Akaike's information criterion in generalized estimating equations

Bleeding in Patients Undergoing Percutaneous Coronary Intervention

Background— Bleeding in patients undergoing percutaneous coronary intervention (PCI) is associated with increased morbidity, mortality, length of hospitalization, and cost. We identified baseline clinical characteristics associated with bleeding complications after PCI and developed a simplified, clinically useful algorithm to predict patient risk.

Methods and Results— Data were analyzed from 302 152 PCI procedures performed at 440 US centers participating in the National Cardiovascular Data Registry. As defined by the National Cardiovascular Data Registry, bleeding required transfusion, prolonged hospital stay, and/or a drop in hemoglobin >3.0 g/dL from any location, including percutaneous entry site, retroperitoneal, gastrointestinal, genitourinary, and other/unknown location. Bleeding complications occurred in 2.4% of patients. From the best-fitting model consisting of 15 clinical elements associated with post-PCI bleeding in a random 80% training cohort, we developed a parsimonious risk algorithm. Predictors of bleeding included age, gender, previous heart failure, glomerular filtration rate, peripheral vascular disease, no previous PCI, New York Heart Association/Canadian Cardiovascular Society Functional Classification class IV heart failure, ST-elevation myocardial infarction, non–ST-elevation myocardial infarction, and cardiogenic shock. The parsimonious model was validated in the remaining 20% of the population (c-statistic, 0.72) and in clinically relevant subgroups of patients. This simplified model was used to derive a clinical risk algorithm, with larger numbers corresponding with greater risk. In 3 categories, bleeding rates were greater in patients with higher estimates (≤7, 0.7%; 8 to 17, 1.8%; ≥18, 5.1%).

Conclusions— This report identifies baseline clinical factors associated with bleeding and proposes a clinically useful algorithm to estimate bleeding risk. This model is potentially actionable in altering therapeutic decision making and improving outcomes in patients undergoing PCI.

Identification of risk factors for the prevalence and persistence of Salmonella in Belgian broiler chicken flocks

According to the European Food Safety Authority, salmonellosis is still one of the main causes of infectious foodborne gastroenteritis in humans. Broilers are an important source of salmonellosis after eggs and pork. Between 1987 and 1999 the trend of human salmonellosis incidence in Belgium increased constantly. However, from 2000 until 2005 a decrease in human cases was observed, probably following the sanitary measures implemented in the poultry breeder and laying sector. In order to decrease human infections it is essential to tackle the problem at the farm level to minimize cross-contamination from farm to fork. This paper seeks to answer two questions: (i) given the Salmonella status of the farm at a certain occasion (equal to the sampling time of the flock), what are the risk factors that the farm will be Salmonella positive at a following occasion? And (ii) what are the risk factors for a farm to be persistently positive for two consecutive flocks? We used surveillance data on 6824 broiler flocks studied for Salmonella infectivity from 2005 to 2006 in Belgium. The farms were tested regularly (3 weeks before slaughter of each broiler flock) for the presence of Salmonella based on multiple faecal samples per flock on a farm yielding clustered data. Generalized estimating equations, alternating logistic regression models, and random-intercept logistic regression models were employed to analyse these correlated binary data. Our results indicated that there are many factors that influence Salmonella risk in broiler flocks, and that they interact. Accounting for interactions between risk factors leads to an improved determination of those risk factors that increase infection with Salmonella. For the conditional analysis, the risk factors found to increase the risk of Salmonella infection on a farm at a current occasion given the previous Salmonella status included: Salmonella infection of day-old chicks (of the current flock); a previously infected flock even though the farm was equipped with a hygiene place to change clothes prior to entering the broiler house; having temporary workmen when there was a separation between birds of different species; and separating birds of different species in the Walloon region relative to the Flanders region. Sanitary measures such as a cleaning and disinfecting procedure conducted by an external cleaning firm, applying the all-in all-out procedure, and hand washing decreased the risk despite their interaction with other factors. From the joint analysis, the most important factors identified for increased risk for persistent Salmonella on a farm involved the interaction between having temporary workmen when there were poultry or farmers in contact with foreign poultry or persons, and the interaction between having temporary workmen when there were poultry or farmers in contact with external poultry or persons.

Dietary compound score and risk of age-related macular degeneration in the age-related eye disease study

PURPOSE

Because foods provide many nutrients that may interact to modify risk for multifactorial diseases such as age-related macular degeneration (AMD), we sought to develop a composite scoring system to summarize the combined effect of multiple dietary nutrients on AMD risk. This has not been done previously.

DESIGN

Cross-sectional study.

PARTICIPANTS

From the 4003 participants in the Age-Related Eye Disease Study (AREDS), there were 7,934 eyes included in this study.

METHODS

Considering dietary intakes of vitamins C and E, zinc, lutein/zeaxanthin, docosahexaenoic acid, eicosapentaenoic acid, and low-dietary glycemic index (dGI) from AREDS baseline information, we assigned each nutrient a percentile rank score then summed them into a compound score for each participant. Using eye as the unit of analysis, we evaluated the association between the compound score and risk of prevalent AMD. Validation, fitness, and performance of the model were evaluated using bootstrapping techniques, adjusted quasi-likelihood under the independence model criterion, and the c-index, respectively.

MAIN OUTCOME MEASURES

Stereoscopic fundus photographs of the macula were taken and graded at baseline using the AREDS protocol and AMD Classification System.

RESULTS

Our results showed that higher compound scores were associated with lower risk for early AMD, indicated by drusen, and advanced AMD. Validation analyses indicated that these relationships are robust (the average 50-time bootstrapping per quartile odds ratios = 0.727, 0.827, and 0.753, respectively, for drusen, and 0.616, 0.536, and 0.572, respectively, for advanced AMD). Model selection analyses suggested that the compound score should be included, but that measures of dietary beta-carotene should not be included.

CONCLUSIONS

We found that consuming diets that provide low dGI and higher intakes of these nutrients were associated with the greatest reduction in risk for prevalent drusen and advanced AMD, whereas dietary beta-carotene did not affect these relationships. These findings warrant further prospective studies.

FINANCIAL DISCLOSURE(S)

Proprietary or commercial disclosure may be found after the references.

A latent model to detect multiple clusters of varying sizes

This article develops a latent model and likelihood-based inference to detect temporal clustering of events. The model mimics typical processes generating the observed data. We apply model selection techniques to determine the number of clusters, and develop likelihood inference and a Monte Carlo expectation-maximization algorithm to estimate model parameters, detect clusters, and identify cluster locations. Our method differs from the classical scan statistic in that we can simultaneously detect multiple clusters of varying sizes. We illustrate the methodology with two real data applications and evaluate its efficiency through simulation studies. For the typical data-generating process, our methodology is more efficient than a competing procedure that relies on least squares.

Using second-order generalized estimating equations to model heterogeneous intraclass correlation in cluster-randomized trials

In cluster-randomized trials, it is commonly assumed that the magnitude of the correlation among subjects within a cluster is constant across clusters. However, the correlation may in fact be heterogeneous and depend on cluster characteristics. Accurate modeling of the correlation has the potential to improve inference. We use second-order generalized estimating equations to model heterogeneous correlation in cluster-randomized trials. Using simulation studies we show that accurate modeling of heterogeneous correlation can improve inference when the correlation is high or varies by cluster size. We apply the methods to a cluster-randomized trial of an intervention to promote breast cancer screening.

Quantitative gait markers and incident fall risk in older adults

BACKGROUND

Identifying quantitative gait markers of falls in older adults may improve diagnostic assessments and suggest novel intervention targets.

METHODS

We studied 597 adults aged 70 and older (mean age 80.5 years, 62% women) enrolled in an aging study who received quantitative gait assessments at baseline. Association of speed and six other gait markers (cadence, stride length, swing, double support, stride length variability, and swing time variability) with incident fall rate was studied using generalized estimation equation procedures adjusted for age, sex, education, falls, chronic illnesses, medications, cognition, disability as well as traditional clinical tests of gait and balance.

RESULTS

Over a mean follow-up period of 20 months, 226 (38%) of the 597 participants fell. Mean fall rate was 0.44 per person-year. Slower gait speed (risk ratio [RR] per 10 cm/s decrease 1.069, 95% confidence interval [CI] 1.001-1.142) was associated with higher risk of falls in the fully adjusted models. Among six other markers, worse performance on swing (RR 1.406, 95% CI 1.027-1.926), double-support phase (RR 1.165, 95% CI 1.026-1.321), swing time variability (RR 1.007, 95% CI 1.004-1.010), and stride length variability (RR 1.076, 95% CI 1.030-1.111) predicted fall risk. The associations remained significant even after accounting for cognitive impairment and disability.

CONCLUSIONS

Quantitative gait markers are independent predictors of falls in older adults. Gait speed and other markers, especially variability, should be further studied to improve current fall risk assessments and to develop new interventions.

High-sensitivity C-reactive protein as a serum predictor of nonalcoholic fatty liver disease based on the Akaike Information Criterion scoring system in the general Japanese population

BACKGROUND

High-sensitivity C-reactive protein (hs-CRP) has been developed and used as a marker to predict coronary vascular diseases in metabolic syndrome (MS). We investigated whether serum hs-CRP concentration was associated with nonalcoholic fatty liver disease (NAFLD) based on the Akaike Information Criterion (AIC) scoring system, using patients from the human dry dock program.

METHODS

From 2004 to 2005, 1254 subjects visited our human dry dock annual checkup program. We excluded from this study individuals with markers of viral hepatitis and those whose alcohol consumption was more than 20 g/week. Finally, 230 subjects (93 men and 137 women) were investigated. Serum hs-CRP concentrations were measured using a highly sensitive latex agglutination assay system. The AIC scoring system with the CATDAP-20 program was introduced to evaluate the parameters that are present frequently in NAFLD.

RESULTS

NAFLD was diagnosed by ultrasound sonography in 35.4% of the men and 18.9% of the women. High serum hs-CRP concentrations were observed in women with NAFLD (normal: NAFLD = 0.45:1.47 mg/l, P < 0.05). Body mass index (BMI), waist circumference, and body weight had the three lowest AIC score (P = 4.5e(-19) to 2.6e(-16)). hs-CRP was the third lowest variable among the serum markers associated with NAFLD (P = 2.3e(-6)) In addition, the hs-CRP concentration was correlated strongly with triglyceride values in females with NAFLD and with fasting blood glucose, HbA1c, and waist/hip ratio in males with NAFLD (P < 0.05).

CONCLUSIONS

The serum hs-CRP concentration was a strong predictor for NAFLD with a low AIC score and correlated with serum markers that indicated lipid and glucose metabolism.

Dimensionality reduction for knowledge discovery in medical claims database: application to antidepressant medication utilization study

Data mining, through its capacity to discover knowledge embedded in large databases to improve organizational decision-making, has the potential to contribute to efficiencies and cost savings in the increasingly costly healthcare industry. One important aspect of the methods of mining medical databases includes reducing dimensionality through feature selection. Traditionally feature selection is accomplished through stepwise regression, which tends to produce an unnecessarily high number of "significant" variables. This paper applies a filter-based feature selection method using inconsistency rate measure and discretization, to a medical claims database to predict the adequacy of duration of antidepressant medication utilization. Compared to traditional stepwise logistic regression, which selected seven variables from a total of nine potential explanatory variables to characterize patients with inadequate antidepressant medication utilization, the filter-based method selected two variables (age and number of claims) to achieve a similar prediction accuracy. This comparison suggests it may be feasible and efficient to apply the filter-based feature selection method to reduce the dimensionality of healthcare databases.

Fractional polynomials and model selection in generalized estimating equations analysis, with an application to a longitudinal epidemiologic study in Australia

In epidemiologic studies, researchers often need to establish a nonlinear exposure-response relation between a continuous risk factor and a health outcome. Furthermore, periodic interviews are often conducted to take repeated measurements from an individual. The authors proposed to use fractional polynomial models to jointly analyze the effects of 2 continuous risk factors on a health outcome. This method was applied to an analysis of the effects of age and cumulative fluoride exposure on forced vital capacity in a longitudinal study of lung function carried out among aluminum workers in Australia (1995-2003). Generalized estimating equations and the quasi-likelihood under the independence model criterion were used. The authors found that the second-degree fractional polynomial models for age and fluoride fitted the data best. The best model for age was robust across different models for fluoride, and the best model for fluoride was also robust. No evidence was found to suggest that the effects of smoking and cumulative fluoride exposure on change in forced vital capacity over time were significant. The trend 1 model, which included the unexposed persons in the analysis of trend in forced vital capacity over tertiles of fluoride exposure, did not fit the data well, and caution should be exercised when this method is used.

Long-term course and effectiveness of combination therapy in Alzheimer disease

OBJECTIVE

To compare the real-world clinical effectiveness and long-term clinical trajectory in patients with Alzheimer disease (AD) treated with combination (COMBO) therapy consisting of cholinesterase-inhibitor (CI) plus memantine (MEM) versus CI alone versus no treatment with either.

METHODS

Three hundred eighty-two subjects with probable AD underwent serial clinical evaluations at a memory disorders unit. Cognition was assessed by the Information-Memory-Concentration subscale of the Blessed Dementia Scale (BDS) and function was assessed by the Weintraub Activities of Daily Living Scale (ADL) at 6-month intervals. One hundred forty-four subjects received standard care without CI or MEM (NO-RX), 122 received CI monotherapy, and 116 received COMBO therapy with CI plus MEM. Mean follow-up was 30 months (4.1 visits) and mean cumulative medication treatment time was 22.5 months. Rates of decline were analyzed using mixed-effects regression models, and Cohen's d effect sizes were calculated annually for years 1 to 4.

RESULTS

Covarying for baseline scores, age, education, and duration of illness, the COMBO group had significantly lower mean annualized rates of deterioration in BDS and ADL scores compared with the CI (P<0.001; Cohen's dBDS=0.10-0.34 and dADL=0.23-0.46 at 1 to 2 y) and NO-RX groups (P<0.001; Cohen's dBDS=0.56-0.73 and dADL=0.32-0.48 at 1 to 2 y). For the COMBO group, Cohen's d effect sizes increased with treatment duration. Similar comparisons significantly favored the CI over the NO-RX group on the BDS.

CONCLUSIONS

COMBO therapy slows cognitive and functional decline in AD compared with CI monotherapy and no treatment. These benefits had small-to-medium effect sizes that increased with time on treatment and were sustained for years.

Predicting worsening asthma control following the common cold

The asthmatic response to the common cold is highly variable, and early characteristics that predict worsening of asthma control following a cold have not been identified. In this prospective multicentric cohort study of 413 adult subjects with asthma, the mini-Asthma Control Questionnaire (mini-ACQ) was used to quantify changes in asthma control and the Wisconsin Upper Respiratory Symptom Survey-21 (WURSS-21) to measure cold severity. Univariate and multivariable models were used to examine demographic, physiological, serological and cold-related characteristics for their relationship to changes in asthma control following a cold. Clinically significant worsening of asthma control was observed following a cold (mean+/-SD increase in mini-ACQ score of 0.69+/-0.93). Univariate analysis demonstrated that season, centre location, cold duration and cold severity measurements were all associated with a change in asthma control. Multivariable analysis of the covariates available within the first 2 days of cold onset revealed that the day 2 and cumulative sum of day 1 and 2 WURSS-21 scores were significant predictors of the subsequent changes in asthma control. In asthmatic subjects, cold severity within the first 2 days can be used to predict subsequent changes in asthma control. This information may help clinicians prevent deterioration in asthma control following a cold.

Epidemiology of highly endemic multiply antibiotic-resistant shigellosis in children in the Peruvian Amazon

OBJECTIVE

Our goal was to estimate the impact of a Shigella vaccine in an area where shigellosis is endemic by characterizing the disease burden and antibiotic-resistance profiles of isolates and by determining the prevalence of Shigella flexneri serotypes.

PATIENTS AND METHODS

We conducted a 43-month-long prospective, community-based diarrheal disease surveillance in 442 children <72 months of age in the Peruvian Amazon between October 1, 2002, and April 15, 2006.

RESULTS

The incidence of diarrheal disease was 4.38 episodes per child-year. The incidence rate for shigellosis was 0.34 episodes per child-year in children <72 months of age and peaked in children between 12 and 23 months at 0.43 episodes per child-year. Maternal education at or beyond the primary grade level, piped water supply, weight-for-age z score, and improved water-storage practices were the most significant determinants of disease in this community with living conditions comparable to many rural areas in the developing world.

CONCLUSIONS

Children living in this region had a 20-fold higher rate of disease incidence detected by active surveillance as those recently estimated by passive detection. Most symptomatic disease was caused by S flexneri, although the diversity of serotypes will require a multivalent vaccine to have a significant impact on the burden of disease caused by shigellosis. Several other public health disease-control interventions targeted at water source and improved storage, nutritional interventions, and improved maternal education seem to have a greater impact than a univalent S flexneri 2a vaccine.

Reciprocal influences between girls' conduct problems and depression, and parental punishment and warmth: a six year prospective analysis

Although the reciprocal effects of parenting and child behavior have long been recognized, the emphasis of empirical study in the field of developmental psychopathology has been on parenting effects on children. For girls in particular, little is known about unique parenting effects on conduct problems in comparison to depression, or vice versa. In the current study, data from the large-scale (n = 2,451) Pittsburgh Girls Study were used to examine the reciprocal relations between parenting and child behavior over a six year period (child ages 7-12 years). Girls and their caregivers (85% of whom were biological mothers) were interviewed annually in their homes. Girls reported on symptoms of conduct disorder and depression, and caregivers reported on level of parent-child warmth and use of harsh punishment. The results of generalized estimating equation regression models demonstrated that both parenting behaviors were uniquely predictive of changes in girls' conduct problems and depressed mood. When the effects of race and poverty on these associations were controlled for, both parenting effects on girls' conduct problems remained significant, but only low parental warmth remained as a significant predictor of depressed mood. Girls' conduct problems, but not depressed mood, predicted changes in harsh punishment over time. The small effect of girls' depressed mood, on changes in parental warmth, was further weakened when socio-demographic factors were also included in the model.

Smooth bootstrap methods for analysis of longitudinal data

In analysis of longitudinal data, the variance matrix of the parameter estimates is usually estimated by the 'sandwich' method, in which the variance for each subject is estimated by its residual products. We propose smooth bootstrap methods by perturbing the estimating functions to obtain 'bootstrapped' realizations of the parameter estimates for statistical inference. Our extensive simulation studies indicate that the variance estimators by our proposed methods can not only correct the bias of the sandwich estimator but also improve the confidence interval coverage. We applied the proposed method to a data set from a clinical trial of antibiotics for leprosy.

Factors associated with worsening and improving urinary incontinence across the menopausal transition

OBJECTIVE

To evaluate whether the menopausal transition is associated with worsening of urinary incontinence symptoms over 6 years in midlife women.

METHODS

We analyzed data from 2,415 women who reported monthly or more incontinence in self-administered questionnaires at baseline and during the first six annual follow-up visits (1995-2002) of the prospective cohort Study of Women's Health Across the Nation. We defined worsening as a reported increase and improving as a reported decrease in frequency of incontinence between annual visits. We classified the menopausal status of women not taking hormone therapy annually from reported menstrual bleeding patterns and hormone therapy use by interviewer questionnaire. We used generalized estimating equations methodology to evaluate factors associated with improving and worsening incontinence from year to year.

RESULTS

Over 6 years, 14.7% of incontinent women reported worsening, 32.4% reported improvement, and 52.9% reported no change in the frequency of incontinence symptoms. Compared with premenopause, perimenopause and postmenopause were not associated with worsening incontinence; for example, early perimenopause was associated with improvement (odds ratio [OR] 1.19; 95% confidence interval [CI] 1.06-1.35) and postmenopause reduced odds of worsening (OR 0.80; 95% CI 0.66-0.95). Meanwhile, each pound of weight gain increased odds of worsening (OR 1.04; 95% CI 1.03-1.05) and reduced odds of improving (OR 0.97; 95% CI 0.96-0.98) incontinence.

CONCLUSION

In midlife incontinent women, worsening of incontinence symptoms was not attributable to the menopausal transition. Modifiable factors such as weight gain account for worsening of incontinence during this life stage.

Population pharmacokinetic data analysis of three phase I studies of matuzumab, a humanised anti-EGFR monoclonal antibody in clinical cancer development

A population pharmacokinetic model based on data from three phase I studies was to be developed including a covariate analysis to describe the concentration–time profiles of matuzumab, a novel humanised monoclonal antibody. Matuzumab was administered as multiple 1 h i.v. infusions with 11 different dosing regimens ranging from 400 to 2000 mg, q1w–q3w. For analysis, 90 patients with 1256 serum concentration–time data were simultaneously fitted using the software NONMEM™. Data were best described using a two-compartment model with the parameters central (V₁) and peripheral distribution volume (V₂), intercompartmental (Q) and linear (CLL) clearance and an additional nonlinear elimination pathway (K_m, V_max). Structural parameters were in agreement with immunoglobulin characteristics. In total, interindividual variability on V_max, CLL, V₁ and V₂ and interoccasion variability on CLL was 22–62% CV. A covariate analysis identified weight having an influence on V₁ (+0.44% per kg) and CLL (+0.87% per kg). All parameters were estimated with good precision (RSE<39%). A robust population pharmacokinetic model for matuzumab was developed, including a nonlinear pharmacokinetic process. In addition, relevant and plausible covariates were identified and incorporated into the model. When correlated to efficacy, this model could serve as a tool to guide dose selection for this ‘targeted’ cancer therapy.

Effects of presentation and electrocardiogram on time to treatment of hyperkalemia

OBJECTIVES

To assess the time to treatment for emergency department (ED) patients with critical hyperkalemia and to determine whether the timing of treatment was associated with clinical characteristics or electrocardiographic abnormalities.

METHODS

The authors performed a retrospective chart review of ED patients with the laboratory diagnosis of hyperkalemia (potassium level > 6.0 mmol/L). Patients presenting in cardiac arrest or who were referred for hyperkalemia or dialysis were excluded. Patient charts were reviewed to find whether patients received specific treatment for hyperkalemia and, if so, what clinical attributes were associated with the time to initiation of treatment.

RESULTS

Of 175 ED visits that occurred over a 1-year time period, 168 (96%) received specific treatment for hyperkalemia. The median time from triage to initiation of treatment was 117 minutes (interquartile range [IQR] = 59 to 196 minutes). The 7 cases in which hyperkalemia was not treated include 4 cases in which the patient was discharged home, with a missed diagnosis of hyperkalemia. Despite initiation of specific therapy for hyperkalemia in 168 cases, 2 patients died of cardiac arrhythmias. Among the patients who received treatment, 15% had a documented systolic blood pressure (sBP) < 90 mmHg, and 30% of treated patients were admitted to intensive care units. The median potassium value was 6.5 mmol/L (IQR = 6.3 to 7.1 mmol/L). The predominant complaints were dyspnea (20%) and weakness (19%). Thirty-six percent of patients were taking angiotensin-converting enzyme (ACE) inhibitors. Initial electrocardiograms (ECGs) were abnormal in 83% of patient visits, including 24% of ECGs with nonspecific ST abnormalities. Findings of peaked T-wave morphology (34%), first-degree atrioventricular block (17%), and interventricular conduction delay (12%) did not lead to early treatment. Vital sign abnormalities, including hypotension (sBP < 90 mmHg), were not associated with early treatment. The chief complaint of "unresponsive" was most likely to lead to early treatment; treatment delays occurred in patients not transported by ambulance, those with a chief complaint of syncope and those with a history of hypertension.

CONCLUSIONS

Recognition of patients with severe hyperkalemia is challenging, and the initiation of appropriate therapy for this disorder is frequently delayed.

Incorporating clinical guidelines through clinician decision-making

Background

It is generally acknowledged that a disparity between knowledge and its implementation is adversely affecting quality of care. An example commonly cited is the failure of clinicians to follow clinical guidelines. A guiding assumption of this view is that adherence should be gauged by a standard of conformance. At least some guideline developers dispute this assumption and claim that their efforts are intended to inform and assist clinical practice, not to function as standards of performance. However, their ability to assist and inform will remain limited until an alternative to the conformance criterion is proposed that gauges how evidence-based guidelines are incorporated into clinical decisions.

Methods

The proposed investigation has two specific aims to identify the processes that affect decisions about incorporating clinical guidelines, and then to develop ad test a strategy that promotes the utilization of evidence-based practices. This paper focuses on the first aim. It presents the rationale, introduces the clinical paradigm of treatment-resistant schizophrenia, and discusses an exemplar of clinician non-conformance to a clinical guideline. A modification of the original study is proposed that targets psychiatric trainees and draws on a cognitively rich theory of decision-making to formulate hypotheses about how the guideline is incorporated into treatment decisions. Twenty volunteer subjects recruited from an accredited psychiatry training program will respond to sixty-four vignettes that represent a fully crossed 2 × 2 × 2 × 4 within-subjects design. The variables consist of criteria contained in the clinical guideline and other relevant factors. Subjects will also respond to a subset of eight vignettes that assesses their overall impression of the guideline. Generalization estimating equation models will be used to test the study's principal hypothesis and perform secondary analyses.

Implications

The original design of phase two of the proposed investigation will be changed in recognition of newly published literature on the relative effectiveness of treatments for schizophrenia. It is suggested that this literature supports the notion that guidelines serve a valuable function as decision tools, and substantiates the importance of decision-making as the means by which general principles are incorporated into clinical practice.

Inference Methods for the Conditional Logistic Regression Model with Longitudinal Data

This paper considers inference methods for case-control logistic regression in longitudinal setups. The motivation is provided by an analysis of plains bison spatial location as a function of habitat heterogeneity. The sampling is done according to a longitudinal matched case-control design in which, at certain time points, exactly one case, the actual location of an animal, is matched to a number of controls, the alternative locations that could have been reached. We develop inference methods for the conditional logistic regression model in this setup, which can be formulated within a generalized estimating equation (GEE) framework. This permits the use of statistical techniques developed for GEE-based inference, such as robust variance estimators and model selection criteria adapted for non-independent data. The performance of the methods is investigated in a simulation study and illustrated with the bison data analysis.

Social network-related risk factors for bloodborne virus infections among injection drug users receiving syringes through secondary exchange

Secondary syringe exchange (SSE) refers to the exchange of sterile syringes between injection drug users (IDUs). To date there has been limited examination of SSE in relation to the social networks of IDUs. This study aimed to identify characteristics of drug injecting networks associated with the receipt of syringes through SSE. Active IDUs were recruited from syringe exchange and methadone treatment programs in Montreal, Canada, between April 2004 and January 2005. Information on each participant and on their drug-injecting networks was elicited using a structured, interviewer-administered questionnaire. Subjects' network characteristics were examined in relation to SSE using regression models with generalized estimating equations. Of 218 participants, 126 were SSE recipients with 186 IDUs in their injecting networks. The 92 non-recipients reported 188 network IDUs. Networks of SSE recipients and non-recipients were similar with regard to network size and demographics of network members. In multivariate analyses adjusted for age and gender, SSE recipients were more likely than non-recipients to self-report being HIV-positive (OR=3.56 [1.54-8.23]); require or provide help with injecting (OR=3.74 [2.01-6.95]); have a social network member who is a sexual partner (OR=1.90 [1.11-3.24]), who currently attends a syringe exchange or methadone program (OR=2.33 [1.16-4.70]), injects daily (OR=1.77 [1.11-2.84]), and shares syringes with the subject (OR=2.24 [1.13-4.46]). SSE is associated with several injection-related risk factors that could be used to help focus public health interventions for risk reduction. Since SSE offers an opportunity for the dissemination of important prevention messages, SSE-based networks should be used to improve public health interventions. This approach can optimize the benefits of SSE while minimizing the potential risks associated with the practice of secondary exchange.

Determinants of injection drug user (IDU) syringe sharing: the relationship between availability of syringes and risk network member characteristics in Winnipeg, Canada

AIMS

Despite the establishment of syringe exchange programmes, syringe-sharing behaviour remains common among some injection drug users (IDU). Previous studies have identified several individual- and social network-level variables associated with syringe sharing. We examine the extent to which each of these variables is related independently to this behaviour within a diverse study population.

DESIGN, SETTING AND PARTICIPANTS

A cross-sectional survey of 435 IDU conducted between December 2003 and September 2004 in Winnipeg, Canada.

MEASUREMENTS

Individual and social-network variables were obtained from a survey instrument administered through a personal interview. Syringe sharing was defined as receptive syringe sharing in the last 6 months. Logistic regression analysis with generalized estimating equations was used to determine simultaneously the role of individual-level and risk network member-level variables on the odds of syringe sharing.

FINDINGS

Individuals' relationship to a risk network member (sex partner, OR: 15.3 95% CI: 7.6-30.8; family member, OR: 3.4 95% CI: 1.3-9.0) and difficulty of access to syringes (OR: 3.6 95% CI: 1.3-9.9) were predictive of syringe sharing. Dyads who 'often' pooled resources to obtain drugs were at 4.9 times (95% CI: 2.1-11.6) the odds of syringe sharing, while those who 'sometimes' pooled resources were at 2.8 times (95% CI: 1.1-6.7) the odds, compared to those who 'never' pooled resources together.

CONCLUSIONS

Syringe sharing in this population depended on both the availability of clean syringes and social network relationships. Adopting interventions that take into account relationships and behaviours that shape social norms present in networks/dyads would be a necessary prevention strategy alongside the provision of clean syringes.

Low IFN-gamma production in the first year of life as a predictor of wheeze during childhood

BACKGROUND

Diminished cytokine production in infancy has been associated with an increased risk for allergen sensitization and early-life wheeze.

OBJECTIVE

We sought to assess the effect of low cytokine production in the first year of life on the development of wheeze through age 13 years.

METHODS

Cytokine production (IFN-gamma and IL-2) by mitogen-stimulated mononuclear cells was determined from peripheral blood samples (9.4 months, n = 118) in a subset of healthy infants enrolled in the Tucson Children's Respiratory Study. The occurrence of wheeze during the previous year was ascertained at ages 2, 3, 6, 8, 11, and 13 years by means of questionnaire. Relative risk for wheeze was computed with generalized estimating equations.

RESULTS

The risk of wheezing between 2 and 13 years was significantly higher for subjects with low 9-month IFN-gamma production (relative risk, 2.29; 95% CI, 1.35-3.89) and borderline significant for those with intermediate IFN-gamma production (relative risk, 1.59; 95% CI, 0.95-2.68) compared with those who produced high levels of IFN-gamma (P value for linear association = .002). Nine-month IL-2 production was unrelated to wheeze. In relation to complex wheezing phenotypes, 9-month IFN-gamma production was inversely related to toddler wheeze (occurring only before age 6 years, P = .03) and chronic wheeze (occurring before and after age 6 years, P = .007) but not school-age wheeze (occurring only after age 6 years, P = .06).

CONCLUSION

The results suggest that characteristics of the immune system present during the first year of life can anticipate the likelihood of development of episodes of airway obstruction characterized by wheezing.

CLINICAL IMPLICATIONS

Immune susceptibility to asthma is established very early during postnatal life.

Covariates affecting spatial variability in bison travel behavior in Yellowstone National Park

Understanding mechanisms influencing the movement paths of animals is essential for comprehending behavior and accurately predicting use of travel corridors. In Yellowstone National Park (USA), the effects of roads and winter road grooming on bison (Bison bison) travel routes and spatial dynamics have been debated for more than a decade. However, no rigorous studies have been conducted on bison spatial movement patterns. We collected 121 380 locations from 14 female bison with GPS collars in central Yellowstone to examine how topography, habitat type, roads, and elevation affected the probability of bison travel year-round. We also conducted daily winter bison road use surveys (2003-2005) to quantify how topography and habitat type influenced spatial variability in the amount of bison road travel. Using model comparison techniques, we found the probability of bison travel and spatial distribution of travel locations were affected by multiple topographic and habitat type attributes including slope, landscape roughness, habitat type, elevation, and distances to streams, foraging areas, forested habitats, and roads. Streams were the most influential natural landscape feature affecting bison travel, and results suggest the bison travel network throughout central Yellowstone is spatially defined largely by the presence of streams that connect foraging areas. Also, the probability of bison travel was higher in regions of variable topography that constrain movements, such as in canyons. Pronounced travel corridors existed both in close association with roads and distant from any roads, and results indicate that roads may facilitate bison travel in certain areas. However, our findings suggest that many road segments used as travel corridors are overlaid upon natural travel pathways because road segments receiving high amounts of bison travel had similar landscape features as natural travel corridors. We suggest that most spatial patterns in bison road travel are a manifestation of general spatial travel trends. Our research offers novel insights into bison spatial dynamics and provides conceptual and analytical frameworks for examining movement patterns of other species.

A predictive model relating daily fluctuations in summer temperatures and mortality rates

Background

In the context of climate change, an efficient alert system to prevent the risk associated with summer heat is necessary. The authors' objective was to describe the temperature-mortality relationship in France over a 29-year period and to define and validate a combination of temperature factors enabling optimum prediction of the daily fluctuations in summer mortality.

Methods

The study addressed the daily mortality rates of subjects aged over 55 years, in France as a whole, from 1975 to 2003. The daily minimum and maximum temperatures consisted in the average values recorded by 97 meteorological stations. For each day, a cumulative variable for the maximum temperature over the preceding 10 days was defined.

The mortality rate was modelled using a Poisson regression with over-dispersion and a first-order autoregressive structure and with control for long-term and within-summer seasonal trends. The lag effects of temperature were accounted for by including the preceding 5 days. A "backward" method was used to select the most significant climatic variables. The predictive performance of the model was assessed by comparing the observed and predicted daily mortality rates on a validation period (summer 2003), which was distinct from the calibration period (1975–2002) used to estimate the model.

Results

The temperature indicators explained 76% of the total over-dispersion. The greater part of the daily fluctuations in mortality was explained by the interaction between minimum and maximum temperatures, for a day t and the day preceding it. The prediction of mortality during extreme events was greatly improved by including the cumulative variables for maximum temperature, in interaction with the maximum temperatures. The correlation between the observed and estimated mortality ratios was 0.88 in the final model.

Conclusion

Although France is a large country with geographic heterogeneity in both mortality and temperatures, a strong correlation between the daily fluctuations in mortality and the temperatures in summer on a national scale was observed. The model provided a satisfactory quantitative prediction of the daily mortality both for the days with usual temperatures and for the days during intense heat episodes. The results may contribute to enhancing the alert system for intense heat waves.

Fuzzy Clusterwise Growth Curve Models via Generalized Estimating Equations: An Application to the Antisocial Behavior of Children

The growth curve model has been a useful tool for the analysis of repeated measures data. However, it is designed for an aggregate-sample analysis based on the assumption that the entire sample of respondents are from a single homogenous population. Thus, this method may not be suitable when heterogeneous subgroups exist in the population with qualitatively distinct patterns of trajectories. In this paper, the growth curve model is generalized to a fuzzy clustering framework, which explicitly accounts for such group-level heterogeneity in trajectories of change over time. Moreover, the proposed method estimates parameters based on generalized estimating equations thereby relaxing the assumption of correct specification of the population covariance structure among repeated responses. The performance of the proposed method in recovering parameters and the number of clusters is investigated based on two Monte Carlo analyses involving synthetic data. In addition, the empirical usefulness of the proposed method is illustrated by an application concerning the antisocial behavior of a sample of children.

Longitudinal variable selection by cross-validation in the case of many covariates

Longitudinal models are commonly used for studying data collected on individuals repeatedly through time. While there are now a variety of such models available (marginal models, mixed effects models, etc.), far fewer options exist for the closely related issue of variable selection. In addition, longitudinal data typically derive from medical or other large-scale studies where often large numbers of potential explanatory variables and hence even larger numbers of candidate models must be considered. Cross-validation is a popular method for variable selection based on the predictive ability of the model. Here, we propose a cross-validation Markov chain Monte Carlo procedure as a general variable selection tool which avoids the need to visit all candidate models. Inclusion of a 'one-standard error' rule provides users with a collection of good models as is often desired. We demonstrate the effectiveness of our procedure both in a simulation setting and in a real application.

Model selection for incomplete and design-based samples

The Akaike information criterion, AIC, is one of the most frequently used methods to select one or a few good, optimal regression models from a set of candidate models. In case the sample is incomplete, the naive use of this criterion on the so-called complete cases can lead to the selection of poor or inappropriate models. A similar problem occurs when a sample based on a design with unequal selection probabilities, is treated as a simple random sample. In this paper, we consider a modification of AIC, based on reweighing the sample in analogy with the weighted Horvitz-Thompson estimates. It is shown that this weighted AIC-criterion provides better model choices for both incomplete and design-based samples. The use of the weighted AIC-criterion is illustrated on data from the Belgian Health Interview Survey, which motivated this research. Simulations show its performance in a variety of settings.

Modeling adolescent drug-use patterns in cluster-unit trials with multiple sources of correlation using robust latent class regressions

PURPOSE

The purpose of the study is to examine variation in adolescent drug-use patterns by using latent class regression analysis and evaluate the properties of an estimating-equations approach under different cluster-unit trial designs.

METHODS

A set of second-order estimating equations for latent class models under the cluster-unit trial design are proposed. This approach models the correlation within subclusters (drug-use behaviors), but ignores the correlation within clusters (communities). A robust covariance estimator is proposed that accounts for within-cluster correlation. Performance of this approach is addressed through a Monte Carlo simulation study, and practical implications are illustrated by using data from the National Evaluation of the Enforcing Underage Drinking Laws Randomized Community Trial.

RESULTS

The example shows that the proposed method provides useful information about the heterogeneous nature of drug use by identifying two subtypes of adolescent problem drinkers. A Monte Carlo simulation study supports the proposed estimation method by suggesting that the latent class model parameters were unbiased for 30 or more clusters. Consistent with other studies of generalized estimating equation (GEE) estimators, the robust covariance estimator tended to underestimate the true variance of regression parameters, but the degree of inflation in the test size was relatively small for 70 clusters and only slightly inflated for 30 clusters.

CONCLUSIONS

The proposed model for studying adolescent drug use provides an alternative to standard diagnostic criteria, focusing on the nature of the drug-use profile, rather than relying on univariate symptom counts. The second-order GEE-type estimation procedure provided a computationally feasible approach that performed well for a moderate number of clusters and was consistent with prior studies of GEE under the generalized linear model framework.

Proximate and landscape factors influence grassland bird distributions

Ecologists increasingly recognize that birds can respond to features well beyond their normal areas of activity, but little is known about the relative importance of landscapes and proximate factors or about the scales of landscapes that influence bird distributions. We examined the influences of tree cover at both proximate and landscape scales on grassland birds, a group of birds of high conservation concern, in the Sheyenne National Grassland in North Dakota, USA. The Grassland contains a diverse array of grassland and woodland habitats. We surveyed breeding birds on 2015 100 m long transect segments during 2002 and 2003. We modeled the occurrence of 19 species in relation to habitat features (percentages of grassland, woodland, shrubland, and wetland) within each 100-m segment and to tree cover within 200-1600 m of the segment. We used information-theoretic statistical methods to compare models and variables. At the proximate scales, tree cover was the most important variable, having negative influences on 13 species and positive influences on two species. In a comparison of multiple scales, models with only proximate variables were adequate for some species, but models combining proximate with landscape information were best for 17 of 19 species. Landscape-only models were rarely competitive. Combined models at the largest scales (800-1600 m) were best for 12 of 19 species. Seven species had best models including 1600-m landscapes plus proximate factors in at least one year. These were Wilson's Phalarope (Phalaropus tricolor), Sedge Wren (Cistothorus platensis), Field Sparrow (Spizella pusilla), Grasshopper Sparrow (Ammodramus savannarum), Bobolink (Dolychonix oryzivorus), Red-winged Blackbird (Agelaius phoeniceus), and Brown-headed Cowbird (Molothrus ater). These seven are small-bodied species; thus larger-bodied species do not necessarily respond most to the largest landscapes. Our findings suggest that birds respond to habitat features at a variety of scales. Models with only landscape-scale tree cover were rarely competitive, indicating that broad-scale modeling alone, such as that based solely on remotely sensed data, is likely to be inadequate in explaining species distributions.

Long-term avifaunal impoverishment in an isolated tropical woodlot

Long-term (> 50 years) extinction patterns and processes in isolated tropical forest patches are poorly understood. Considering that forest fragments are rapidly becoming the common feature of most tropical landscapes, data on the long-term conservation value of such fragments are urgently needed. We report on avifaunal turnover in a tropical woodlot (Bogor Botanical Gardens; 86 ha; 54% native and 46% introduced plants; mean 83,649 visitors/month) that has been surveyed several times before and after its isolation in 1936. By 2004 the original avifaunal richness of this woodlot declined by 59% (97 to 40 species) and its forest-dependent avifauna declined by 60% (30 to 12 species). Large-bodied birds were particularly prone to extinction before 1987, but following this time none of the species traits we studied could be considered predictive of extinction proneness. All seven forest-dependent bird species that attempted to colonize this woodlot by 1987 perished thereafter. Our results show that area reduction, isolation, intense human use, and perverse management (e.g., understory removal) of this patch have probably negatively affected the long-term sustainability of its forest avifauna.

Enrollment of racial/ethnic minorities in NIAID-funded networks of HIV vaccine trials in the United States, 1988 to 2002

OBJECTIVE

The purpose of this study was to analyze enrollment of racial/ethnic minorities in Phase I and Phase II HIV vaccine trials in the U.S. conducted by National Institute of Allergy and Infectious Diseases (NIAID)-funded networks from 1988 to 2002.

METHODS

A centralized database was searched for all NIAID-funded networks of HIV vaccine trial enrollment data in the U.S. from 1988 through 2002. The authors reviewed data from Phase I or Phase II preventive HIV vaccine trials that included HIV-1 uninfected participants at low to moderate or high risk for HIV infection based on self-reported risk behaviors. Of 66 identified trials, 55 (52 Phase I, 3 Phase II) met selection criteria and were used for analyses. Investigators extracted data on participant demographics using statistical software.

RESULTS

A total of 3,731 volunteers enrolled in U.S. NIAID-funded network HIV vaccine trials from 1988 to 2002. Racial/ethnic minority participants represented 17% of the overall enrollment. By pooling data across all NIAID-funded networks from 1988 to 2002, the proportion of racial/ethnic minority participants was significantly greater (Fisher's exact test p-value < 0.001) in Phase II trials (278/1,061 or 26%) than in Phase I trials (347/2,670 or 13%). By generalized estimating equations, the proportion of minorities in Phase I trials increased over time (p = 0.017), indicating a significant increase in racial/ethnic minority participants from 1988 to 2002.

CONCLUSIONS

There has been a gradual increase in racial/ethnic minority participation in NIAID-funded network HIV vaccine trials in the U.S. since 1988. In the light of recent efficacy trial results, it is essential to continue to increase the enrollment of diverse populations in HIV vaccine research.