Goodness-of-fit for GEE: an example with mental health service utilization

Goodness-of-fit tests for modified Poisson regression possibly producing fitted values exceeding one in binary outcome analysis

Modified Poisson regression, which estimates the regression parameters in the log-binomial regression model using the Poisson quasi-likelihood estimating equation and robust variance, is a useful tool for estimating the adjusted risk and prevalence ratio in binary outcome analysis. Although several goodness-of-fit tests have been developed for other binary regressions, few goodness-of-fit tests are available for modified Poisson regression. In this study, we proposed several goodness-of-fit tests for modified Poisson regression, including the modified Hosmer-Lemeshow test with empirical variance, Tsiatis test, normalized Pearson chi-square tests with binomial variance and Poisson variance, and normalized residual sum of squares test. The original Hosmer-Lemeshow test and normalized Pearson chi-square test with binomial variance are inappropriate for the modified Poisson regression, which can produce a fitted value exceeding 1 owing to the unconstrained parameter space. A simulation study revealed that the normalized residual sum of squares test performed well regarding the type I error probability and the power for a wrong link function. We applied the proposed goodness-of-fit tests to the analysis of cross-sectional data of patients with cancer. We recommend the normalized residual sum of squares test as a goodness-of-fit test in the modified Poisson regression.

Factors associated with the outcome of root canal treatment-A cohort study conducted in a private practice

AIM

To investigate the association of various pre-, intra- and post-operative factors on root canal treatment outcome.

METHODOLOGY

In this cohort study, primary or secondary root canal treatment of mature permanent teeth was performed by a single endodontist in a private practice over 13 years, and followed 1-4 years after treatment. Treatment details and clinical and radiographic data were collected. The proportion of successfully treated teeth and roots based on strict radiographic (periapical index (PAI) ≤2) and clinical criteria (absence of pain, swelling or sinus tract) was estimated. To evaluate joint associations of prognostic factors and treatment success probability, 44 pre, intra- and post-operative factors were investigated using bivariate associations, and a multiple logistic regression model was fitted using Generalized Estimating Equations.

RESULTS

1259 teeth (2445 roots, 3149 canals) were assessed with a recall rate of 91%. The proportion of successfully treated teeth was 79.9% [95% confidence interval 77.7-82.1]. Eleven prognostic factors were identified that significantly reduced the odds ratio (OR) for treatment success at tooth level. Six were preoperative: injury history (OR = 0.05[0.01-0.24]), root PAI (OR = 0.29[0.20-0.42], 0.21[0.13-0.34] and 0.22[0.12-0.42] for PAI = 3, 4 and 5, respectively, against PAI = 1), lesion diameter (OR = 0.30[0.21-0.43] and 0.24[0.16-0.37] for diameters of 1-5 mm and ≥6 mm, respectively, against no lesion), tooth type (OR = 0.51[0.27-0.97] and OR = 0.45[0.24-0.83] for premolars and molars, respectively, against incisors or canines), tenderness to periapical palpation (OR = 0.64[0.43-0.94]) and two canals per root (OR = 0.67[0.54-0.83]). Four factors were intraoperative: root filling of unsatisfactory quality (OR = 0.18[0.08-0.40]) or extending beyond or shorter than 2 mm from the apex (OR = 0.44[0.26-0.75] and 0.62[0.40-0.97] respectively), resin sealer (OR = 0.58[0.39-0.87] against bioceramic sealer) and single visit treatment (OR = 0.40[0.21-0.75] against multiple visits). One factor was post-operative: defective coronal restoration (OR = 0.35[0.21-0.56]).

CONCLUSION

The following factors were associated with unsuccessful root canal treatment: (i) history of injury, apical periodontitis with increased severity (larger lesion, higher PAI, tenderness to periapical palpation), or complicated anatomic conditions (premolar or molar, two canals in a single root); (ii) technically suboptimal root filling (of unsatisfactory quality or not ending within 2 mm of radiographic apex) performed in a single-visit, or use of resin sealer instead of novel bioceramic sealer; (iii) suboptimal quality coronal restoration.

Analysing cluster randomised controlled trials using GLMM, GEE1, GEE2, and QIF: results from four case studies

BACKGROUND

Using four case studies, we aim to provide practical guidance and recommendations for the analysis of cluster randomised controlled trials.

METHODS

Four modelling approaches (Generalized Linear Mixed Models with parameters estimated by maximum likelihood/restricted maximum likelihood; Generalized Linear Models with parameters estimated by Generalized Estimating Equations (1st order or second order) and Quadratic Inference Function, for analysing correlated individual participant level outcomes in cluster randomised controlled trials were identified after we reviewed the literature. We systematically searched the online bibliography databases of MEDLINE, EMBASE, PsycINFO (via OVID), CINAHL (via EBSCO), and SCOPUS. We identified the above-mentioned four statistical analytical approaches and applied them to four case studies of cluster randomised controlled trials with the number of clusters ranging from 10 to 100, and individual participants ranging from 748 to 9,207. Results were obtained for both continuous and binary outcomes using R and SAS statistical packages.

RESULTS

The intracluster correlation coefficient (ICC) estimates for the case studies were less than 0.05 and are consistent with the observed ICC values commonly reported in primary care and community-based cluster randomised controlled trials. In most cases, the four methods produced similar results. However, in a few analyses, quadratic inference function produced different results compared to the generalized linear mixed model, first-order generalized estimating equations, and second-order generalized estimating equations, especially in trials with small to moderate numbers of clusters.

CONCLUSION

This paper demonstrates the analysis of cluster randomised controlled trials with four modelling approaches. The results obtained were similar in most cases, however, for trials with few clusters we do recommend that the quadratic inference function should be used with caution, and where possible a small sample correction should be used. The generalisability of our results is limited to studies with similar features to our case studies, for example, studies with a similar-sized ICC. It is important to conduct simulation studies to comprehensively evaluate the performance of the four modelling approaches.

Racial Inequities in Drug Tests Ordered by Clinicians for Pregnant People Who Disclose Prenatal Substance Use

OBJECTIVE

To measure racial inequities in drug testing among pregnant people during the first prenatal visit based on their drug use disclosure pattern.

METHODS

We used data from a cohort study of patient-clinician communication patterns regarding substance use in first prenatal visits from February 2011 to August 2014. We assessed racial differences (Black-White) in the receipt of urine toxicology testing, stratifying on patients' drug use disclosure to the clinician.

RESULTS

Among 341 study participants (205 Black [60.1%] and 136 White [39.9%] participants), 70 participants (33 Black [47.1%] and 37 White [52.9%] participants) disclosed drug use, and 271 participants (172 Black [63.5%] and 99 White [36.5%] participants) did not disclose drug use during their first obstetric visit. Of 70 participants who disclosed drug use, 50 (28 Black [56.0%] and 22 White [44.0%] White) had urine drug testing conducted. Black pregnant patients who disclosed drug use were more likely to be tested for drugs than their White counterparts in the adjusted regression analysis (adjusted odds ratio [aOR] 8.9, 95% CI 1.3-58.6). Among the 271 participants who did not disclose drug use, 38 (18 Black [47.4%] and 20 White [52.6%] participants) had urine drug testing conducted. For those who did not disclose drug use, the adjusted model showed no statistically significant differences in urine drug testing by patients' race (aOR 0.7, 95% CI 0.3-1.6).

CONCLUSION

When pregnant people disclosed drug use, clinicians were more likely to order urine drug testing for Black pregnant people compared with their White counterparts, suggesting clinician racial bias. Current practice patterns and protocols such as urine drug testing in pregnancy care deserve review to identify and mitigate areas of potential clinician discrimination.

Family Structure and Adolescent Mental Health Service Utilization During the COVID-19 Pandemic

PURPOSE

This study evaluated the relationship between sociodemographic factors including family structure and mental health service (MHS) utilization before and during the COVID-19 pandemic. We also investigated the moderation effects of the COVID-19 pandemic on MHS utilization.

METHODS

Our retrospective cohort study analyzed adolescents aged 12-17 years with a mental health diagnosis as identified in the electronic medical record enrolled in Kaiser Permanente Mid-Atlantic States in Maryland and Virginia, a comprehensive integrated health system. We used logistic regression models with an interaction term for the COVID-19 pandemic year to determine the relationship between family structure and adolescent MHS utilization ≥ one outpatient behavioral health visit within the measurement year, while adjusting for age, chronic medical condition (= physical illness lasting > 12 months), mental health condition, race, sex, and state of residence.

RESULTS

Among 5,420 adolescents, only those in two-parent households significantly increased MHS utilization during COVID-19 compared to the prepandemic year (McNemar's χ2 = 9.24, p < .01); however, family structure was not a significant predictor. Overall, the odds of adolescents using MHS were associated with a 12% increase during COVID-19 (odds ratio 1.12, 95% confidence interval [CI]: 1.02-1.22, p < .01). Higher odds of using MHS was associated with chronic medical condition (adjusted odds ratio = 1.15; 95% CI: 1.05-1.26, p < .01) and with White adolescents compared to all racial/ethnic minorities. The odds ratio of females using MHS compared to their male counterparts increased by 63% (ratio of adjusted odds ratio = 1.63; 95% CI: 1.39-1.91, p < .01) during the COVID-19 pandemic.

DISCUSSION

Individual-level demographic factors served as predictors of MHS utilization with effects moderated by COVID-19.

Factors Associated with Insomnia Symptoms in a Longitudinal Study among New York City Healthcare Workers during the COVID-19 Pandemic

Background: Few studies have examined the longer-term psychological impact of COVID-19 in healthcare workers (HCWs). Purpose: We examined the 10-week trajectory of insomnia symptoms in HCWs during the COVID-19 pandemic. Methods: HCWs completed a web-based survey at baseline (9 April–11 May 2020) and every 2 weeks for 10 weeks. The main outcome was the severity of insomnia symptoms in the past week. Multivariable-adjusted generalized estimating equation analyses examined factors associated with insomnia symptoms. Results: n = 230 completed surveys at baseline. n = 155, n = 130, n = 118, n = 95, and n = 89 completed follow-ups at weeks 2, 4, 6, 8, and 10, respectively. Prevalence of insomnia symptoms of at least moderate severity was 72.6% at baseline, and 63.2%, 44.6%, 40.7%, 34.7%, and 39.3% at weeks 2, 4, 6, 8, and 10, respectively. In multivariable analyses, factors significantly associated with increased odds of insomnia symptoms were younger age (OR: 0.98, 95% CI: 0.96–1.00), working in a COVID-facing environment (OR: 1.75, 95% CI: 1.15–2.67) and hours worked (OR: 1.16, 95% CI: 1.06–1.27). Conclusions: The initial high rates of insomnia symptoms improved as time passed from the peak of local COVID-19 cases but four out of ten HCWs still had moderate-to-severe insomnia symptoms ten weeks after baseline.

Weighted cumulative sum tests for random effect models with binary responses

Correlated binary responses are very commonly encountered in many disciplines like, for example, medical studies. The development of goodness-of-fit tests is essential for examining the adequacy of the fitted models. The objective of this article is to provide weighted modifications of cumulative sums or moving cumulative sums of residuals for testing goodness-of-fit of random effects logistic regression models. The proposed weights can be interpreted as the residuals of a weighted linear regression of an omitted covariate on the covariates already included in the fixed part of the model. These processes lead to supremum statistics whose null distribution is derived using simulation. Results from a simulation study suggest better performance of the weighted when compared to the unweighted supremum statistics. The proposed tests are illustrated using a real data example.

A GEE-type approach to untangle structural and random zeros in predictors

Count outcomes with excessive zeros are common in behavioral and social studies, and zero-inflated count models such as zero-inflated Poisson (ZIP) and zero-inflated Negative Binomial (ZINB) can be applied when such zero-inflated count data are used as response variable. However, when the zero-inflated count data are used as predictors, ignoring the difference of structural and random zeros can result in biased estimates. In this paper, a generalized estimating equation (GEE)-type mixture model is proposed to jointly model the response of interest and the zero-inflated count predictors. Simulation studies show that the proposed method performs well for practical settings and is more robust for model misspecification than the likelihood-based approach. A case study is also provided for illustration.

Comparison of Efficiency GEE and QIF Methods for Predicting Factors Affecting on Bipolar I Disorder Under Complete-case in a Longitudinal Studies

BACKGROUND

Mood variation in manic and depression phases during time is common in type I of Bipolar disorder. Analyzing recurrence require to the related statistical methods. In this paper, we compare the two methods of estimating the GEE and the QIF in recurrence data.

METHODS

In this study, data of 255 patients with Bipolar I disorder hospitalized during years of 2007-2011. Recurrence in Bipolar I disorder was as outcome. Patients' characteristics were gender, age of onset, recurrence history in first degree family, and economic status. Under simulation, percentage of missing were generated to vary and handled by complete-case(cc) strategy. Data were analyzed using GEE and QIF methods. Performance of the methods was assessed using Relative Efficiency.

RESULTS

QIF method had more efficiency than GEE method in the data with missing /without missing. Odds of recurrence in a first-degree family history was 30% more than those without a family history (p=0.009). Also, odds of recurrence in high/moderate level of economic status was 23% more than low level status (p=0.014).

CONCLUSION

QIF method was more appropriated for modeling recurrence during time with the structure of more correlation and low dropout rate in data. Family history and economic status were more affected recurrence in type I of Bipolar disorder.

Untangle the Structural and Random Zeros in Statistical Modelings

Count data with structural zeros are common in public health applications. There are considerable researches focusing on zero-inflated models such as zero-inflated Poisson (ZIP) and zero-inflated Negative Binomial (ZINB) models for such zero-inflated count data when used as response variable. However, when such variables are used as predictors, the difference between structural and random zeros is often ignored and may result in biased estimates. One remedy is to include an indicator of the structural zero in the model as a predictor if observed. However, structural zeros are often not observed in practice, in which case no statistical method is available to address the bias issue. This paper is aimed to fill this methodological gap by developing parametric methods to model zero-inflated count data when used as predictors based on the maximum likelihood approach. The response variable can be any type of data including continuous, binary, count or even zero-inflated count responses. Simulation studies are performed to assess the numerical performance of this new approach when sample size is small to moderate. A real data example is also used to demonstrate the application of this method.

Between-litter variation in developmental studies of hormones and behavior: Inflated false positives and diminished power

Developmental studies of hormones and behavior often include littermates-rodent siblings that share early-life experiences and genes. Due to between-litter variation (i.e., litter effects), the statistical assumption of independent observations is untenable. In two literatures-natural variation in maternal care and prenatal stress-entire litters are categorized based on maternal behavior or experimental condition. Here, we (1) review both literatures; (2) simulate false positive rates for commonly used statistical methods in each literature; and (3) characterize small sample performance of multilevel models (MLM) and generalized estimating equations (GEE). We found that the assumption of independence was routinely violated (>85%), false positives (α=0.05) exceeded nominal levels (up to 0.70), and power (1-β) rarely surpassed 0.80 (even for optimistic sample and effect sizes). Additionally, we show that MLMs and GEEs have adequate performance for common research designs. We discuss implications for the extant literature, the field of behavioral neuroendocrinology, and provide recommendations.

Provider-Induced Demand in the Treatment of Carotid Artery Stenosis: Variation in Treatment Decisions Between Private Sector Fee-for-Service vs Salary-Based Military Physicians

Importance

Although many factors influence the management of carotid artery stenosis, it is not well understood whether a preference toward procedural management exists when procedural volume and physician compensation are linked in the fee-for-service environment.

Objective

To explore evidence for provider-induced demand in the management of carotid artery stenosis.

Design, Setting, and Participants

The Department of Defense Military Health System Data Repository was queried for individuals diagnosed with carotid artery stenosis between October 1, 2006, and September 30, 2010. A hierarchical multivariable model evaluated the association of the treatment system (fee-for-service physicians in the private sector vs salary-based military physicians) with the odds of procedural intervention (carotid endarterectomy or carotid artery stenting) compared with medical management. Subanalysis was performed by symptom status at the time of presentation. The association of treatment system and of management strategy with clinical outcomes, including stroke and death, was also evaluated. Data analysis was conducted from August 15, 2015, to August 2, 2016.

Main Outcomes and Measures

The odds of procedural intervention based on treatment system was the primary outcome used to indicate the presence and effect of provider-induced demand.

Results

Of 10 579 individuals with a diagnosis of carotid artery stenosis (4615 women and 5964 men; mean [SD] age, 65.6 [11.4] years), 1307 (12.4%) underwent at least 1 procedure. After adjusting for demographic and clinical factors, the odds of undergoing procedural management were significantly higher for patients in the fee-for-service system compared with those in the salary-based setting (odds ratio, 1.629; 95% CI, 1.285-2.063; P < .001). This finding remained true when patients were stratified by symptom status at presentation (symptomatic: odds ratio, 2.074; 95% CI, 1.302-3.303; P = .002; and asymptomatic: odds ratio, 1.534; 95% CI, 1.186-1.984; P = .001).

Conclusions and Relevance

Individuals treated in a fee-for-service system were significantly more likely to undergo procedural management for carotid stenosis compared with those in the salary-based setting. These findings remained consistent for individuals with and without symptomatic disease.

Standardized mean differences in individually-randomized and cluster-randomized trials, with applications to meta-analysis

The magnitude of the effect of an intervention on a quantitative outcome may be expressed as a standardized mean difference by dividing the difference in means by the standard deviation of the outcome. This is useful to compare outcomes measured using different scales, especially in meta-analysis. However, uncertainty about the standard deviation leads to complicated formulae to avoid bias and to compute the correct standard error. We review approximate and exact formulae and argue for the use of the exact formulae. We then extend the formulae to cluster-randomized trials, and show how the calculations may be implemented using published results. We also describe methods for estimating the standard deviation. Various pitfalls are identified which can lead to major errors especially in the cluster-randomized setting.

Using Generalized Estimating Equations for Longitudinal Data Analysis

The generalized estimating equation (GEE) approach of Zeger and Liang facilitates analysis of data collected in longitudinal, nested, or repeated measures designs. GEEs use the generalized linear model to estimate more efficient and unbiased regression parameters relative to ordinary least squares regression in part because they permit specification of a working correlation matrix that accounts for the form of within-subject correlation of responses on dependent variables of many different distributions, including normal, binomial, and Poisson. The author briefly explains the theory behind GEEs and their beneficial statistical properties and limitations and compares GEEs to suboptimal approaches for analyzing longitudinal data through use of two examples. The first demonstration applies GEEs to the analysis of data from a longitudinal lab study with a counted response variable; the second demonstration applies GEEs to analysis of data with a normally distributed response variable from subjects nested within branch offices ofan organization.

Nursing Home Deficiency Citations for Medication Use

Deficiency citations for medication use in nursing homes, including those for psychoactive drug use, are examined. The variables of interest include eight structural and market factors. Data primarily came from the 1997 through 2003 Online Survey, Certification and Recording data and the 2004 Area Resource File. Multivariate logistic regression analyses were used with generalized estimating equations and multinomial logistic regression models with Huber-White robust estimation. Smaller nursing homes and high Medicaid reimbursement rates were consistently significantly associated with fewer deficiency citations in general and fewer repeat deficiency citations. For the other structural and market factors, varying results were identified depending on whether the deficiency citation was specifically for psychoactive drugs, medication errors, or medication administration. Relatively few facilities received psychoactive-specific deficiency citations, whereas numerous facilities were identified with medication error deficiency citations and medication administration deficiency citations. In addition, a relatively large number of facilities received these same citations repeatedly.

Occupational determinants of methicillin-resistant Staphylococcus aureus colonization among healthcare workers: a longitudinal study in a rehabilitation center

BACKGROUND Staphylococcus aureus carriage among healthcare workers (HCWs) is a concern in hospital settings, where it may provide a reservoir for later infections in both patients and staff. Earlier studies have shown that the prevalence of methicillin-resistant Staphylococcus aureus (MRSA) carriage in HCWs is highly variable, depending notably on location, hospital department type, MRSA prevalence among patients, and type of contacts with patients. However, MRSA incidence in HCWs and its occupational determinants have seldom been studied. METHODS A prospective, observational cohort study was conducted between May and October 2009 in a French rehabilitation center hospital. HCWs and patients were screened weekly for S. aureus nasal carriage. Methicillin-susceptible S. aureus and MRSA prevalence and incidence were estimated and factors associated with MRSA acquisition were identified using generalized estimating equation regression methods. RESULTS Among 343 HCWs included in the analysis, the average prevalence was 27% (95% CI, 24%-29%) for methicillin-susceptible S. aureus and 10% (8%-11%) for MRSA. We observed 129 MRSA colonization events. According to the multivariable analysis, high MRSA prevalence level among patients and HCW occupation were significantly associated with MRSA acquisition in HCWs, with assistant nurses being more at risk than nurses (odds ratio, 2.2; 95% CI, 1.4-3.6). CONCLUSIONS Our findings may help further our understanding of the transmission dynamics of MRSA carriage acquisition in HCWs, suggesting that it is notably driven by carriage among patients and by the type of contact with patients.

Predictors of prescribed medication use for depression, anxiety, stress, and sleep problems in mid-aged Australian women

OBJECTIVE

The study examined prevalence of self-reported use of medication recommended or prescribed by a doctor for depression, anxiety, stress, and sleep problems; and modelled baseline factors that predicted use over 3 years for each condition.

METHODS

Analyses were undertaken on the 2001 and 2004 surveys of mid-aged women in the Australian Longitudinal Study on Women's Health. Dependent variables were self-reported use in past 4 weeks of medications recommended or prescribed by a doctor for depression, anxiety, stress, or sleep problems in 2001 and 2004. Generalized Estimating Equations (GEE) were used to predict medication use for each condition over 3 years.

RESULTS

Prevalence of prescribed medication use (2001, 2004) for each condition was depression (7.2, 8.9 %), anxiety (7.4, 9.0 %), stress (4.8, 5.7 %), and sleep problems (8.7, 9.5 %). Multivariable analyses revealed that odds of medication use across 3 years in all four conditions were higher for women with poorer mental and physical health, using hormone replacement therapy (HRT), or having seen a counsellor; and increased over time for depression, anxiety, and stress models. Medication use for depression was also higher for overweight/obese women, ex-smokers, and unmarried. Medication use for anxiety was higher for unmarried and non-working/low occupational women. Medication use for stress was higher for non-working women. Additional predictors of medication for sleep were surgical menopause, and area of residence.

CONCLUSIONS

Self-reported use of prescribed medication for four mental health conditions is increased over time after controlling for mental and physical health and other variables. Research needs to explore decision-making processes influencing differential rates of psychoactive medication use and their relationship with health outcomes.

Comparison of the abiotic preferences of macroinvertebrates in tropical river basins

We assessed and compared abiotic preferences of aquatic macroinvertebrates in three river basins located in Ecuador, Ethiopia and Vietnam. Upon using logistic regression models we analyzed the relationship between the probability of occurrence of five macroinvertebrate families, ranging from pollution tolerant to pollution sensitive, (Chironomidae, Baetidae, Hydroptilidae, Libellulidae and Leptophlebiidae) and physical-chemical water quality conditions. Within the investigated physical-chemical ranges, nine out of twenty-five interaction effects were significant. Our analyses suggested river basin dependent associations between the macroinvertebrate families and the corresponding physical-chemical conditions. It was found that pollution tolerant families showed no clear abiotic preference and occurred at most sampling locations, i.e. Chironomidae were present in 91%, 84% and 93% of the samples taken in Ecuador, Ethiopia and Vietnam. Pollution sensitive families were strongly associated with dissolved oxygen and stream velocity, e.g. Leptophlebiidae were only present in 48%, 2% and 18% of the samples in Ecuador, Ethiopia and Vietnam. Despite some limitations in the study design, we concluded that associations between macroinvertebrates and abiotic conditions can be river basin-specific and hence are not automatically transferable across river basins in the tropics.

Sex offender registration and notification policy increases juvenile plea bargains

The aim of this study was to test the hypothesis that South Carolina's sex offender registration and notification policy influenced juvenile sex offense case plea bargains. Two types of plea bargains were examined: initial sex offense charges amended to nonsex offense charges and amended to lower severity charges. Comparison analyses were conducted with juvenile assault and robbery offense cases. Archival data on cases involving 19,215 male youth charged with sex, assault, and/or robbery offenses between 1990 and 2004 informed analyses. Of these youth, 2,991 were charged with one or more sex offense, 16,091 were charged with one or more assault offense, and 2,036 were charged with at one or more robbery offense. Generalized estimating equations (GEE) were used to model changes in the probabilities of plea bargain outcomes across three time intervals: before policy implementation (1990 to 1994), after initial policy implementation (1995 to 1998), and after implementation of a revised policy that included online registration requirements (1999 to 2004). Results indicate significant increases in the probability of plea bargains for sex offense cases across subsequent time periods, supporting the hypothesis that South Carolina's initial and revised registration and notification policies were associated with significant increases the likelihood of plea bargains to different types of charges and to lower severity charges. Results were either nonsignificant or of much lower magnitude for the comparison assault and robbery analyses. Suggestions for revising South Carolina and national registration and notification policies are discussed.

Intraprocedural C-arm dual-phase cone-beam CT: can it be used to predict short-term response to TACE with drug-eluting beads in patients with hepatocellular carcinoma?

PURPOSE

To investigate whether C-arm dual-phase cone-beam computed tomography (CT) performed during transcatheter arterial chemoembolization (TACE) with doxorubicin-eluting beads can help predict tumor response at 1-month follow-up in patients with hepatocellular carcinoma (HCC).

MATERIALS AND METHODS

This prospective study was compliant with HIPAA and approved by the institutional review board and animal care and use committee. Analysis was performed retrospectively on 50 targeted HCC lesions in 29 patients (16 men, 13 women; mean age, 61.9 years ± 10.7) treated with TACE with drug-eluting beads. Magnetic resonance (MR) imaging was performed at baseline and 1 month after TACE. Dual-phase cone-beam CT was performed before and after TACE. Tumor enhancement at dual-phase cone-beam CT in early arterial and delayed venous phases was assessed retrospectively with blinding to MR findings. Tumor response at MR imaging was assessed according to European Association for the Study of the Liver (EASL) guidelines. Two patients were excluded from analysis because dual-phase cone-beam CT scans were not interpretable. Logistic regression models for correlated data were used to compare changes in tumor enhancement between modalities. The radiation dose with dual-phase cone-beam CT was measured in one pig.

RESULTS

At 1-month MR imaging follow-up, complete and/or partial tumor response was seen in 74% and 76% of lesions in the arterial and venous phases, respectively. Paired t tests used to compare images obtained before and after TACE showed a significant reduction in tumor enhancement with both modalities (P < .0001). The decrease in tumor enhancement seen with dual-phase cone-beam CT after TACE showed a linear correlation with MR findings. Estimated correlation coefficients were excellent for first (R = 0.89) and second (R = 0.82) phases. A significant relationship between tumor enhancement at cone-beam CT after TACE and complete and/or partial tumor response at MR imaging was found for arterial (odds ratio, 0.95; 95% confidence interval [CI]: 0.91, 0.99; P = .023) and venous (odds ratio, 0.96; 95% CI: 0.93, 0.99; P = .035) phases with the multivariate logistic regression model. Radiation dose for two dual-phase cone-beam CT scans was 3.08 mSv.

CONCLUSION

Intraprocedural C-arm dual-phase cone-beam CT can be used immediately after TACE with doxorubicin-eluting beads to predict HCC tumor response at 1-month MR imaging follow-up.

Heterogeneity of inter-rater reliabilities of grant peer reviews and its determinants: a general estimating equations approach

Background

One of the most important weaknesses of the peer review process is that different reviewers’ ratings of the same grant proposal typically differ. Studies on the inter-rater reliability of peer reviews mostly report only average values across all submitted proposals. But inter-rater reliabilities can vary depending on the scientific discipline or the requested grant sum, for instance.

Goal

Taking the Austrian Science Fund (FWF) as an example, we aimed to investigate empirically the heterogeneity of inter-rater reliabilities (intraclass correlation) and its determinants.

Methods

The data consisted of N = 8,329 proposals with N = 23,414 overall ratings by reviewers, which were statistically analyzed using the generalized estimating equations approach (GEE).

Results

We found an overall intraclass correlation (ICC) of reviewer? ratings of ρ = .259 with a 95% confidence interval of [.249,.279]. In humanities the ICCs were statistically significantly higher than in all other research areas except technical sciences. The ICC in biosciences deviated statistically significantly from the average ICC. Other factors (besides the research areas), such as the grant sum requested, had negligible influence on the ICC.

Conclusions

Especially in biosciences, the number of reviewers of each proposal should be increased so as to increase the ICC.

Individual participant data meta-analysis of mechanical workplace risk factors and low back pain

OBJECTIVES

We used individual participant data from multiple studies to conduct a comprehensive meta-analysis of mechanical exposures in the workplace and low back pain.

METHODS

We conducted a systematic literature search and contacted an author of each study to request their individual participant data. Because outcome definitions and exposure measures were not uniform across studies, we conducted 2 substudies: (1) to identify sets of outcome definitions that could be combined in a meta-analysis and (2) to develop methods to translate mechanical exposure onto a common metric. We used generalized estimating equation regression to analyze the data.

RESULTS

The odds ratios (ORs) for posture exposures ranged from 1.1 to 2.0. Force exposure ORs ranged from 1.4 to 2.1. The magnitudes of the ORs differed according to the definition of low back pain, and heterogeneity was associated with both study-level and individual-level characteristics.

CONCLUSIONS

We found small to moderate ORs for the association of mechanical exposures and low back pain, although the relationships were complex. The presence of individual-level OR modifiers in such an area can be best understood by conducting a meta-analysis of individual participant data.

A shared care model pilot for palliative home care in a rural area: impact on symptoms, distress, and place of death

CONTEXT

Shared care models integrating family physician services with interdisciplinary palliative care specialist teams are critical to improve access to quality palliative home care and address multiple domains of end-of-life issues and needs.

OBJECTIVES

To examine the impact of a shared care pilot program on the primary outcomes of symptom severity and emotional distress (patient and family separately) over time and, secondarily, the concordance between patient preferences and place of death.

METHODS

An inception cohort of patients (n = 95) with advanced, progressive disease, expected to die within six months, were recruited from three rural family physician group practices (21 physicians) and followed prospectively until death or pilot end. Serial measurement of symptoms, emotional distress (patient and family), and preferences for place of death was performed, with analysis of changes in distress outcomes assessed using t-tests and general linear models.

RESULTS

Symptoms trended toward improvement, with a significant reduction in anxiety from baseline to 14 days noted. Symptom and emotional distress were maintained below high severity (7-10), and a high rate of home death compared with population norms was observed.

CONCLUSION

Future controlled studies are needed to examine outcomes for shared care models with comparison groups. Shared care models build on family physician capacity and as such are promising in the development of palliative home care programs to improve access to quality palliative home care and foster health system integration.

Modic changes and associated features in Southern European chronic low back pain patients

BACKGROUND CONTEXT

Conflicting reports exist regarding the prevalence of Modic changes among low back pain (LBP) patients and factors associated with their existence.

PURPOSE

To assess the prevalence of Modic changes and other findings on lumbar magnetic resonance imaging (MRI) among Spanish adult chronic LBP patients and the patient characteristics and radiological findings associated with Modic changes.

STUDY DESIGN

A cross-sectional imaging study among chronic LBP patients.

PATIENT SAMPLE

Four hundred eighty-seven patients (263 women and 224 men) undergoing lumbar spine MRI examination for chronic LBP.

OUTCOME MEASURES

Gender, age, body mass index (BMI), lifetime smoking exposure, degree of physical activity, and image features (disc degeneration, type and extension of Modic changes, disc contour, annular tears, spinal stenosis, and spondylolisthesis).

METHODS

Ten radiologists from six hospitals across six cities in Spain consecutively recruited adult patients in whom lumbar MRI had been prescribed for LBP lasting ≥3 months. Patients' characteristics and imaging findings were assessed through previously validated instruments. A multivariate logistic regression model was developed to assess the features associated with Modic changes.

RESULTS

Modic changes were found in 81% of the patients. The most common was Type II (51.3%), affecting only the end plate. Variables associated with Type I changes were disc contour abnormalities, spondylolisthesis, and disc degeneration. The same variables were associated with a higher risk of Type II or any type of Modic changes, as well as being male, and having a higher BMI.

CONCLUSIONS

Modic changes are found in 81% (95% confidence interval, 77-85) of adult Spanish patients in whom an MRI is prescribed for chronic LBP. Modic changes are more likely to be found in males with a high BMI, who also show disc contour abnormalities, spondylolisthesis, or disc degeneration.

Predictors of initial stability of orthodontic miniscrew implants

The purpose of this retrospective study was to elucidate potential confounding factors affecting initial stability of miniscrews inserted to enhance orthodontic anchorage. Four hundred and seven miniscrews inserted in 168 patients treated by 17 orthodontic residents were analysed in a consecutive chart review. The outcome variable was the stability of the miniscrew, measured as a dichotomous variable, 0 if the miniscrew loosened during a 1 week period after insertion to the time of orthodontic force application and a value of 1 otherwise. Potential confounding variables examined were gender, age, jaw, insertion site, tissue type, length and diameter of the miniscrew, and number of previous insertions. Generalized estimating equations (GEE) methods were used to estimate the influence of each factor on stability for the correlated binary outcomes of each patient. A weighted analysis for the GEE approach was also performed for the convergence calculation of the estimation procedure due to a value of 0 in one of the cells. Crude odds ratio (cOR) and adjusted odds ratio (aOR) and their 95 per cent confidence intervals (CI) were calculated for this purpose. The overall success rate after 1 week was 93.1 per cent (379/407). The screws inserted by more experienced clinicians (more than 20 miniscrews) were found to have approximately a 3.6-fold higher success rate of initial stability compared with those inserted by less experienced clinicians after adjusting for the insertion site (aOR = 3.63, P = 0.015). The results of the present study suggest that the initial stability depends on insertion site and clinician experience.

Periodontal disease progression and glycaemic control among Gullah African Americans with type-2 diabetes

AIM

To evaluate associations between glycaemic control and periodontitis progression among Gullah African Americans with type-2 diabetes mellitus (T2DM).

MATERIALS AND METHODS

From an ongoing clinical trial among T2DM Gullah, we extracted a cohort previously in a cross-sectional study (N=88). Time from baseline (previous study) to follow-up (trial enrollment, before treatment interventions) ranged 1.93-4.08 years [mean=2.99, standard deviation (SD)=0.36]. We evaluated tooth site-level periodontitis progression [clinical attachment loss (CAL) worsening of > or =2 mm, periodontal probing depth (PPD) increases of > or =2 mm and bleeding on probing (BOP) from none to present] by glycaemic control status (well-controlled=HbA(1c)<7%, poorly-controlled=HbA(1c)> or =7%) using multivariable generalized estimating equations logistic regression, nesting tooth sites/person.

RESULTS

Poorly-controlled T2DM (68.18%) was more prevalent than well-controlled T2DM (31.82%). Proportions of tooth sites/person with CAL progression between baseline and follow-up ranged 0.00-0.59 (mean=0.12, SD=0.12), while PPD and BOP progression ranged 0.00-0.44 (mean=0.09, SD=0.11) and 0.00-0.96 (mean=0.24, SD=0.18), respectively. Site-level PPD at baseline was a significant effect modifier of associations between poorly-controlled T2DM and site-level CAL and PPD progression [adjusted odds ratios (OR) according to poorly-controlled T2DM among PPD at baseline=3, 5 and 7 mm, respectively: CAL progression=1.93, 2.64, and 3.62, PPD progression=1.98, 2.76, and 3.84; p<0.05 for all]. Odds of site-level BOP progression were increased (OR=1.24) for poorly-controlled T2DM, yet the results were not significant (p=0.32).

CONCLUSIONS

These findings from a distinct, homogenous population further support the clinical relevance of identifying patients with poor glycaemic control and periodontitis, particularly among those with disparities for both diseases.

Patient retention and adherence to antiretrovirals in a large antiretroviral therapy program in Nigeria: a longitudinal analysis for risk factors

BACKGROUND

Substantial resources and patient commitment are required to successfully scale-up antiretroviral therapy (ART) and provide appropriate HIV management in resource-limited settings. We used pharmacy refill records to evaluate risk factors for loss to follow-up (LTFU) and non-adherence to ART in a large treatment cohort in Nigeria.

METHODS AND FINDINGS

We reviewed clinic records of adult patients initiating ART between March 2005 and July 2006 at five health facilities. Patients were classified as LTFU if they did not return >60 days from their expected visit. Pharmacy refill rates were calculated and used to assess non-adherence. We identified risk factors associated with LTFU and non-adherence using Cox and Generalized Estimating Equation (GEE) regressions, respectively. Of 5,760 patients initiating ART, 26% were LTFU. Female gender (p < 0.001), post-secondary education (p = 0.03), and initiating treatment with zidovudine-containing (p = 0.004) or tenofovir-containing (p = 0.05) regimens were associated with decreased risk of LTFU, while patients with only primary education (p = 0.02) and those with baseline CD4 counts (cell/ml(3)) >350 and <100 were at a higher risk of LTFU compared to patients with baseline CD4 counts of 100-200. The adjusted GEE analysis showed that patients aged <35 years (p = 0.005), who traveled for >2 hours to the clinic (p = 0.03), had total ART duration of >6 months (p<0.001), and CD4 counts >200 at ART initiation were at a higher risk of non-adherence. Patients who disclosed their HIV status to spouse/family (p = 0.01) and were treated with tenofovir-containing regimens (p < or = 0.001) were more likely to be adherent.

CONCLUSIONS

These findings formed the basis for implementing multiple pre-treatment visit preparation that promote disclosure and active community outreaching to support retention and adherence. Expansion of treatment access points of care to communities to diminish travel time may have a positive impact on adherence.

Quadratic inference functions in marginal models for longitudinal data

The quadratic inference function (QIF) is a new statistical methodology developed for the estimation and inference in longitudinal data analysis using marginal models. This method is an alternative to the popular generalized estimating equations approach, and it has several useful properties such as robustness, a goodness-of-fit test and model selection. This paper presents an introductory review of the QIF, with a strong emphasis on its applications. In particular, a recently developed SAS MACRO QIF is illustrated in this paper to obtain numerical results.

Family physicians' personal and practice characteristics that are associated with improved utilization of bone mineral density testing and osteoporosis medication prescribing

Family physicians' personal and practice characteristics may influence how osteoporosis is managed. Thus, we evaluated the impact of family physicians' personal and practice characteristics on the appropriate use of bone mineral density testing and osteoporosis therapy. The physician questionnaire assessed 13 personal and practice characteristics of the physicians. The patient questionnaire was used to collect data to ascertain how family physicians managed osteoporosis. A total of 225 family physicians from 7 provinces across Canada completed both the physician and patient questionnaires. The family physicians evaluated a total of 5601 patients. The generalized estimating equations technique was utilized to model the associations between family physicians' personal and practice characteristics and appropriate use of bone mineral density testing and osteoporosis therapy. Odds ratios (OR) and corresponding 95% confidence intervals (CI) are reported. Findings indicated that female family physicians have higher odds of administering appropriate bone density testing compared to male family physicians (OR: 1.28; 95% CI: 1.05, 1.55), and that physicians who have hospital privileges (OR: 0.77; 95% CI: 0.62, 0.97) and who graduated more recently from medical school (OR: 0.87; 95% CI: 0.77, 0.99) have lower odds of administering appropriate bone mineral density tests. Physicians who use electronic health records have higher odds of administering appropriate therapy (OR: 1.30; 95% CI: 1.06, 1.59) as compared to physicians who do not use them. Several family physicians' personal and practice characteristics are associated with appropriate utilization of bone mineral density testing and therapy. The education of both clinicians and policy makers regarding these new insights may translate to enhanced individual practices and an improved overall health care system to optimize the environment for managing osteoporosis.

The utilization of appropriate osteoporosis medications improves following a multifaceted educational intervention: the Canadian quality circle project (CQC)

BACKGROUND

Osteoporosis is a serious but treatable condition. However, appropriate therapy utilization of the disease remains suboptimal. Thus, the objective of the study was to change physicians' therapy administration behavior in accordance with the Osteoporosis Canada 2002 guidelines.

METHODS

The Project was a two year cohort study that consisted of five Quality Circle (QC) phases that included: 1) Training & Baseline Data Collection, 2) First Educational Intervention & First Follow-Up Data Collection 3) First Strategy Implementation Session, 4) Final Educational Intervention & Final Follow-up Data Collection, and 5) Final Strategy Implementation Session. A total of 340 family physicians formed 34 QCs and participated in the study. Physicians evaluated a total of 8376, 7354 and 3673 randomly selected patient charts at baseline, follow-up #1 and the final follow-up, respectively. Patients were divided into three groups; the high-risk, low-risk, and low-risk without fracture groups. The generalized estimating equations technique was utilized to model the change over time of whether physicians

RESULTS

The odds of appropriate therapy was 1.29 (95% CI: 1.13, 1.46), and 1.41 (95% CI: 1.20, 1.66) in the high risk group, 1.15 (95% CI: 0.97, 1.36), and 1.16 (95% CI: 0.93, 1.44) in the low risk group, and 1.20 (95% CI: 1.01, 1.43), and 1.23 (95% CI: 0.97, 1.55) in the low risk group without fractures at follow-up #1 and the final follow-up, respectively.

CONCLUSION

QCs methodology was successful in increasing physicians' appropriate use of osteoporosis medications in accordance with Osteoporosis Canada guidelines.

Adherence to antiretroviral medications and medical care in HIV-infected adults diagnosed with mental and substance abuse disorders

This paper examines factors associated with adherence to antiretroviral medications (ARVs) in an HIV-infected population at high risk for non-adherence: individuals living with psychiatric and substance abuse disorders. Data were examined from baseline interviews of a multisite cohort intervention study of 1138 HIV-infected adults with both a psychiatric and substance abuse disorder (based on a structured psychiatric research interview using DSM-IV criteria). The baseline interview documented mental illness and substance use in the past year, mental illness and substance abuse severity, demographics, service utilization in the past three months, general health and HIV-related conditions, self-reported spirituality and self-reported ARV medication use. Among the participants, 62% were prescribed ARVs at baseline (n = 542) and 45% of those on ARVs reported skipping medications in the past three days. Reports of non-adherence were significantly associated with having a detectable viral load (p<.01). The factors associated with non-adherence were current drug and alcohol abuse, increased psychological distress, less attendance at medical appointments, non-adherence to psychiatric medications and lower self-reported spirituality. Increased psychological distress was significantly associated with non-adherence, independent of substance abuse (p<.05). The data suggest that both mental illness and substance use must be addressed in HIV-infected adults living with these co-morbid illnesses to improve adherence to ARVs.

A nonparametric smoothing method for assessing GEE models with longitudinal binary data

Studies involving longitudinal binary responses are widely applied in the health and biomedical sciences research and frequently analyzed by generalized estimating equations (GEE) method. This article proposes an alternative goodness-of-fit test based on the nonparametric smoothing approach for assessing the adequacy of GEE fitted models, which can be regarded as an extension of the goodness-of-fit test of le Cessie and van Houwelingen (Biometrics 1991; 47:1267-1282). The expectation and approximate variance of the proposed test statistic are derived. The asymptotic distribution of the proposed test statistic in terms of a scaled chi-squared distribution and the power performance of the proposed test are discussed by simulation studies. The testing procedure is demonstrated by two real data.

Optimizing care in osteoporosis: the Canadian quality circle project

BACKGROUND

While the Osteoporosis Canada 2002 Canadian guidelines provided evidence based strategies in preventing, diagnosing, and managing this condition, publication and distribution of guidelines have not, in and of themselves, been shown to alter physicians clinical approaches. We hypothesize that primary care physicians enrolled in the Quality Circle project would change their patient management of osteoporosis in terms of awareness of osteoporosis risk factors and bone mineral density testing in accordance with the guidelines.

METHODS

The project consisted of five Quality Circle phases that included: 1) Training & Baseline Data Collection, 2) First Educational Intervention & First Follow-Up Data Collection 3) First Strategy Implementation Session, 4) Final Educational Intervention & Final Follow-up Data Collection, and 5) Final Strategy Implementation Session. A total of 340 circle members formed 34 quality circles and participated in the study. The generalized estimating equations approach was used to model physician awareness of risk factors for osteoporosis and appropriate utilization of bone mineral density testing pre and post educational intervention (first year of the study). Odds ratios (OR) and 95% confidence intervals (95% CI) were calculated.

RESULTS

After the 1st year of the study, physicians' certainty of their patients' risk factor status increased. Certainty varied from an OR of 1.4 (95% CI: 1.1, 1.8) for prior vertebral fracture status to 6.3 (95% CI: 2.3, 17.9) for prior hip fracture status. Furthermore, bone mineral density testing increased in high risk as compared with low risk patients (OR: 1.4; 95% CI: 1.2, 1.7).

CONCLUSION

Quality Circle methodology was successful in increasing both physicians' awareness of osteoporosis risk factors and appropriate bone mineral density testing in accordance with the 2002 Canadian guidelines.