic scientific conclusions. In this paper, we discuss several statistical issues that are relevant to the case-control/family sampling design with a focus on the use of this design in psychiatric research. Specifically, we 1) discuss the relative merit of matched versus unmatched designs; 2) present statistical methods that are useful for analyzing family data and 3) present sample size formulas for studies of quantitative and qualitative traits. A genetic epidemiologic study of schizophrenia is used for illustrative purposes.

Adolescent↗

PubMed1996

Determining linkage and mode of inheritance: mode scores and other methods.

The lod score method remains a popular approach for detecting linkage and estimating the recombination fraction theta between a marker locus and a trait locus. However, its implementation requires knowledge of all parameters of the genetic mechanism, including the number of loci involved and the genotype specific penetrance, which could depend on factors such as age. When some of the penetrance parameters phi are unknown, several methods are available, and have been reviewed by Hodge and Elston [(1994) Genet Epidemiol 11:329-342]. These include the "wrod score" (lod score maximized over theta under a wrong value of phi) and "mod score" (lod score maximized over both theta and phi) methods for inference on theta. It has further been proposed that the mod score also be used for estimating phi. In this paper, we review and assess the adequacy of these two methods for inferences on both phi and theta. In particular, all of the methods can be seen as variations on likelihood inference, using the information in the conditional likelihood for the marker data, given the trait data. The loss of efficiency of the mod is compared to that of the full likelihood, which utilizes all information available in the trait data. We also propose an alternative, based on the pseudo-likelihood, where phi is estimated via the trait information and plugged into the conditional likelihood. This method is compared to the mod score method, and the advantages and disadvantages of each are elucidated. In particular, it is seen that the pseudo-likelihood method can be more efficient than the mod score method if the ascertainment scheme can be modeled. As examples, both a random sample and a multiplex ascertainment scheme are considered. In addition, the pseudo-likelihood method leads to likelihood ratio tests for detecting linkage with a simple, known asymptotic reference distribution, a feature not shared by the mod score. Finally, we discuss the advantages of using the pseudo-likelihood method over the full likelihood method, both of which are valid when the ascertainment scheme is known.

Genetic Linkage↗

PubMed1996

Quasilikelihood estimation in measurement error models with correlated replicates.

We consider quasilikelihood models when some of the predictors are measured with error. In many cases, the true but fallible predictor is impossible to measure, and the best one can do is to obtain replicates of the fallible predictor. We consider the case that the replicates are not independent. If one assumes that replicates are independent and they are not, one typically underestimates the extent of the measurement error, leading to an inconsistent errors in variables correction. We devise techniques for estimating the measurement error covariance matrix. In addition, we discuss how one might perform a quasilikelihood analysis by computing the mean and variance functions of the observed data, both using approximations and also exactly through a Monte Carlo method. The methods are illustrated on a data set involving systolic blood pressure and urinary sodium chloride, where the measurement errors appear to be approximately normally distributed but highly correlated, and the distribution of the true predictor is reasonably modeled as a mixture of normals.

Analysis of Variance↗

PubMed1995

Structure and course of positive and negative symptoms in schizophrenia.

BACKGROUND: This article focuses on the underlying structure and temporal course of the symptoms of schizophrenia. METHODS: Ratings of symptoms in 90 schizophrenic patients were made each month for 10 years following the first hospitalization. The analytic methods consisted of cross-tabulation, dichotomous factor analysis, and bivariate dichotomous time series. RESULTS: The factor analyses revealed positive and negative factors with a slight tendency to merge over time. The prevalence of positive and negative symptoms declined in the year following first hospitalization and was stable thereafter. Positive and negative symptoms in 1 month were highly predictive of the same type of symptoms in the next month. Neither type of symptom was strongly associated with the other type in the following month when both types were included in the model. The predictability of the process increased with time. CONCLUSIONS: With a few minor caveats, the results suggest that the positive and negative symptom clusters are independent, both cross-sectionally and longitudinally.

Adult↗

PubMed1995

Some recent developments for regression analysis of multivariate failure time data.

Cox's seminal 1972 paper on regression methods for possibly censored failure time data popularized the use of time to an event as a primary response in prospective studies. But one key assumption of this and other regression methods is that observations are independent of one another. In many problems, failure times are clustered into small groups where outcomes within a group are correlated. Examples include failure times for two eyes from one person or for members of the same family. This paper presents a survey of models for multivariate failure time data. Two distinct classes of models are considered: frailty and marginal models. In a frailty model, the correlation is assumed to derive from latent variables ("frailties") common to observations from the same cluster. Regression models are formulated for the conditional failure time distribution given the frailties. Alternatively, marginal models describe the marginal failure time distribution of each response while separately modelling the association among responses from the same cluster. We focus on recent extensions of the proportional hazards model for multivariate failure time data. Model formulation, parameter interpretation and estimation procedures are considered.

Clinical Trials as Topic↗

PubMed1995

Sudden infant death syndrome in relation to weather and optimetrically measured air pollution in Taiwan.

OBJECTIVE: To examine the possible role of weather and air pollution in sudden infant death syndrome (SIDS) and suffocation. METHODS: Poisson regression analysis was carried out to measure the association between daily rates of SIDS per 1000 live births and daily average values of visibility and temperature in Taiwan between 1981 and 1991. The optimetrical measure of air pollution was used to represent pollution over a whole area rather than at a point source. RESULTS: Mortality from SIDS per 1000 live births was 3.3 times greater in the lowest category of visibility on the day of death than in the highest category; this rate ratio was 3.4 for the average visibility during the 9 days before death. Adjustment for population size, season, level of urbanization, incidence of deaths from respiratory tract infections, temperature, air pressure, sunshine, rainfall, relative humidity, and windspeed increased these rate ratios to 3.8 and 5.1, respectively. This suggests that the relationship between air pollution and SIDS is not biased by ecological confounders. For temperature the rate ratios were between 3.3 and 4.0. CONCLUSIONS: Our findings confirm the association of climatic temperature and air pollution with SIDS.

Air Pollution↗

PubMed1994

Minimum sample size estimation to detect gene-environment interaction in case-control designs.

As genetic markers become more available, case-control studies will be increasingly important in defining the role of genetic factors in disease causality. The authors estimate the minimum sample size needed to assure adequate statistical power to detect gene-environment interaction. One assumption is made: the prevalence of exposure is independent of marker genotypes among controls. Given the assumption, six parameters (three odds ratios, the prevalence of exposure, the proportion of those with the susceptible genotype, and the ratio of controls to cases) dictate the expected cell sizes in a 2 x 2 x 2 table contrasting genetic susceptibility, exposure, and disease. The three odds ratios reflect the association between disease and 1) exposure among non-susceptibles; 2) susceptible genotypes among nonexposed individuals; and 3) the gene-environment interaction itself, respectively. Given these parameters, the number of cases and controls needed to assure any particular Type I and Type II error rates can be estimated. Results presented here demonstrate that case-control designs can be used to detect gene-environment interaction when there is both a common exposure and a highly polymorphic marker of susceptibility.

Case-Control Studies↗

PubMed1994

Extended latent class approach to the study of familial/sporadic forms of a disease: its application to the study of the heterogeneity of schizophrenia.

When no method exists for detecting genetic forms of a disorder, epidemiologists classify probands according to the presence or absence of an affected relative (familial or sporadic). Not only is this a surrogate measure but if the risk for the disorder is associated with characteristics such as age and gender, then probands with varied distributions of these characteristics among their relatives are subject to misclassification. A latent class approach is presented which explicitly models the relationship between the affected status of the relatives and the unobservable familial/sporadic status of the proband in order to adjust for these characteristics. Lastly, an approach is introduced to correct for attenuation in measures of association between familial/sporadic status and other variables that could result if probands are misclassified. This approach incorporates the latent class probabilities directly into the regression model without classifying probands. These methods are applied to a study of the heterogeneity of schizophrenia.

Adolescent↗

PubMed1994

Temperament as a potential predictor of mortality: evidence from a 41-year prospective study.

Psychological factors were hypothesized to influence mortality, in particular, early versus later mortality. To explore the relationship between temperament, a psychological factor, and mortality in a prospective study of 1337 medical students, we constructed a measure portraying three temperament types, using latent class analysis. Death occurred in 113 subjects over 25-41 years of follow-up. In univariate survival analysis, subjects tending to direct tension "inward" when under stress ("Tension-In") had a higher risk of mortality than "Tension-Out" or "Stable" types. These associations persisted after adjustment for age, smoking, cholesterol level, and Quetelet Index. The relative risk (RR) of mortality for Tension-In was 1.56 (95% confidence interval, 1.00-2.44) compared with the Stable group. The risk was due entirely to the excess risk in persons under 55 years of age (RR, 2.59; 95% confidence interval, 1.46-4.62); the corresponding risk of death in older persons was 0.66 (0.30-1.48). Thus temperament is a significant risk factor for mortality, in particular, premature death.

Adult↗

PubMed1994

Coffee intake and coronary heart disease.

We examined the risk of coronary heart disease (CHD) associated with coffee intake in 1040 male medical students followed for 28 to 44 years. During the follow-up, CHD developed in 111 men. The relative risks (95% confidence interval) associated with drinking 5 cups of coffee/d were 2.94 (1.27, 6.81) for baseline, 5.52 (1.31, 23.18) for average, and 1.95 (0.86, 4.40) for most recent intake after adjustment for baseline age, serum cholesterol levels, calendar time, and the time-dependent covariates number of cigarettes, body mass index, and incident hypertension and diabetes. Risks were elevated in both smokers and nonsmokers and were stronger for myocardial infarction. Most of the excess risk was associated with coffee drinking prior to 1975. The diagnosis of hypertension was associated with a subsequent reduction in coffee intake. Negative results in some studies may be due to the assessment of coffee intake later in life or to differences in methods of coffee preparation between study populations or over calendar time.

Adult↗