Finite mixture regression model with random effects: application to neonatal hospital length of stay

A two-component mixture regression model that allows simultaneously for heterogeneity and dependency among observations is proposed. By specifying random effects explicitly in the linear predictor of the mixture probability and the mixture components, parameter estimation is achieved by maximising the corresponding best linear unbiased prediction type log-likelihood. Approximate residual maximum likelihood estimates are obtained via an EM algorithm in the manner of generalised linear mixed model (GLMM). The method can be extended to a g-component mixture regression model with the component density from the exponential family, leading to the development of the class of finite mixture GLMM. For illustration, the method is applied to analyse neonatal length of stay (LOS). It is shown that identification of pertinent factors that influence hospital LOS can provide important information for health care planning and resource allocation. (C) 2002 Elsevier Science B.V. All rights reserved.

Polymorphisms in glutathione S-transferases and non-melanoma skin cancer risk in Australian renal transplant recipients

Caucasian renal transplant recipients from Queensland, Australia have the highest non-melanoma skin cancer (NMSC) risk worldwide. Although ultraviolet light (UVR) exposure is critical, genetic factors also appear important. We and others have shown that polymorphism in the glutathione S-transferases (GST) is associated with NMSC in UK recipients. However, the effect of high UVR exposure and differences in immunosuppressive regimen on these associations is unknown. In this study, we examined allelism in GSTM1, GSTM3, GSTT1 and GSTP1 in 361 Queensland renal transplant recipients. Data on squamous (SCC) and basal cell carcinoma (BCC), UVR/tobacco exposure and genotype were obtained. Associations with both NMSC risk and numbers were examined using logistic and negative binomial regression, respectively. In the total group, GSTM1 AB [P = 0.049, rate ratio (RR) = 0.23] and GSTM3 AA (P = 0.015, RR = 0.50) were associated with fewer SCC. Recipients were then stratified by prednisolone dose (less than or equal to7 versus >7 mg/day). In the low-dose group, GSTT1 null (P = 0.006, RR = 0.20) and GSTP1 Val/Val (P = 0.021, RR = 0.20) were associated with SCC numbers. In contrast, in the high-dose group, GSTM1 AB (P = 0.009, RR = 0.05), GSTM3 AB (P = 0.042, RR = 2.29) and BB (P = 0.014, RR = 5.31) and GSTP1 Val/Val (P = 0.036, RR = 2.98) were associated with SCC numbers. GSTM1 AB (P = 0.016) and GSTP1 Val/Val (P = 0.046) were also associated with fewer BCC in this group. GSTP1 associations were strongest in recipients with lower UVR/tobacco exposure. The data confirm our UK findings, suggesting that protection against UVR-induced oxidative stress is important in NMSC development in recipients, but that this effect depends on the immunosuppressant regimen.

Food intake and risk of squamous cell carcinoma of the skin in a community: The Nambour skin cancer cohort study

There is some evidence that dietary factors may modify the risk of squamous cell carcinoma (SCC) of the skin, but the association between food intake and SCC has not been evaluated prospectively. We examined the association between food intake and SCC incidence among 1,056 randomly selected adults living in an Australian sub-tropical community. Measurement-error corrected estimates of intake in 15 food groups were defined from a validated food frequency questionnaire in 1992. Associations with SCC risk were assessed using Poisson and negative binomial regression to the persons affected and tumour counts, respectively, based on incident, histologically confirmed tumours occurring between 1992 and 2002. After multivariable adjustment, none of the food groups was significantly associated with SCC risk. Stratified analysis in participants with a past history of skin cancer showed a decreased risk of SCC tumours for high intakes of green leafy vegetables (RR = 0.45, 95% CI = 0.22-0.91; p for trend = 0.02) and an increased risk for high intake of unmodified dairy products (RR = 2.53, 95% CI: 1.15-5.54; p for trend = 0.03). Food intake was not associated with SCC risk in persons who had no past history of skin cancer. These findings suggest that consumption of green leafy vegetables may help prevent development of subsequent SCCs of the skin among people with previous skin cancer and that consumption of unmodified dairy products, such as whole milk, cheese and yoghurt, may increase SCC risk in susceptible persons.

Improving precision and reducing bias in biological surveys: estimating false-negative error rates

The use of presence/absence data in wildlife management and biological surveys is widespread. There is a growing interest in quantifying the sources of error associated with these data. We show that false-negative errors (failure to record a species when in fact it is present) can have a significant impact on statistical estimation of habitat models using simulated data. Then we introduce an extension of logistic modeling, the zero-inflated binomial (ZIB) model that permits the estimation of the rate of false-negative errors and the correction of estimates of the probability of occurrence for false-negative errors by using repeated. visits to the same site. Our simulations show that even relatively low rates of false negatives bias statistical estimates of habitat effects. The method with three repeated visits eliminates the bias, but estimates are relatively imprecise. Six repeated visits improve precision of estimates to levels comparable to that achieved with conventional statistics in the absence of false-negative errors In general, when error rates are less than or equal to50% greater efficiency is gained by adding more sites, whereas when error rates are >50% it is better to increase the number of repeated visits. We highlight the flexibility of the method with three case studies, clearly demonstrating the effect of false-negative errors for a range of commonly used survey methods.

Transmission dynamics of the Echinococcus granulosus sheep-dog strain (G1 genotype) in camels in Tunisia

Cystic echinococcosis, caused by Echinococcus grantilosus, is highly endemic in North Africa and the Middle East. This paper examines the abundance and prevalence of infection of E. granulosus in camels in Tunisia. No cysts were found in 103 camels from Kebili, whilst 19 of 188 camels from Benguerden (10.1%) were infected. Of the cysts found 95% were considered fertile with the presence of protoscolices and 80% of protoscolices were considered viable by their ability to exclude aqueous eosin. Molecular techniques were used on cyst material from camels and this demonstrated that the study animals were infected with the G1 sheep strain of E. granulosus. Observed data were fitted to a mathematical model by maximum likelihood techniques to define the parameters and their confidence limits and the negative binomial distribution was used to define the error variance in the observed data. The infection pressure to camels was somewhat lower in comparison to sheep reported in an earlier study. However, because camels are much longer-lived animals, the results of the model fit suggested that older camels have a relatively high prevalence rate, reaching a most likely value of 32% at age 15 years. This could represent an important source of transmission to dogs and hence indirectly to man of this zonotic strain. In common with similar studies on other species, there was no evidence of parasite-induced immunity in camels. (C) 2004 Elsevier B.V. All rights reserved.

Determinants of exercise capacity in patients with type 2 diabetes

OBJECTIVE - Type 2 diabetes is associated with reduced exercise capacity, but the cause of this association is unclear. We sought the associations of impaired exercise capacity in type 2 diabetes. RESEARCH DESIGN AND METHODS - Subclinical left ventricular (LV) dysfunction was sought from myocardial strain rate and the basal segmental diastolic velocity (Em) of each wall in 170 patients with type 2 diabetes (aged 56 +/- 10 years, 91 men), good quality echocardiographic images, and negative exercise echocardiograms. The same measurements were made in 56 control subjects (aged 53 +/- 10 years, 29 men). Exercise capacity was calculated in metabolic equivalents, and heart rate recovery (HRR) was measured as the heart rate difference between peak and 1 min after exercise. In subjects with type 2 diabetes, exercise capacity was correlated with clinical, therapeutic, biochemical, and echocardiographic variables, and significant independent associations were sought using a multiple linear regression model. RESULTS - Exercise capacity, strain rate, Em, and HRR were significantly reduced in type 2 diabetes. Exercise capacity was associated with age (r- = -0.37, P < 0.001), male sex (r = 0.26, P = 0.001), BMI (r = -0.19, P = 0.012), HbA(1c) (AlC; r = -0.22, P = 0.009), Em (r = 0.43, P < 0.001), HRR (r = 0.42, P < 0.001), diabetes duration (r = -0.18, P = 0.021), and hypertension history (r = -0.28, P < 0.001). Age (P < 0.001), male sex (P = 0.007), BMI (P = 0.001), Em (P = 0.032), HRR (P = 0.013), and AlC (P = 0.0007) were independent predictors of exercise capacity. CONCLUSIONS - Reduced exercise capacity in patients with type 2 diabetes is associated with diabetes control, subclinical LV dysfunction, and impaired HRR.

Multi-level zero-inflated Poisson regression modelling of correlated count data with excess zeros

Count data with excess zeros relative to a Poisson distribution are common in many biomedical applications. A popular approach to the analysis of such data is to use a zero-inflated Poisson (ZIP) regression model. Often, because of the hierarchical Study design or the data collection procedure, zero-inflation and lack of independence may occur simultaneously, which tender the standard ZIP model inadequate. To account for the preponderance of zero counts and the inherent correlation of observations, a class of multi-level ZIP regression model with random effects is presented. Model fitting is facilitated using an expectation-maximization algorithm, whereas variance components are estimated via residual maximum likelihood estimating equations. A score test for zero-inflation is also presented. The multi-level ZIP model is then generalized to cope with a more complex correlation structure. Application to the analysis of correlated count data from a longitudinal infant feeding study illustrates the usefulness of the approach.

Optimal crossover designs for logistic regression models in pharmacodynamics

Pharmacodynamics (PD) is the study of the biochemical and physiological effects of drugs. The construction of optimal designs for dose-ranging trials with multiple periods is considered in this paper, where the outcome of the trial (the effect of the drug) is considered to be a binary response: the success or failure of a drug to bring about a particular change in the subject after a given amount of time. The carryover effect of each dose from one period to the next is assumed to be proportional to the direct effect. It is shown for a logistic regression model that the efficiency of optimal parallel (single-period) or crossover (two-period) design is substantially greater than a balanced design. The optimal designs are also shown to be robust to misspecification of the value of the parameters. Finally, the parallel and crossover designs are combined to provide the experimenter with greater flexibility.