Akaike's information criterion in generalized estimating equations

Extended key-factor/key-stage analysis for longitudinal data

Key-factor/key-stage analysis was originally a descriptive approach to analyze life tables. However, this method can be extended to analyze longitudinal data in pharmaceutical experiments. By dividing the variance into components, the extended key-factor/key-stage analysis indicates which factor is influential, and through which stage the factor generates its influence in determining the outcome of treatments. Such knowledge helps us in constructing a class of nonlinear longitudinal models that can be interpretable than linear models. Example SAS programs and R programs are provided for the calculation. Supplemental materials are available for this article. Go to the publisher's online edition of Journal of Biopharmaceutical Statistics to view the supplemental files.

Variation in the use of urgent splenectomy after blunt splenic injury in adults

OBJECTIVE

Recent reports indicate that mortality after trauma center admission may be directly related to the rate of operative intervention after blunt solid organ injury. These findings bring into question the role of urgent splenectomy after blunt splenic injury (BSI). The purpose of this study was to determine the role of urgent splenectomy (defined as splenectomy within 6 hours of admission) in the management of BSI as well as the relationship between urgent splenectomy and in-hospital mortality.

METHODS

The National Trauma Data Bank for 2007 was queried for adults (18-81) who suffered BSI. Patients who died in or were transferred from the emergency department were excluded. Hierarchical multivariate models were used to account for clustering of patients within hospitals and to identify hospital and patient factors associated with urgent splenectomy. Propensity score matching was used to analyze the relationship between urgent splenectomy and mortality.

RESULTS

There were 507,202 total incidents identified. Of those, 11,793 met inclusion criteria. Urgent splenectomy was performed on 1,104 (9.4%). Hierarchical models revealed that age ≥55 years, arrival systolic blood pressure ≤90 mm Hg, no or mild head injury, increasing injury severity, and massive disruption of the spleen were associated with urgent splenectomy. Hospitals level factors associated with urgent splenectomy included hospital region, hospital type, and trauma center level. The propensity-matched cohorts revealed no association between urgent splenectomy and in-hospital mortality (odds ratio, 1.08; 95% confidence interval, 0.82-1.42).

CONCLUSION

Despite ongoing variation in the use of urgent splenectomy after BSI in adults, urgent splenectomy was not associated with in-hospital mortality.

An overview of longitudinal data analysis methods for neurological research

The purpose of this article is to provide a concise, broad and readily accessible overview of longitudinal data analysis methods, aimed to be a practical guide for clinical investigators in neurology. In general, we advise that older, traditional methods, including (1) simple regression of the dependent variable on a time measure, (2) analyzing a single summary subject level number that indexes changes for each subject and (3) a general linear model approach with a fixed-subject effect, should be reserved for quick, simple or preliminary analyses. We advocate the general use of mixed-random and fixed-effect regression models for analyses of most longitudinal clinical studies. Under restrictive situations or to provide validation, we recommend: (1) repeated-measure analysis of covariance (ANCOVA), (2) ANCOVA for two time points, (3) generalized estimating equations and (4) latent growth curve/structural equation models.

Semiparametric Transformation Rate Model for Recurrent Event Data

In this article, we propose a class of semiparametric transformation rate models for recurrent event data subject to right-censoring and potentially stopped by a terminating event (e.g., death). These transformation models include both additive rates model and proportional rates model as special cases. Respecting the property that no recurrent events can occur after the terminating event, we model the conditional recurrent event rate given survival. Weighted estimating equations are constructed to estimate the regression coefficients and baseline rate function. In particular, the baseline rate function is approximated by wavelet function. Asymptotic properties of the proposed estimators are derived and a data-dependent criterion is proposed for selecting the most suitable transformation. Simulation studies show that the proposed estimators perform well for practical sample sizes. The proposed methods are used in two real-data examples: a randomized trial of rhDNase and a community trial of Vitamin A.

Model-based determination of abatacept exposure in support of the recommended dose for Japanese rheumatoid arthritis patients

The objective of this study was to provide support for a body weight-tiered dosing regimen by characterizing abatacept pharmacokinetics (PK) and the relationship between exposure and the ACR20 (American College of Rheumatology criteria for 20% improvement) response in Japanese patients with rheumatoid arthritis (RA). A population PK model was developed using NONMEM with 2,535 samples from 344 Japanese RA patients in two clinical trials. The exposure-response relationship was characterized using a Generalized Estimating Equation (GEE) logistic regression model, with time-varying actual trough concentrations and ACR20 responder rates over 6 months in a randomized, placebo-controlled phase 2 trial for stable methotrexate. Abatacept exposure was well characterized using a linear, two-compartment model, in which body weight and the empirically calculated glomerular filtration rate were significant covariates for clearance. The ACR20 response model was developed by examining the quasi-likelihood information criterion, and the cumulative logit in the final model was specified by the log-transformed trough concentration. The predicted ACR20 responder rate was consistent with the actual values in the clinical trial and this model revealed trough concentrations higher than the recommended body weight-tiered dose are unlikely to result in substantial increases in clinical efficacy. Considering that ACR20 is a longitudinal binary variable and the response to RA treatment is delayed, the GEE model was useful for predicting the probability of an ACR20 response. In conclusion, the same dosing regimen as non-Japanese patients is recommended because a body weight-tiered dosing regimen achieves similar exposures across the wide range of body weight.

Determinants of optimal adherence over time to antiretroviral therapy amongst HIV positive adults in South Africa: a longitudinal study

Highly active antiretroviral therapy (HAART) requires strict adherence to achieve optimal clinical and survival benefits. A study was done to explore the factors affecting HAART adherence among HIV positive adults by reviewing routinely collected patient information in the Centre for the AIDS Programme of Research in South Africa's (CAPRISA) AIDS Treatment Programme. Records of 688 patients enrolled between 2004 and 2006 were analysed. Patients were considered adherent if they had taken at least 95% of their prescribed drugs. Generalized estimating equations were used to analyse the data. The results showed that HAART adherence increased over time, however, the rate of increase differed by some of the socio-demographic and behavioural characteristics of the patients. For instance, HAART adherence increased in both urban and rural treatment sites over time, but the rate of increase was higher in the rural site. This helped identify sub-populations, such as the urban population, that required ongoing adherence counseling.

Acute respiratory inflammation in children and black carbon in ambient air before and during the 2008 Beijing Olympics

BACKGROUND

Epidemiologic evidence for a causative association between black carbon (BC) and health outcomes is limited.

OBJECTIVES

We estimated associations and exposure-response relationships between acute respiratory inflammation in schoolchildren and concentrations of BC and particulate matter with an aerodynamic diameter of ≤ 2.5 μm (PM2.5) in ambient air before and during the air pollution intervention for the 2008 Beijing Olympics.

METHODS

We measured exhaled nitric oxide (eNO) as an acute respiratory inflammation biomarker and hourly mean air pollutant concentrations to estimate BC and PM2.5 exposure. We used 1,581 valid observations of 36 subjects over five visits in 2 years to estimate associations of eNO with BC and PM2.5 according to generalized estimating equations with polynomial distributed-lag models, controlling for body mass index, asthma, temperature, and relative humidity. We also assessed the relative importance of BC and PM2.5 with two-pollutant models.

RESULTS

Air pollution concentrations and eNO were clearly lower during the 2008 Olympics. BC and PM2.5 concentrations averaged over 0-24 hr were strongly associated with eNO, which increased by 16.6% [95% confidence interval (CI), 14.1-19.2%] and 18.7% (95% CI, 15.0-22.5%) per interquartile range (IQR) increase in BC (4.0 μg/m3) and PM2.5 (149 μg/m3), respectively. In the two-pollutant model, estimated effects of BC were robust, but associations between PM2.5 and eNO decreased with adjustment for BC. We found that eNO was associated with IQR increases in hourly BC concentrations up to 10 hr after exposure, consistent with effects primarily in the first hours after exposure.

CONCLUSIONS

Recent exposure to BC was associated with acute respiratory inflammation in schoolchildren in Beijing. Lower air pollution levels during the 2008 Olympics also were associated with reduced eNO.

Condom use and sexually transmitted infections among Malagasy sex workers

We evaluated whether use of a short- or longer-term recall period for condom use was superior for assessing risk of acquisition of incident sexually transmitted infection (STI). Female sex workers ( n = 1000) in Madagascar took part in a randomized trial comparing counselling strategies for male and female condom promotion. We explored associations between women's self-reported condom use with clients and non-paying partners and incident STI, examining both short-term recall (last sex act) and longer-term recall (over the past month and year). Self-reported condom use was generally not associated with reduced STI risk, whether measured at the last act or over longer time periods; with clients or non-paying partners; and through unadjusted and adjusted modelling. No condom use measure (short versus longer time periods, with one or both partner types) was strongly predictive of STI, suggesting poor validity.

Penalized generalized estimating equations for high-dimensional longitudinal data analysis

We consider the penalized generalized estimating equations (GEEs) for analyzing longitudinal data with high-dimensional covariates, which often arise in microarray experiments and large-scale health studies. Existing high-dimensional regression procedures often assume independent data and rely on the likelihood function. Construction of a feasible joint likelihood function for high-dimensional longitudinal data is challenging, particularly for correlated discrete outcome data. The penalized GEE procedure only requires specifying the first two marginal moments and a working correlation structure. We establish the asymptotic theory in a high-dimensional framework where the number of covariates p(n) increases as the number of clusters n increases, and p(n) can reach the same order as n. One important feature of the new procedure is that the consistency of model selection holds even if the working correlation structure is misspecified. We evaluate the performance of the proposed method using Monte Carlo simulations and demonstrate its application using a yeast cell-cycle gene expression data set.

Particle-associated organic compounds and symptoms in myocardial infarction survivors

CONTEXT

The aerosol components responsible for the adverse health effects of the exposure to particulate matter (PM) have not been conclusively identified, and there is especially little information on the role of particulate organic compounds (POC).

OBJECTIVE

This study evaluated the role of PM and POC with regard to daily symptoms.

METHODS

One hundred and fifty-three myocardial infarction survivors from Augsburg, Germany, recorded daily occurrence of different symptoms in winter 2003/2004. Ambient concentrations of PM with a diameter <2.5 μm (PM(2.5)), particle number concentration (PNC), PM(2.5)-bound hopanes, and polycyclic aromatic hydrocarbons (PAH) were quantified. Data were analyzed using generalized estimating equations adjusting for meteorological and other time-variant confounders.

RESULTS

The odds for avoidance of physically demanding activities due to heart problems increased immediately associated with most POC measures (e.g. 5% per 1.08 ng/m(3) increase in benzo[a]pyrene, 95%-confidence interval (CI):1-9%) and tended to a delayed decrease. After a 2-day delayed decrease associated with hopanes, the odds for shortness of breath increased consistently after 3 days with almost all POC measures (e.g. 4% per 0.21 ng/m(3) increase in 17α(H), 21β(H)-hopane, CI: 1-8%). The odds for heart palpitations marginally increased immediately in association with PNC (8% per 8146 cm(-3) increase in PNC, CI: 0-16%).

CONCLUSIONS

The study showed an association between PM, particle-bound POC, and daily symptoms. The organic compounds may be causally related with cardiovascular health or act rather as indicators for traffic- and combustion-related particles.

The impact of environmental experiences on symptoms of anxiety and depression across the life span

Symptoms of anxiety and depression are relatively stable over time. Can this stability be explained by genetic influences, or is it caused by the long-lasting effects of accumulating environmental experiences? To address this question, we analyzed longitudinally assessed symptoms of anxiety and depression in eight samples of monozygotic twins of widely varying ages. These samples were drawn from American and European population-based registries. Using hierarchical linear modeling, we examined individual differences and individual changes in the level of symptoms over time. This method enabled us to decompose the variance into the predictable variance shared by both members of each pair of twins, the differences between individuals within pairs, and the residual variance. We then modeled how these components of individual variation changed over time. Within pairs, the twins' predicted levels of symptoms increasingly diverged from childhood until late adulthood, at which point the divergence ceased. By middle adulthood, environmental experiences contributed substantially to stable and predictable interindividual differences in levels of anxiety and depression.

Estimating functions and the generalized method of moments

Estimating functions provide a very general framework for statistical inference, and are particularly useful when one is either unable or unwilling to specify a likelihood function. This paper aims to provide an accessible review of estimating function theory that has potential for application to the analysis and modelling of a wide range of complex systems. Assumptions are given in terms that can be checked relatively easily in practice, and some of the more technical derivations are relegated to an online supplement for clarity of exposition. The special case of the generalized method of moments is considered in some detail. The main points are illustrated by considering the problem of inference for a class of stochastic rainfall models based on point processes, with simulations used to demonstrate the performance of the methods.

Co-occurrence patterns of the dengue vector Aedes aegypti and Aedes mediovitattus, a dengue competent mosquito in Puerto Rico

Aedes aegypti is implicated in dengue transmission in tropical and subtropical urban areas around the world. Ae. aegypti populations are controlled through integrative vector management. However, the efficacy of vector control may be undermined by the presence of alternative, competent species. In Puerto Rico, a native mosquito, Ae. mediovittatus, is a competent dengue vector in laboratory settings and spatially overlaps with Ae. aegypti. It has been proposed that Ae. mediovittatus may act as a dengue reservoir during inter-epidemic periods, perpetuating endemic dengue transmission in rural Puerto Rico. Dengue transmission dynamics may therefore be influenced by the spatial overlap of Ae. mediovittatus, Ae. aegypti, dengue viruses, and humans. We take a landscape epidemiology approach to examine the association between landscape composition and configuration and the distribution of each of these Aedes species and their co-occurrence. We used remotely sensed imagery from a newly launched satellite to map landscape features at very high spatial resolution. We found that the distribution of Ae. aegypti is positively predicted by urban density and by the number of tree patches, Ae. mediovittatus is positively predicted by the number of tree patches, but negatively predicted by large contiguous urban areas, and both species are predicted by urban density and the number of tree patches. This analysis provides evidence that landscape composition and configuration is a surrogate for mosquito community composition, and suggests that mapping landscape structure can be used to inform vector control efforts as well as to inform urban planning.

C-reactive protein and prediction of 1-year mortality in prevalent hemodialysis patients

BACKGROUND AND OBJECTIVES

Measurement of C-reactive protein (CRP) levels remains uncommon in North America, although it is now routine in many countries. Using Dialysis Outcomes and Practice Patterns Study data, our primary aim was to evaluate the value of CRP for predicting mortality when measured along with other common inflammatory biomarkers.

DESIGN, SETTING, PARTICIPANTS, & MEASUREMENTS

We studied 5061 prevalent hemodialysis patients from 2005 to 2008 in 140 facilities routinely measuring CRP in 10 countries. The association of CRP with mortality was evaluated using Cox regression. Prediction of 1-year mortality was assessed in logistic regression models with differing adjustment variables.

RESULTS

Median baseline CRP was lower in Japan (1.0 mg/L) than other countries (6.0 mg/L). CRP was positively, monotonically associated with mortality. No threshold below which mortality rate leveled off was identified. In prediction models, CRP performance was comparable with albumin and exceeded ferritin and white blood cell (WBC) count based on measures of model discrimination (c-statistics, net reclassification improvement [NRI]) and global model fit (generalized R(2)). The primary analysis included age, gender, diabetes, catheter use, and the four inflammatory markers (omitting one at a time). Specifying NRI ≥5% as appropriate reclassification of predicted mortality risk, NRI for CRP was 12.8% compared with 10.3% for albumin, 0.8% for ferritin, and <0.1% for WBC.

CONCLUSIONS

These findings demonstrate the value of measuring CRP in addition to standard inflammatory biomarkers to improve mortality prediction in hemodialysis patients. Future studies are indicated to identify interventions that lower CRP and to identify whether they improve clinical outcomes.

Optimal cutoff values for high-risk waist circumference in older adults based on related health outcomes

The authors aimed to explore optimal cutoffs for high-risk waist circumference (WC) in older adults to assess the health risks of obesity. Prospective data from 4,996 measurements in 2,232 participants aged ≥70 years were collected during 5 triennial measurement cycles (1992/1993-2005/2006) of a population-based cohort study, the Longitudinal Aging Study Amsterdam (Amsterdam, the Netherlands). Cross-sectional associations of WC with pain, mobility limitations, incontinence, knee osteoarthritis, cardiovascular disease, and diabetes were studied. Generalized estimating equations models were fitted with restricted cubic spline functions in order to carefully study the shapes of the associations. Model fits for applying different cutoffs to categorize WC in the association with all outcomes were tested using the quasi-likelihood under the Independence Criterion (QIC). On the basis of the spline regression curves, potential WC cutoffs of approximately 109 cm in men and 98 cm in women were proposed. Based on the model fit, cutoffs between 100 cm and 106 cm were equally applicable in men but should not be higher. In women, the QIC confirmed an optimal cutoff of 99 cm.

Randomized clinical trial of brief risk reduction counseling for sexually transmitted infection clinic patients in Cape Town, South Africa

OBJECTIVES

We examined the effects of a brief counseling intervention designed to reduce HIV risk behaviors and sexually transmitted infections (STIs) among patients receiving STI services in Cape Town, South Africa.

METHODS

After randomization to either a 60-minute risk reduction counseling session or a 20-minute HIV-STI educational session, patients completed computerized sexual behavior assessments. More than 85% of the participants were retained at the 12-month follow-up.

RESULTS

There were 24% fewer incident STIs and significant reductions in unprotected vaginal and anal intercourse among participants who received risk reduction counseling relative to members of the control condition. Moderator analyses showed shorter lived outcomes for heavy alcohol drinkers than for lighter drinkers. The results were not moderated by gender.

CONCLUSIONS

Brief single-session HIV prevention counseling delivered to STI clinic patients has the potential to reduce HIV infections. Counseling should be enhanced for heavier drinkers, and sustained outcomes will require relapse prevention techniques. Disseminating effective, brief, and feasible behavioral interventions to those at highest risk for HIV infection should remain a public health priority.

Increased gonorrhoea and chlamydia testing did not increase case detection in an HIV clinical cohort 1999-2007

OBJECTIVES

Since 2003, US organisations have recommended universal screening, rather than targeted screening, of HIV-infected persons for gonorrhoea and chlamydia. The objective of this study was to determine whether wider testing resulting from these guidelines would produce an increase in gonorrhoea/chlamydia diagnoses.

METHODS

3283 patients receiving HIV care in 1999-2007 in the Johns Hopkins Hospital HIV clinic were studied. The two primary outcomes were the occurrence of any gonorrhoea/chlamydia testing in each year of care and the occurrence of any positive result(s) in years of testing. The proportion of all patients in care who were diagnosed with gonorrhoea/chlamydia was defined as the number of patients with positive results divided by the number of patients in care. Trends were analysed with repeated measures logistic regression.

RESULTS

The proportion of patients tested for gonorrhoea/chlamydia increased steadily from 0.12 in 1999 to 0.33 in 2007 (OR per year for being tested 1.17, 95% CI 1.15 to 1.19). The proportion positive among those tested decreased significantly after 2003 (OR per year 0.67, 95% CI 0.55 to 0.81). The proportion of all patients in care diagnosed with gonorrhoea/chlamydia therefore remained generally stable in 1999-2007 (OR per year 0.97, 95% CI 0.91 to 1.04).

CONCLUSIONS

Universal annual screening, as implemented, did not increase the proportion of all patients in care who were diagnosed with gonorrhoea/chlamydia. Similarly low implementation rates have been reported in cross-sectional studies. If future efforts to enhance implementation do not yield increases in diagnoses, then guidelines focusing on targeted screening of high-risk groups rather than universal screening may be warranted.

Mechanisms of functional connectivity: the case of free-ranging bison in a forest landscape

Functional connectivity is a key determinant of animal distributions in heterogeneous landscapes. Patch connectivity depends on both patch preference and accessibility, but few studies have integrated habitat selection and movement analyses to gain a general understanding of functional connectivity. In this paper, we define functional connectivity by identifying which factors influence the choice of the patch that is visited next, the location from which animals leave the current patch, and the inter-patch trajectory. Our study provides tools to anticipate movement trajectories and, therefore, animal distribution in patchy landscapes. We followed 23 radio-collared bison across the meadow network of Prince Albert National Park between 2005 and 2008. Selection of the next meadow visited over available meadows was assessed by comparing their characteristics and land cover composition of the area separating them from the departure meadow. Additionally, we used 196 bison trails originating from 29 meadows to evaluate movement rules during inter-patch travels. Bison preferred to travel in deciduous rather than in conifer stands during summer and fall but displayed no preference during winter and spring. They also selected meadows offering higher plant biomass in winter than in other seasons. Throughout the year, meadow proximity was an important determinant of meadow selection. Inter-patch trajectory was influenced by directional persistence, as well as movement biases toward the next meadow and toward canopy gaps. Unlike the choices individuals made in selecting their next meadow, bison displayed no preference between forest stands during inter-meadow travel, indicating that functional connectivity involves hierarchical movement decisions. We showed that the behavioral determinants of functional connectivity varied over spatiotemporal scales. First, forest stand composition between meadows influenced the next target, but not the trajectory during inter-meadow travels. Second, meadow selection varied among seasons. Therefore, although structural connectivity may be immutable to these behaviorally induced changes in inter-patch movements, functional connectivity would adequately account for such modifications in animal spatial dynamics.

Associations of ionizing radiation and breast cancer-related serum hormone and growth factor levels in cancer-free female A-bomb survivors

Levels of exposure to ionizing radiation are increasing for women worldwide due to the widespread use of CT and other radiologic diagnostic modalities. Exposure to ionizing radiation as well as increased levels of estradiol and other sex hormones are acknowledged breast cancer risk factors, but the effects of whole-body radiation on serum hormone levels in cancer-free women are unknown. This study examined whether ionizing radiation exposure is associated with levels of serum hormones and other markers that may mediate radiation-associated breast cancer risk. Serum samples were measured from cancer-free women who attended biennial health examinations with a wide range of past radiation exposure levels (N  =  412, ages 26-79). The women were selected as controls for separate case-control studies from a cohort of A-bomb survivors. Outcome measures included serum levels of total estradiol, bioavailable estradiol, testosterone, progesterone, prolactin, insulin-like growth factor-1 (IGF1), insulin-like growth factor-binding protein 3 (IGFBP-3), and ferritin. Relationships were assessed using repeated-measures regression models fitted with generalized estimating equations. Geometric mean serum levels of total estradiol and bioavailable estradiol increased with 1 Gy of radiation dose among samples collected from postmenopausal women (17%(1Gy), 95% CI: 1%-36% and 21%(1Gy), 95% CI: 4%-40%, respectively), while they decreased in samples collected from premenopausal women (-11%(1Gy), 95% CI: -20%-1% and -12%(1Gy), 95% CI: -20%- -2%, respectively). Interactions by menopausal status were significant (P  =  0.003 and P < 0.001, respectively). Testosterone levels increased with radiation dose in postmenopausal samples (30.0%(1Gy), 95% CI: 13%-49%) while they marginally decreased in premenopausal samples (-10%(1Gy), 95% CI: -19%-0%) and the interaction by menopausal status was significant (P < 0.001). Serum levels of IGF1 increased linearly with radiation dose (11%(1Gy), 95% CI: 2%-18%) and there was a significant interaction by menopausal status (P  =  0.014). Radiation-associated changes in serum levels of estradiol, bioavailable estradiol, testosterone and IGF1 were modified by menopausal status at the time of collection. No associations with radiation were observed in serum levels of progesterone, prolactin, IGFBP-3 or ferritin.

Habitat selection, reproduction and predation of wintering lemmings in the Arctic

Snow cover has dramatic effects on the structure and functioning of Arctic ecosystems in winter. In the tundra, the subnivean space is the primary habitat of wintering small mammals and may be critical for their survival and reproduction. We have investigated the effects of snow cover and habitat features on the distributions of collared lemming (Dicrostonyx groenlandicus) and brown lemming (Lemmus trimucronatus) winter nests, as well as on their probabilities of reproduction and predation by stoats (Mustela erminea) and arctic foxes (Vulpes lagopus). We sampled 193 lemming winter nests and measured habitat features at all of these nests and at random sites at two spatial scales. We also monitored overwinter ground temperature at a subsample of nest and random sites. Our results demonstrate that nests were primarily located in areas with high micro-topography heterogeneity, steep slopes, deep snow cover providing thermal protection (reduced daily temperature fluctuations) and a high abundance of mosses. The probability of reproduction increased in collared lemming nests at low elevation and in brown lemming nests with high availability of some graminoid species. The probability of predation by stoats was density dependent and was higher in nests used by collared lemmings. Snow cover did not affect the probability of predation of lemming nests by stoats, but deep snow cover limited predation attempts by arctic foxes. We conclude that snow cover plays a key role in the spatial structure of wintering lemming populations and potentially in their population dynamics in the Arctic.

The effect of sentinel injury on Medicare expenditures over time

OBJECTIVES

To examine the long-term effect of sentinel injury (unintentional injury involving serious health-related consequences) among older adults on Medicare expenditures.

DESIGN

Secondary data analysis of the Medicare Current Beneficiary Survey, a nationally representative survey of Medicare Beneficiaries.

SETTING

Noninstitutionalized community dwellers.

PARTICIPANTS

Older adults (N = 12,318) continuously enrolled in Medicare Fee-for-Service under Old Age Survivors Insurance Benefits surveyed between October 1998 and December 2004.

MEASUREMENTS

Monthly total Medicare expenditures served as the dependent variable. Injury status (preinjury, injury episode, postinjury) was identified from Medicare claims and specified as a set of dummy variables. Injury episodes began with the first index injury claim identified and ended when no further injury claims were found within 180 days. Population-averaged models using generalized estimating equation techniques were estimated to explore changes in Medicare expenditures over time after adjusting for casemix differences. A case-crossover design was used to compare monthly Medicare expenditures before and after sentinel injury events.

RESULTS

Fifteen percent of beneficiaries sustained at least one sentinel injury. Medicare expenditures increased sharply during sentinel injury episodes (β = 1.703, P < .001) and remained at least 28% higher than would otherwise be expected for 27 uninterrupted months following injury. Additive Medicare expenditures associated with sentinel injury over 3 years were estimated at $28,885.

CONCLUSION

Consequences of sentinel injury in older adults extend well beyond the period typically considered to be an acute injury episode. Better understanding of the long-term consequences of injury-related outcomes is needed to achieve public health goals of reducing injury and improving injury-related medical care.

Hospitalizations for varicella in children and adolescents in a referral hospital in Hong Kong, 2004 to 2008: a time series study

Background

Varicella accounts for significant morbidities and remains a public health issue worldwide. Climatic factors have been shown to associate with the incidence and transmission of various infectious diseases. We describe the epidemiology of varicella in paediatric patients hospitalized at a tertiary referral hospital in Hong Kong from 2004 to 2008, and to explore the possible association between the occurrence of varicella infection and various climatic factors.

Methods

The hospital discharge database of Princess Margaret Hospital was retrospectively analyzed for admissions associated with varicella from 2004 to 2008. Meteorological data were obtained from the monthly meteorological reports from the Hong Kong Observatory website. Time series analysis was performed with Poisson regression using a Generalized Estimating Equation (GEE) approach.

Results

During the study period, 598 children were hospitalized for varicella. The mean age on admission was 57.6 months, and the mean duration of hospitalization was 3.7 days. The overall complication rate was 47%. The mean monthly relative humidity, especially in cool seasons, was inversely correlated with the monthly varicella cases of the same month.

Conclusions

Varicella can lead to serious complications and prolonged hospitalization, even in previously healthy children. Lower relative humidity in cool seasons is associated with higher number of paediatric varicella hospital admissions. These findings are useful for a better understanding of the pattern of paediatric varicella hospitalization in Hong Kong.

Asymmetric responsiveness of physician prescription behavior to drug promotion of competitive brands within an established therapeutic drug class

This article examines the impact of direct-to-physician, direct-to-consumer, and other marketing activities by pharmaceutical companies on a mature drug category which is in the later stage of its life cycle and in which generics have accrued a significant market share. The main objective of this article is to quantitatively estimate the impact of pharmaceutical promotions on physician prescribing behavior for three different statin brands, after controlling for factors such as patient, physician and physician practice characteristics, generic pressure, et cetera. Using unique panel data of physicians, combined with patient pharmacy prescription records, the authors developed a physician level generalized linear regression model. The generalized estimating equations method was used to account for within physician serial correlations and estimate physician population averaged effects. The findings reveal that even though on average the marketing efforts affect the brand share positively, the magnitude of the effects is very brand specific. Generally, each statin brand has its own trend and because of this, the best choice of predictors for one brand could be suboptimal for another.

High disease activity is associated with an increased risk of infection in patients with rheumatoid arthritis

OBJECTIVE

To determine the relationship of disease activity to infections in patients with rheumatoid arthritis (RA).

METHODS

From the CORRONA database, the incidence of physician-reported infections in RA patients on stable disease-modifying antirheumatic drug, biological, and corticosteroid therapy for at least 6 months was ascertained. Two composite measures of disease activity were defined: clinical disease activity index (CDAI) and disease activity score 28 (DAS28). Incident rate ratios (IRR) were calculated using generalised estimating equation Poisson regression models adjusted for demographics, medications and clinical factors.

RESULTS

Of 1 6242 RA patients, 6242 were on stable therapy for at least 6 months and were eligible for analysis. 2282 infections were reported in the cohort, followed over 7290 patient-years. After controlling for possible confounders, disease activity was associated with an increased rate of infections. Each 0.6 unit increase in DAS28 score corresponded to a 4% increased rate of outpatient infections (IRR 1.04, p=0.01) and a 25% increased rate of infections requiring hospitalisation (IRR 1.25, p=0.03). There was a dichotomy in the relationship between infections and CDAI scores. For CDAI <10 (mild disease activity) patients had a 12% increased rate of outpatient infections with each 5 unit increase in CDAI score (IRR 1.12, p=0.003). At CDAI scores ≥10, there was no further increase in the rate of outpatient infections associated with higher disease activity. The relationship of CDAI to hospitalised infections showed similar trends to outpatient data but did not reach statistical significance after multivariate analysis (CDAI <10: IRR 1.56, p=0.08).

CONCLUSIONS

In this large cohort of RA patients, higher disease activity was associated with a higher probability of developing infections.

Exploring the influence of proximity to death on disease-specific hospital expenditures: a carpaccio of red herrings

It has been demonstrated repeatedly that time to death is a much better predictor of health care expenditures than age. This is known as the 'red herring' hypothesis. In this article, we investigate whether this is also the case regarding disease-specific hospital expenditures. Longitudinal data samples from the Dutch hospital register (n=11 253 455) were used to estimate 94 disease-specific two-part models. Based on these models, Monte Carlo simulations were used to assess the predictive value of proximity to death and age on disease-specific expenditures. Results revealed that there was a clear effect of proximity of death on health care expenditures. This effect was present for most diseases and was strongest for most cancers. However, even for some less fatal diseases, proximity to death was found to be an important predictor of expenditures. Controlling for proximity to death, age was found to be a significant predictor of expenditures for most diseases. However, its impact is modest when compared to proximity to death. Considering the large variation in the degree to which proximity to death and age matter for each specific disease, we may speak not only of age as a 'red herring' but also of a 'carpaccio of red herrings'.

Traumatic bereavement, acute dissociation, and posttraumatic stress: 14 years after the MS Estonia disaster

This prospective longitudinal study aimed to examine posttraumatic stress in survivors 14 years after a ferry disaster, and estimate short- and long-term changes in stress associated with traumatic bereavement and acute dissociation. There were 852 people who perished in the disaster, 137 survived. The 51 Swedish survivors were surveyed with the Impact of Event Scale-Revised (IES-R) at 3 months, 1, 3, and 14 years (response rates 82%, 65%, 51%, and 69%). Symptoms decreased from 3 months to 1 year; no change was found thereafter. After 14 years, 27% reported significant symptoms. Traumatic bereavement, but not acute dissociation, was associated with long-term symptom elevation. Chronic posttraumatic stress can persist in a minority of survivors, and traumatic bereavement appears to hinder recovery.

Water use practices limit the effectiveness of a temephos-based Aedes aegypti larval control program in Northern Argentina

BACKGROUND

A five-year citywide control program based on regular application of temephos significantly reduced Aedes aegypti larval indices but failed to maintain them below target levels in Clorinda, northern Argentina. Incomplete surveillance coverage and reduced residuality of temephos were held as the main putative causes limiting effectiveness of control actions.

METHODOLOGY

The duration of temephos residual effects in household-owned water-holding tanks (the most productive container type and main target for control) was estimated prospectively in two trials. Temephos was applied using spoons or inside perforated small zip-lock bags. Water samples from the study tanks (including positive and negative controls) were collected weekly and subjected to larval mortality bioassays. Water turnover was estimated quantitatively by adding sodium chloride to the study tanks and measuring its dilution 48 hs later.

PRINCIPAL FINDINGS

The median duration of residual effects of temephos applied using spoons (2.4 weeks) was significantly lower than with zip-lock bags (3.4 weeks), and widely heterogeneous between tanks. Generalized estimating equations models showed that bioassay larval mortality was strongly affected by water type and type of temephos application depending on water type. Water type and water turnover were highly significantly associated. Tanks filled with piped water had high turnover rates and short-lasting residual effects, whereas tanks filled with rain water showed the opposite pattern. On average, larval infestations reappeared nine weeks post-treatment and seven weeks after estimated loss of residuality.

CONCLUSIONS

Temephos residuality in the field was much shorter and more variable than expected. The main factor limiting temephos residuality was fast water turnover, caused by householders' practice of refilling tanks overnight to counteract the intermittence of the local water supply. Limited field residuality of temephos accounts in part for the inability of the larval control program to further reduce infestation levels with a treatment cycle period of 3 or 4 months.

Age, period, and cohort analysis of regular dental care behavior and edentulism: a marginal approach

Background

To analyze the regular dental care behavior and prevalence of edentulism in adult Danes, reported in sequential cross-sectional oral health surveys by the application of a marginal approach to consider the possible clustering effect of birth cohorts.

Methods

Data from four sequential cross-sectional surveys of non-institutionalized Danes conducted from 1975-2005 comprising 4330 respondents aged 15+ years in 9 birth cohorts were analyzed. The key study variables were seeking dental care on an annual basis (ADC) and edentulism. For the analysis of ADC, survey year, age, gender, socio-economic status (SES) group, denture-wearing, and school dental care (SDC) during childhood were considered. For the analysis of edentulism, only respondents aged 35+ years were included. Survey year, age, gender, SES group, ADC, and SDC during childhood were considered as the independent factors. To take into account the clustering effect of birth cohorts, marginal logistic regressions with an independent correlation structure in generalized estimating equations (GEE) were carried out, with PROC GENMOD in SAS software.

Results

The overall proportion of people seeking ADC increased from 58.8% in 1975 to 86.7% in 2005, while for respondents aged 35 years or older, the overall prevalence of edentulism (35+ years) decreased from 36.4% in 1975 to 5.0% in 2005. Females, respondents in the higher SES group, in more recent survey years, with no denture, and receiving SDC in all grades during childhood were associated with higher probability of seeking ADC regularly (P < 0.05). The interaction of SDC and age (P < 0.0001) was significant. The probabilities of seeking ADC were even higher among subjects with SDC in all grades and aged 45 years or older. Females, older age group, respondents in earlier survey years, not seeking ADC, lower SES group, and not receiving SDC in all grades were associated with higher probability of being edentulous (P < 0.05).

Conclusions

With the use of GEE, the potential clustering effect of birth cohorts in sequential cross-sectional oral health survey data could be appropriately considered. The success of Danish dental health policy was demonstrated by a continued increase of regular dental visiting habits and tooth retention in adults because school dental care was provided to Danes in their childhood.

Sex differences in grey matter atrophy patterns among AD and aMCI patients: results from ADNI

We used longitudinal magnetic resonance imaging (MRI) data to determine whether there are any gender differences in grey matter atrophy patterns over time in 197 individuals with probable Alzheimer's disease (AD) and 266 with amnestic mild cognitive impairment (aMCI), compared with 224 healthy controls participating in the Alzheimer's Disease Neuroimaging Initiative (ADNI). While previous research has differentiated probable AD and aMCI groups from controls in brain atrophy, it is unclear whether and how sex plays a role in patterns of change over time. Using regional volumetric maps, we fit longitudinal models to the grey matter data collected at repeated occasions, seeking differences in patterns of volume change over time by sex and diagnostic group in a voxel-wise analysis. Additionally, using a region-of-interest approach, we fit longitudinal models to the global volumetric data of predetermined brain regions to determine whether this more conventional approach is sufficient for determining sex and group differences in atrophy. Our longitudinal analyses revealed that, of the various grey matter regions investigated, males and females in the AD group and the aMCI group showed different patterns of decline over time compared to controls in the bilateral precuneus, bilateral caudate nucleus, right entorhinal gyrus, bilateral thalamus, bilateral middle temporal gyrus, left insula, and right amygdala. As one of the first investigation to model more than two time points of structural MRI data over time, our findings add insight into how AD and aMCI males and females differ from controls and from each other over time.

Electrocardiographic ST-segment depression and exposure to traffic-related aerosols in elderly subjects with coronary artery disease

BACKGROUND

Air pollutants have not been associated with ambulatory electrocardiographic evidence of ST-segment depression ≥ 1 mm (probable cardiac ischemia). We previously found that markers of primary (combustion-related) organic aerosols and gases were positively associated with circulating biomarkers of inflammation and ambulatory blood pressure in the present cohort panel study of elderly subjects with coronary artery disease.

OBJECTIVES

We specifically aimed to evaluate whether exposure markers of primary organic aerosols and ultrafine particles were more strongly associated with ST-segment depression of ≥ 1 mm than were secondary organic aerosols or PM2.5 (particulate matter with aerodynamic diameter ≤ 2.5 µm) mass.

METHODS

We evaluated relations of air pollutants to ambulatory electrocardiographic evidence of cardiac ischemia over 10 days in 38 subjects without ST depression on baseline electrocardiographs. Exposures were measured outdoors in retirement communities in the Los Angeles basin, including daily size-fractionated particle mass and hourly markers of primary and secondary organic aerosols and gases. Generalized estimating equations were used to estimate odds of hourly ST-segment depression (≥ 1 mm) from hourly air pollution exposures and to estimate relative rates of daily counts of ST-segment depression from daily average exposures, controlling for potential confounders.

RESULTS

We found significant positive associations of hourly ST-segment depression with markers of combustion-related aerosols and gases averaged 1-hr through 3-4 days, but not secondary (photochemically aged) organic aerosols or ozone. The odds ratio per interquartile increase in 2-day average primary organic carbon (5.2 µg/m3) was 15.4 (95% confidence interval, 3.5-68.2). Daily counts of ST-segment depression were consistently associated with primary combustion markers and 2-day average quasi-ultrafine particles < 0.25 µm.

CONCLUSIONS

Results suggest that exposure to quasi-ultrafine particles and combustion-related pollutants (predominantly from traffic) increase the risk of myocardial ischemia, coherent with our previous findings for systemic inflammation and blood pressure.

Effects of policies to restrict malt liquor sales on neighborhood crime

We examined the effectiveness of malt liquor sales restrictions adopted in 2005 in three liquor stores in a large Midwestern U.S. city. We hypothesized that the restrictions would be associated with crime reductions in adjacent neighborhoods. Using Poisson regression modeling, we compared crime rates two years prior to, and two years following policy adoption. Findings were mixed; malt liquor restrictions were associated with reductions in disorderly conduct citations, but increases in larceny/theft, beyond citywide trends. Limitations and implications of our study are discussed, and further research suggested. The study was funded by the U.S. Centers for Disease Control and Prevention.

Patient age, ethnicity, medical history, and risk factor profile, but not drug insurance coverage, predict successful attainment of glycemic targets: Time 2 Do More Quality Enhancement Research Initiative (T2DM QUERI)

OBJECTIVE

To identify factors in patients with type 2 diabetes and A1C >7.0% associated with attainment of A1C ≤ 7.0%.

RESEARCH DESIGN AND METHODS

We used a prospective registry of 5,280 Canadian patients in primary care settings enrolled in a 12-month glycemic pharmacotherapy optimization strategy based on national guidelines.

RESULTS

At close out, median A1C was 7.1% (vs. 7.8% at baseline) with 48% of subjects achieving A1C ≤ 7.0% (P < 0.0001). Older patients of Asian or black origin, those with longer diabetes duration, those with lower baseline A1C, BMI, LDL cholesterol, and blood pressure, and those on angiotensin receptor blockers and a lower number of antihyperglycemic agents, were more likely to achieve A1C ≤ 7.0% at some point during the study (all P < 0.0235). Access to private versus public drug coverage did not impact glycemic target realization.

CONCLUSIONS

Patient demography, cardiometabolic health, and ongoing pharmacotherapy, but not access to private drug insurance coverage, contribute to the care gap in type 2 diabetes.

Statistical Modeling of the Time Course of Tantrum Anger

Although anger is an important emotion that underlies much overt aggression at great social cost, little is known about how to quantify anger or to specify the relationship between anger and the overt behaviors that express it. This paper proposes a novel statistical model which provides both a metric for the intensity of anger and an approach to determining the quantitative relationship between anger intensity and the specific behaviors that it controls. From observed angry behaviors, we reconstruct the time course of the latent anger intensity and the linkage between anger intensity and the probability of each angry behavior. The data on which this analysis is based consist of observed tantrums had by 296 children in the Madison WI area during the period 1994-1996. For each tantrum, eight angry behaviors were recorded as occurring or not within each consecutive 30-second unit. So, the data can be characterized as a multivariate, binary, longitudinal (MBL) dataset with a latent variable (anger intensity) involved. Data such as these are common in biomedical, psychological, and other areas of the medical and social sciences. Thus, the proposed modeling approach has broad applications.

Healthcare use after screening for lung cancer

BACKGROUND

To evaluate the benefits of lung cancer screening, all effects of screening need to be considered. The aim of this study was to determine whether screening had an effect on healthcare use, specifically whether use increased for those with a false-positive or indeterminate screening result.

METHODS

Recruited were 400 individuals participating in a lung cancer screening study at the University of Pittsburgh. Self-reported outpatient healthcare use information was collected for the 6 months before, 0 to 6 months after, and 6 to 12 months after screening. The screening outcomes were negative, indeterminate, and suspicious. Repeated-measures Poisson regression models were used to examine changes in use over time and how changes over time varied among the screening outcome groups.

RESULTS

Approximately 58% of participants had a negative screening result, 36% had an indeterminate result, and approximately 6% had a suspicious result. The percentage of individuals who had any incidence of each type of outpatient use increased after screening, with the greatest increase noted for those with a suspicious screening result. Adjusted mean use significantly increased for nearly all types of use and for all 3 screening results categories in the 6 months after screening, but mostly declined to prescreening levels in the next 6 months.

CONCLUSIONS

Outpatient healthcare use was found to increase after screening for all individuals who were screened for lung cancer, regardless of the screening finding. The cost of the lung-related visits alone was substantial. Therefore, if lung cancer screening prevalence is increased, attendant follow-up healthcare costs are also likely to increase.

Correlation and studies of habitat selection: problem, red herring or opportunity?

With the advent of new technologies, animal locations are being collected at ever finer spatio-temporal scales. We review analytical methods for dealing with correlated data in the context of resource selection, including post hoc variance inflation techniques, 'two-stage' approaches based on models fit to each individual, generalized estimating equations and hierarchical mixed-effects models. These methods are applicable to a wide range of correlated data problems, but can be difficult to apply and remain especially challenging for use-availability sampling designs because the correlation structure for combinations of used and available points are not likely to follow common parametric forms. We also review emerging approaches to studying habitat selection that use fine-scale temporal data to arrive at biologically based definitions of available habitat, while naturally accounting for autocorrelation by modelling animal movement between telemetry locations. Sophisticated analyses that explicitly model correlation rather than consider it a nuisance, like mixed effects and state-space models, offer potentially novel insights into the process of resource selection, but additional work is needed to make them more generally applicable to large datasets based on the use-availability designs. Until then, variance inflation techniques and two-stage approaches should offer pragmatic and flexible approaches to modelling correlated data.