Model-checking techniques based on cumulative residuals

Impact of sickle cell disease and thalassemias in infants on birth outcomes

OBJECTIVE

The contribution of sickle cell disease (SCD) and other common thalassemias in infants to adverse birth outcomes is under-studied. We therefore sought to compare adverse birth outcomes in infants with and without hemoglobinopathy.

STUDY DESIGN

Retrospective cohort study utilizing a population-based dataset from Florida (1998-2007, n=1,564,038). The primary outcomes were low birthweight (LBW), very low birthweight (VLBW), preterm birth (PTB), very preterm birth (VPTB) and small for gestational age (SGA). We used propensity scores to match infants with hemoglobinopathy to those without hemoglobinopathy on selected variables. To approximate relative risks, we generated adjusted odds ratios (AOR) and 95% confidence intervals (CI) from logistic regression models and accounted for the matched design using generalized estimating equations framework.

RESULTS

Infants with SCD or thalassemia had a heightened risk for LBW (AOR=1.58, 95% CI: 1.29-1.93), VLBW (AOR=3.01, 95% CI: 2.12-4.25), PTB (AOR=1.36, 95% CI: 1.12-1.65), VPTB (AOR=2.70, 95% CI: 1.93-3.78), and neurological conditions (AOR=2.04, 95% CI: 1.48-2.81) compared to infants without hemoglobinopathy.

CONCLUSION

Infants with SCD or thalassemia experience considerably higher risks for multiple infant morbidities. Our findings are potentially important in prenatal counseling, as well as for targeted care of affected pregnancies in the prenatal period.

Risk of colorectal high-grade dysplasia and cancer in a prospective observational cohort of patients with inflammatory bowel disease

BACKGROUND & AIMS

There is an unclear risk of colonic high-grade dysplasia (HGD) and colorectal cancer (CRC) among patients with inflammatory bowel disease (IBD) treated with immunosuppressants. We analyzed data on CRC development among patients with IBD enrolled in the observational cohort Cancers et Surrisque Associé aux Maladies Inflammatoires Intestinales En France (CESAME).

METHODS

We followed and collected data from 19,486 patients with IBD (60.3% with Crohn's disease, 30.1% receiving thiopurine therapy) enrolled in CESAME from May 2004 and June 2005, and followed them until December 2007. When the study began, 2841 patients (14.6%) were characterized as having long-standing extensive colitis (ie, >10 years and involving ≥50% of the colon). Early lesions (HGD and CRC) were defined as those diagnosed within 10 years after diagnosis of IBD.

RESULTS

Thirty-seven patients developed CRC during the follow-up period, and 20 developed colorectal HGD. The standardized incidence ratios of CRC were 2.2 for all IBD patients (95% confidence interval [CI]: 1.5-3.0; P < .0001), 7.0 for patients with long-standing extensive colitis (95% CI: 4.4-10.5; P < .001), and 1.1 for patients without long-standing extensive colitis (95% CI: 0.6-1.8; P = .84). Among patients with long-standing extensive colitis, the multivariate adjusted hazard ratio for colorectal HGD and cancer was 0.28 for those who received thiopurines compared with those who never received thiopurine therapy (95% CI: 0.1-0.9; P = .03). Twenty-two patients developed early lesions; 7 of these were related to IBD, based on histologic analysis.

CONCLUSIONS

Patients with IBD and long-standing extensive colitis are at increased risk for CRC, although the risk is lower among patients receiving thiopurine therapy. Patients without long-standing extensive colitis have a risk for CRC similar to that of the general population, but they can develop IBD-related lesions within 10 years after diagnosis of IBD.

Types of fish consumed and fish preparation methods in relation to pancreatic cancer incidence: the VITAL Cohort Study

The associations of types of fish and fish preparation methods with pancreatic cancer risk remain unknown. The authors conducted a prospective cohort study in western Washington State among 66,616 adults, aged 50-76 years, who participated in the VITamins And Lifestyle cohort study. Diet was assessed by a food frequency questionnaire. Pancreatic cancer cases were identified by linkage to the Surveillance, Epidemiology, and End Results cancer registry. During an average follow-up of 6.8 years, 151 participants developed pancreatic cancer (adenocarcinoma). Long-chain (n-3) polyunsaturated fatty acids (LC-PUFAs) and nonfried fish intake were inversely associated with pancreatic cancer incidence. When the highest and lowest tertiles of exposure were compared, the multivariable-adjusted hazard ratio of pancreatic cancer was 0.62 (95% confidence interval: 0.40, 0.98) (P(trend) = 0.08) for LC-PUFAs and 0.55 (95% confidence interval: 0.34, 0.88) (P(trend) = 0.045) for nonfried fish. Docosahexaenoic acid showed a greater inverse association with pancreatic cancer than eicosapentaenoic acid. No statistically significant associations were observed with fried fish and shellfish consumption. The potential health impact of fish consumption may depend on the types of fish consumed and fish preparation methods. LC-PUFAs, particularly docosahexaenoic acid, and nonfried fish, but not shellfish or fried fish, may be beneficial in the primary prevention of pancreatic cancer.

Physicians compliance during maintenance therapy in children with Down syndrome and acute lymphoblastic leukemia

Children with Down syndrome (DS) and acute lymphoblastic leukemia (ALL) have an inferior prognosis compared with non-DS ALL patients. We reviewed methotrexate (MTX)/mercaptopurine (6MP) maintenance therapy data for children with DS treated according to the Nordic Society of Pediatric Hematology and Oncology (NOPHO) ALL92 or the NOPHO ALL2000 protocols between 1992 and 2007. The 5-year event-free survival probability (pEFS(5 yr)) for the 66 DS patients was inferior to the 2602 non-DS patients (0.50 ± 0.07 vs 0.77 ± 0.01 (P<0.001)). The 48 DS patients in first remission at the beginning of maintenance therapy had pEFS(10 yr) below that of the 522 non-DS control patients (pEFS(10 yr): 0.58 (95% confidence interval (CI) 0.43-0.77) vs 0.83 (95% CI 0.80-0.86), respectively (P<0.0001)). The DS patients received lower median doses of MTX (median: 11.8 vs 15.4 (P<0.0001)) and 6MP (median: 43.6 vs 59.4 (P<0.0001)). In Cox regression analysis, male gender, presence of DS and high median maintenance therapy white blood cell levels (mWBC) were associated with increased risk for relapse. DS-ALL patients with mWBC above or below 3.5 × 10(9)/l (protocol target) had pEFS(10 yr) of 0.31 and 0.72 (P=0.02), and the mWBC hazard ratio for DS-ALL patients was 2.0 (P<0.0005). We conclude that insufficient treatment intensity during maintenance therapy of DS-ALL patients may contribute to their poor prognosis.

Person mobility in the design and analysis of cluster-randomized cohort prevention trials

Person mobility is an inescapable fact of life for most cluster-randomized (e.g., schools, hospitals, clinic, cities, state) cohort prevention trials. Mobility rates are an important substantive consideration in estimating the effects of an intervention. In cluster-randomized trials, mobility rates are often correlated with ethnicity, poverty and other variables associated with disparity. This raises the possibility that estimated intervention effects may generalize to only the least mobile segments of a population and, thus, create a threat to external validity. Such mobility can also create threats to the internal validity of conclusions from randomized trials. Researchers must decide how to deal with persons who leave study clusters during a trial (dropouts), persons and clusters that do not comply with an assigned intervention, and persons who enter clusters during a trial (late entrants), in addition to the persons who remain for the duration of a trial (stayers). Statistical techniques alone cannot solve the key issues of internal and external validity raised by the phenomenon of person mobility. This commentary presents a systematic, Campbellian-type analysis of person mobility in cluster-randomized cohort prevention trials. It describes four approaches for dealing with dropouts, late entrants and stayers with respect to data collection, analysis and generalizability. The questions at issue are: 1) From whom should data be collected at each wave of data collection? 2) Which cases should be included in the analyses of an intervention effect? and 3) To what populations can trial results be generalized? The conclusions lead to recommendations for the design and analysis of future cluster-randomized cohort prevention trials.

Model Checking Techniques for Assessing Functional Form Specifications in Censored Linear Regression Models

In this paper we develop model checking techniques for assessing functional form specifications of covariates in censored linear regression models. These procedures are based on a censored data analog to taking cumulative sums of "robust" residuals over the space of the covariate under investigation. These cumulative sums are formed by integrating certain Kaplan-Meier estimators and may be viewed as "robust" censored data analogs to the processes considered by Lin, Wei & Ying (2002). The null distributions of these stochastic processes can be approximated by the distributions of certain zero-mean Gaussian processes whose realizations can be generated by computer simulation. Each observed process can then be graphically compared with a few realizations from the Gaussian process. We also develop formal test statistics for numerical comparison. Such comparisons enable one to assess objectively whether an apparent trend seen in a residual plot reects model misspecification or natural variation. We illustrate the methods with a well known dataset. In addition, we examine the finite sample performance of the proposed test statistics in simulation experiments. In our simulation experiments, the proposed test statistics have good power of detecting misspecification while at the same time controlling the size of the test.

Regression calibration when foods (measured with error) are the variables of interest: markedly non-Gaussian data with many zeroes

Regression calibration has been described as a means of correcting effects of measurement error for normally distributed dietary variables. When foods are the items of interest, true distributions of intake are often positively skewed, may contain many zeroes, and are usually not described by well-known statistical distributions. The authors considered the validity of regression calibration assumptions where data are non-Gaussian. Such data (including many zeroes) were simulated, and use of the regression calibration algorithm was evaluated. An example used data from Adventist Health Study 2 (2002-2008). In this special situation, a linear calibration model does (as usual) at least approximately correct the parameter that captures the exposure-disease association in the "disease" model. Poor fit in the calibration model does not produce biased calibrated estimates when the "disease" model is linear, and it produces little bias in a nonlinear "disease" model if the model is approximately linear. Poor fit will adversely affect statistical power, but more complex linear calibration models can help here. The authors conclude that non-Gaussian data with many zeroes do not invalidate regression calibration. Irrespective of fit, linear regression calibration in this situation at least approximately corrects bias. More complex linear calibration equations that improve fit may increase power over that of uncalibrated regressions.

PERTURBATION AND SCALED COOK'S DISTANCE

Cook's (Cook, 1977) distance is one of the most important diagnostic tools for detecting influential individual or subsets of observations in linear regression for cross-sectional data. However, for many complex data structures (e.g., longitudinal data), no rigorous approach has been developed to address a fundamental issue: deleting subsets with different numbers of observations introduces different degrees of perturbation to the current model fitted to the data and the magnitude of Cook's distance is associated with the degree of the perturbation. The aim of this paper is to address this issue in general parametric models with complex data structures. We propose a new quantity for measuring the degree of the perturbation introduced by deleting a subset. We use stochastic ordering to quantify the stochastic relationship between the degree of the perturbation and the magnitude of Cook's distance. We develop several scaled Cook's distances to resolve the comparison of Cook's distance for different subset deletions. Theoretical and numerical examples are examined to highlight the broad spectrum of applications of these scaled Cook's distances in a formal influence analysis.

Prevalence and temporal trends of hepatitis B, hepatitis C, and HIV/AIDS co-infection during pregnancy across the decade, 1998-2007

BACKGROUND

Limited data are available on hepatitis rates during pregnancy by socio-demographic characteristics. This study examined temporal trends in hepatitis B virus (HBV) and hepatitis C virus (HCV) mono-infections and HIV/HBV and HIV/HCV co-infections in subpopulations among pregnant women in Florida between 1998 and 2007.

METHODS

We analyzed all Florida live births from 1998 to 2007 using hospital discharge data linked to birth records.

RESULTS

The total sample size was 1,700,734 singleton live births. The prevalance of HBV in pregnancy rose from 65.4 per 100,000 births to 123.5 per 100,000 births (p<0.0001 for trend), and the prevalence of HCV in pregnancy increased from 17.0 per 100,000 births to 125.1 per 100,000 births (p<0.0001 for trend). Compared with white mothers, black mothers were more than twice as likely to have HBV in pregnancy (adjusted rate ratios [ARR]=2.24; 95% CI=1.97-2.53). Black mothers were 69% (ARR=0.31, 95% CI=0.25-0.39) and Hispanic mothers were 51% (ARR=0.49, 95% CI=0.41-0.60) less likely to have HCV compared with white mothers.

CONCLUSIONS

Although the overall prevalence rate of HBV increased over the past decade, black women still had a noticeably higher rate of infection. Similarly, white women and those with HIV co-infection had noticeably higher rates of HCV infection over the study period. Our findings call for improved and increased HBV/HCV prevention, screening, and immunization programs among minority women of childbearing age.

Improving coping skills for self-management of treatment side effects can reduce antiretroviral medication nonadherence among people living with HIV

Background

Human immunodeficiency virus (HIV) treatment side effects have a deleterious impact on treatment adherence, which is necessary to optimize treatment outcomes including morbidity and mortality.

Purpose

To examine the effect of the Balance Project intervention, a five-session, individually delivered HIV treatment side effects coping skills intervention on antiretroviral medication adherence.

Methods

HIV+ men and women (N = 249) on antiretroviral therapy (ART) with self-reported high levels of ART side effect distress were randomized to intervention or treatment as usual. The primary outcome was self-reported ART adherence as measured by a combined 3-day and 30-day adherence assessment.

Results

Intent-to-treat analyses revealed a significant difference in rates of nonadherence between intervention and control participants across the follow-up time points such that those in the intervention condition were less likely to report nonadherence. Secondary analyses revealed that intervention participants were more likely to seek information about side effects and social support in efforts to cope with side effects.

Conclusions

Interventions focusing on skills related to ART side-effects management show promise for improving ART adherence among persons experiencing high levels of perceived ART side effects.

Distortion-product otoacoustic emission test performance for ototoxicity monitoring

INTRODUCTION

A nonbehavioral method for monitoring ototoxicity in patients treated with cisplatin is needed because patients enduring chemotherapy may not be well or cooperative enough to undergo repeated hearing tests. Distortion-product otoacoustic emissions (DPOAEs) provide a nonbehavioral measure of auditory function that is sensitive to cisplatin exposure. However, interpreting DPOAE findings in the context of ototoxicity monitoring requires that their accuracy be determined in relation to a clinically accepted gold standard test.

OBJECTIVES

Among patients receiving cisplatin for the treatment of cancer, we sought to (1) identify the combination of DPOAE metrics and ototoxicity risk factors that best classified ears with and without ototoxic-induced hearing changes; and (2) evaluate the test performance achieved by the composite measure as well as by DPOAEs alone.

DESIGN

Odds of experiencing hearing changes at a given patient visit were determined using data collected prospectively from 24 Veterans receiving cisplatin. Pure-tone thresholds were examined within an octave of each subject's high-frequency hearing limit. DPOAE were collected as a set of four response growth (input/output) functions near the highest f2 frequency that yielded a robust response at L2 = L1 = 65 dB SPL. Logistic regression modeled the risk of hearing change using several DPOAE metrics, drug treatment factors, and other patient factors as independent variables. An optimal discriminant function was derived by reducing the model so that only statistically significant variables were included. Receiver operating characteristic curve analyses were used to evaluate test performance.

RESULTS

At higher cisplatin doses, ears with better hearing at baseline were more likely to exhibit ototoxic hearing changes than those with poorer hearing. Measures of pre-exposure hearing, cumulative drug dose, and DPOAEs generated a highly accurate discriminant function with a cross-validated area under the receiver operating characteristic curve of 0.9. DPOAEs alone also provided an indication of ototoxic hearing change when measured at the highest DPOAE test frequency that yielded a robust response.

CONCLUSIONS

DPOAEs alone and especially in combination with pre-exposure hearing and cisplatin dose provide an indication of whether or not hearing has changed as a result of cisplatin administration. These promising results need to be validated in a separate sample.

Interaction between tacrolimus and intravenous nicardipine in the treatment of post-kidney transplant hypertension at pediatric hospitals

TAC is commonly prescribed in KTX recipients, though overexposure can be nephrotoxic. CIVN, used to treat post-KTX hypertension, may inhibit TAC metabolism resulting in overexposure and potential toxicity. We present two case reports and analysis of 2068 KTXs from the PHIS to characterize post-KTX intravenous anti-hypertensive use and to determine whether CIVN in TAC-treated patients would predict "immunosuppressive drug causing adverse effects in therapeutic use" (E-code E9331). CIVN was ordered in 11% of KTXs and prescribing increased from 6.2% in 2003 to 10.3% in 2008 (p=0.003, Mantel-Haenszel chi-square test). AEI were reported in 7.1% of TAC-treated patients with CIVN orders compared to 3% of those without (p=0.003, chi-square test). In univariate analysis using GEEs, AEI were twofold more likely in patients with CIVN orders than patients without (AOR 2.1, 95% CI 1.03-4.17) and threefold more likely than patients with orders for other continuous intravenous anti-hypertensives (AOR 3.2, 95% CI 1.09-9.08). In multivariate analysis, only CIVN significantly predicted AEI (AOR 2.8, 95% CI 1.29-6.04). Thus, until clinical studies to fully characterize this interaction are completed, CIVN should be used with caution in TAC-treated individuals.

Impact of neoadjuvant chemotherapy on breast reconstruction

BACKGROUND

With advances in oncologic treatment, cosmesis after mastectomy has assumed a pivotal role in patient and provider decision making. Multiple studies have confirmed the safety of both chemotherapy before breast surgery and immediate reconstruction. Little has been written about the effect of neoadjuvant chemotherapy on decisions about reconstruction.

METHODS

The authors identified 665 patients with stage I through III breast cancer who received chemotherapy and underwent mastectomy at Dana-Farber/Brigham & Women's Cancer Center from 1997 to 2007. By using multivariate logistic regression, reconstruction rates were compared between patients who received neoadjuvant chemotherapy (n = 180) and patients who underwent mastectomy before chemotherapy (n = 485). The rate of postoperative complications after mastectomy was determined for patients who received neoadjuvant chemotherapy compared with those who did not.

RESULTS

Reconstruction was performed immediately in 44% of patients who did not receive neoadjuvant chemotherapy but in only 23% of those who did. Twenty-one percent of neoadjuvant chemotherapy recipients and 14% of adjuvant-only chemotherapy recipients underwent delayed reconstruction. After controlling for age, receipt of radiotherapy, and disease stage, neoadjuvant recipients were less likely to undergo immediate reconstruction (odds ratio [OR], 0.57; 95% confidence interval [CI], 0.37, 0.87) but were no more likely to undergo delayed reconstruction (OR, 1.29; 95% CI, 0.75, 2.20). Surgical complications occurred in 30% of neoadjuvant chemotherapy recipients and in 31% of adjuvant chemotherapy recipients.

CONCLUSIONS

The current results suggest that patients who receive neoadjuvant chemotherapy are less likely to undergo immediate reconstruction and are no more likely to undergo delayed reconstruction than patients who undergo surgery before they receive chemotherapy.

Association between risk-assessment scores and individual-cow Johne's disease-test status over time on seven Michigan, USA dairy herds

To evaluate the effectiveness of management practices implemented to control the spread of Johne's disease (JD), we conducted a 5-year observational study (January 2003 to December 2007) on seven Michigan, USA dairy herds containing cows infected with Mycobacterium avium subsp. paratuberculosis (MAP; the causative agent of the disease). The JD incidence and prevalence was monitored in each herd annually by serum ELISA and/or fecal culture of all adult cows. A JDCP was designed specifically for each herd based on the results of an initial risk-assessment. The risk-assessment was repeated annually and the control program updated as needed. Herd risk-assessment scores were used to measure compliance with the control program and create JD-risk profiles for individual cows raised on the farms. The association between specific risk-assessment scores and the JD-test status of individual cows was evaluated using logistic regression. We accounted for clustering of cows within herds using generalized estimating equations (GEE). Multivariable models were built with purposeful selection of risk factors assessed on univariable analyses. The dataset analyzed consisted of 3707 cows raised on the respective farms, of which 616 were classified as infected with MAP based on testing positive on fecal culture or serum ELISA. Of the cows that were not exposed to the control program, 20% were classified as infected, while only 7% of cows that were exposed to the control program were infected. The final multivariable model consisted of two factors: exposure to adult cows other than dam at birth (OR=1.09, 95% CI: 1.06, 1.13), and feeding colostrum from one cow to multiple calves (OR=1.10, 95% CI: 1.09, 1.12). Based on this study, implementing practices that minimize the exposure of newborn calves to MAP being shed by infected adult cows should take priority.

Nesiritide in acute decompensated heart failure: a pooled analysis of randomized controlled trials

BACKGROUND

Previous randomized controlled trials (RCTs) evaluating nesiritide for the treatment of acute decompensated heart failure (ADHF) have reported wide variances in mortality hazard ratios for nesiritide vs controls, but these individual trials were neither designed nor powered to evaluate mortality. This study used relevant data from all RCTs of nesiritide in ADHF completed as of June 2006 to independently estimate the effect of nesiritide on 30- and 180-day mortality.

HYPOTHESIS

Administration of nesiritide to treat patients with ADHF does not significantly increase mortality at 30 or 180 days.

METHODS

Six trials met prespecified criteria for inclusion in this analysis. Primary data from these trials were obtained from Scios Inc. (Fremont, CA). Statistical models were fitted to estimate 4 effects: dose response, differing control groups, vulnerable subgroup interactions, and time of death relative to nesiritide administration. All models included 4 baseline covariates that were significantly and independently associated with mortality.

RESULTS

Complete covariate data were available in 1472 of 1538 subjects (96%). The risk-adjusted hazard ratio for mortality was 1.05 (95% confidence interval [CI]: 0.85-1.30) at 30 and 1.00 (95% CI: 0.88-1.14) at 180 days with no clear relationship to nesiritide dose. In addition to consistent results across 2 time points, no significant evidence of sensitivity to control group or baseline risk factors was found.

CONCLUSIONS

Currently available data suggest nesiritide does not significantly increase mortality at 30 or 180 days.

Choline intake and genetic polymorphisms influence choline metabolite concentrations in human breast milk and plasma

BACKGROUND

Choline is essential for infant nutrition, and breast milk is a rich source of this nutrient. Common single nucleotide polymorphisms (SNPs) change dietary requirements for choline intake.

OBJECTIVE

The aim of this study was to determine whether total choline intake and/or SNPs influence concentrations of choline and its metabolites in human breast milk and plasma.

DESIGN

We gave a total of 103 pregnant women supplemental choline or a placebo from 18 wk gestation to 45 d postpartum and genotyped the women for 370 common SNPs. At 45 d postpartum, we measured choline metabolite concentrations in breast milk and plasma and assessed the dietary intake of choline by using a 3-d food record.

RESULTS

On average, lactating women in our study ate two-thirds of the recommended intake for choline (Adequate Intake = 550 mg choline/d). Dietary choline intake (no supplement) correlated with breast-milk phosphatidylcholine and plasma choline concentrations. A supplement further increased breast-milk choline, betaine, and phosphocholine concentrations and increased plasma choline and betaine concentrations. We identified 5 SNPs in MTHFR that altered the slope of the intake-metabolite concentration relations, and we identified 2 SNPs in PEMT that shifted these curves upward. Individuals who shared sets of common SNPs were outliers in plots of intake-metabolite concentration curves; we suggest that these SNPs should be further investigated to determine how they alter choline metabolism.

CONCLUSION

Total intake of choline and genotype can influence the concentrations of choline and its metabolites in the breast milk and blood of lactating women and thereby affect the amount of choline available to the developing infant. This study was registered at clinicaltrials.gov as NCT00678925.

Super-obesity and risk for early and late pre-eclampsia

OBJECTIVE

To examine the association between obesity subtypes and risk of early and late pre-eclampsia.

DESIGN

Population-based retrospective study.

SETTING

State of Missouri maternally linked birth cohort files.

POPULATION

All singleton live births in the state of Missouri from 1989 to 2005.

METHODS

The body mass index (BMI) was used to classify women as normal weight (BMI = 18.5-24.9 kg/m(2)), class I obesity (BMI = 30-34.9 kg/m(2)), class II obesity (BMI = 35-39.9 kg/m(2)), class III obesity (BMI = 40-49.9 kg/m(2)) or super-obesity (BMI > or = 50 kg/m(2)). Adjusted odds ratios (OR) and 95% confidence intervals (CI) for the association between obesity and the risk of pre-eclampsia were obtained from logistic regression models with adjustment for intracluster correlation.

RESULTS

The rate of pre-eclampsia increased with increasing BMI, with super-obese women having the highest incidence (13.4%). Compared with normal weight women, obese women (BMI > or = 30 kg/m(2)) had a higher risk for pre-eclampsia (OR = 2.59, 95% CI = 2.87-3.01). This risk remained approximately the same for late-onset pre-eclampsia (pre-eclampsia occurring at 34 weeks or more of gestation) and was slightly reduced for early-onset pre-eclampsia (pre-eclampsia occurring at 34 weeks or less of gestation). Within each BMI category, the risk of pre-eclampsia increased with the rate of weight gain. Compared with normal weight mothers with moderate weight gain, super-obese women with a high rate of weight gain had the greatest risk for pre-eclampsia (OR = 7.52, 95% CI = 2.70-21.0).

CONCLUSION

BMI and rate of weight gain are synergistic risk factors that amplify the burden of pre-eclampsia among super-obese women.

The impact of race/ethnicity on preoperative time to hip stabilization procedure after hip fracture

BACKGROUND

We sought to examine the preoperative time for hip stabilization procedure among Hispanics, non-Hispanic blacks (blacks) and non-Hispanic whites (whites).

METHODS

This was a secondary data analysis using Medicare claims data. Our analysis included 40,321 patients admitted for hip fracture hospitalization from 2001-2005. Our primary analysis was generalized linear modeling, and our dependent variable was preoperative time. Our independent variable was race/ethnicity (Hispanics, blacks versus whites), and covariates were age, gender, income, type of hip fracture and comorbidities.

RESULTS

Bivariate analyses showed that both Hispanics and blacks experienced a longer preoperative time (P < 0.01). The average (mean) of days to surgery was 1.2 for whites, 1.6 for blacks and 1.7 for Hispanics. The delayed preoperative time among Hispanics and blacks persisted after adjusting for covariates.

CONCLUSIONS

The delayed preoperative time among minorities suggests the need to closely monitor care among minorities with hip fracture to determine how to best address their developing needs.

Lack of association between thiazolidinediones and macular edema in type 2 diabetes: the ACCORD eye substudy

OBJECTIVE

To assess the cross-sectional association of thiazolidinediones with diabetic macular edema (DME).

METHODS

The cross-sectional association of DME and visual acuity with thiazolidinediones was examined by means of baseline fundus photographs and visual acuity measurements from the Action to Control Cardiovascular Risk in Diabetes (ACCORD) trial. Visual acuity was assessed in 9690 participants in the ACCORD trial, and 3473 of these participants had fundus photographs that were centrally read in a standardized fashion by masked graders to assess DME and retinopathy from October 23, 2003, to March 10, 2006.

RESULTS

Among the subsample, 695 (20.0%) people had used thiazolidinediones, whereas 217 (6.2%) people had DME. Thiazolidinedione use was not associated with DME in unadjusted (odds ratio [OR], 1.01; 95% confidence interval [CI], 0.71-1.44; P = .95) and adjusted (OR, 0.97; 95% CI, 0.67-1.40; P = .86) analyses. Significant associations with DME were found for retinopathy severity (P < .001) and age (OR, 0.97; 95% CI, 0.952-0.997; P = .03) but not for hemoglobin A(1c) (P = .06), duration of diabetes (P = .65), sex (P = .72), and ethnicity (P = .20). Thiazolidinedione use was associated with slightly greater visual acuity (0.79 letter; 95% CI, 0.20-1.38; P = .009) of uncertain clinical significance.

CONCLUSIONS

In a cross-sectional analysis of data from the largest study to date, no association was observed between thiazolidinedione exposure and DME in patients with type 2 diabetes; however, we cannot exclude a modest protective or harmful association. Trial Registration clinicaltrials.gov Identifier: NCT00542178.

Spatial cluster detection for repeatedly measured outcomes while accounting for residential history

Spatial cluster detection has become an important methodology in quantifying the effect of hazardous exposures. Previous methods have focused on cross-sectional outcomes that are binary or continuous. There are virtually no spatial cluster detection methods proposed for longitudinal outcomes. This paper proposes a new spatial cluster detection method for repeated outcomes using cumulative geographic residuals. A major advantage of this method is its ability to readily incorporate information on study participants relocation, which most cluster detection statistics cannot. Application of these methods will be illustrated by the Home Allergens and Asthma prospective cohort study analyzing the relationship between environmental exposures and repeated measured outcome, occurrence of wheeze in the last 6 months, while taking into account mobile locations.

Diagnostic Measures for Generalized Linear Models with Missing Covariates

In this paper, we carry out an in-depth investigation of diagnostic measures for assessing the influence of observations and model misspecification in the presence of missing covariate data for generalized linear models. Our diagnostic measures include case-deletion measures and conditional residuals. We use the conditional residuals to construct goodness-of-fit statistics for testing possible misspecifications in model assumptions, including the sampling distribution. We develop specific strategies for incorporating missing data into goodness-of-fit statistics in order to increase the power of detecting model misspecification. A resampling method is proposed to approximate the p-value of the goodness-of-fit statistics. Simulation studies are conducted to evaluate our methods and a real data set is analysed to illustrate the use of our various diagnostic measures.

Lymphoproliferative disorders in patients receiving thiopurines for inflammatory bowel disease: a prospective observational cohort study

BACKGROUND

Reports of an increased risk of lymphoproliferative disorders in patients receiving thiopurines for inflammatory bowel disease are controversial. We assessed this risk in a prospective observational cohort study.

METHODS

19,486 patients with inflammatory bowel disease, of whom 11,759 (60.3%) had Crohn's disease and 7727 (39.7%) had ulcerative colitis or unclassified inflammatory bowel disease, were enrolled in a nationwide French cohort by 680 gastroenterologists, who reported details of immunosuppressive therapy during the observation period, cases of cancer, and deaths. The risk of lymphoproliferative disorder was assessed according to thiopurine exposure. Median follow-up was 35 months (IQR 29-40).

FINDINGS

At baseline, 5867 (30.1%) of patients were receiving, 2809 (14.4%) had discontinued, and 10,810 (55.5%) had never received thiopurines. 23 new cases of lymphoproliferative disorder were diagnosed, consisting of one case of Hodgkin's lymphoma and 22 cases of non-Hodgkin lymphoproliferative disorder. The incidence rates of lymphoproliferative disorder were 0.90 per 1000 (95% CI 0.50-1.49) patient-years in those receiving, 0.20/1000 (0.02-0.72) patient-years in those who had discontinued, and 0.26/1000 (0.10-0.57) patient-years in those who had never received thiopurines (p=0.0054). The multivariate-adjusted hazard ratio of lymphoproliferative disorder between patients receiving thiopurines and those who had never received the drugs was 5.28 (2.01-13.9, p=0.0007). Most cases associated with thiopurine exposure matched the pathological range of post-transplant disease.

INTERPRETATION

Patients receiving thiopurines for inflammatory bowel disease have an increased risk of developing lymphoproliferative disorders.

FUNDING

Programme Hospitalier de Recherche Clinique National (AOM05157), Association François Aupetit, Délégation Inter-régionale de la Recherche clinique Ile de France-Assistance Publique Hôpitaux de Paris (AP-HP), Ligue contre le Cancer, and Fonds de Recherche de la Société Nationale Française de Gastro-entérologie.

Trends in primary care clinician perceptions of a new electronic health record

BACKGROUND

Clinician perceptions of a newly implemented electronic health record play an important role in its success or failure.

OBJECTIVE

To measure changes in primary care clinician attitudes toward an electronic health record during the first year following implementation.

DESIGN

Longitudinal survey.

PARTICIPANTS

86 primary care clinicians surveyed between December 2006 and January 2008.

MEASUREMENTS

Perceived impact on overall quality of care, patient safety, communication, and efficiency at 1, 3, 6, and 12 months following implementation.

RESULTS

Response rates for months 1, 3, 6, and 12 were 92%, 95%, 90%, and 82%, respectively. The proportion of clinicians agreeing that the EHR improved the overall quality of care (63% to 86%; p < 0.001), reduced medication-related errors (72% to 81%; p = 0.03), improved follow-up of test results (62% to 87%; p < 0.001), and improved communication among clinicians (72% to 93%; p < 0.001) increased from month 1 to month 12. During the same time period, a decreasing proportion of clinicians agreed that the EHR reduced the quality of patient interactions (49% to 33%; p = 0.001), resulted in longer patient visits (68% to 51%; p = 0.001), and increased time spent on medical documentation (78% to 68%; p = 0.006). Significant improvements in perceptions related to test result follow-up were first detected at 6 months, while those related to overall quality, efficiency, and communication were first identified at 12 months.

CONCLUSIONS

Primary care clinicians report increasingly positive perceptions of a new electronic health record within 1 year of implementation across a spectrum of domains of care.

Socioeconomic differences in patient survival are increasing for acute myeloid leukemia and multiple myeloma in sweden

PURPOSE

An association between socioeconomic status (SES) and survival in acute myeloid leukemia (AML) and multiple myeloma (MM) has not been established in developed countries. We assessed the impact of SES on survival in two large population-based cohorts of AML and MM patients diagnosed in Sweden 1973 to 2005.

PATIENTS AND METHODS

The relative risk of death (all cause and cause specific) in relation to SES was estimated using Cox's proportional hazards regression. We also conducted analyses stratified by calendar periods (1973 to 1979, 1980 to 1989, 1990 to 1999, and 2000 to 2005).

RESULTS

We identified a total of 9,165 and 14,744 patients with AML and MM, respectively. Overall, higher white-collar workers had a lower mortality than other SES groups for both AML (P = .005) and MM (P < .005). In AML patients, a consistently higher overall mortality was observed in blue-collar workers compared with higher white-collar workers in the last three periods (hazard ratio [HR], 1.26; 95% CI, 1.05 to 1.51; HR, 1.23; 95% CI, 1.05 to 1.45; HR, 1.28; 95% CI, 1.04 to 1.57, respectively). In MM, no difference was observed in the first two calendar periods. However, in 1990 to 1999, self-employed (HR, 1.18; 95% CI, 1.02 to 1.37), blue-collar workers (HR, 1.18; 95% CI, 1.04 to 1.32), and retired (HR, 1.45; 95% CI, 1.16 to 1.80) had a higher mortality compared to higher white-collar workers. In 2000 to 2005, blue-collar workers had a higher mortality (HR, 1.31; 95% CI, 1.07 to 1.60) compared with higher white-collar workers.

CONCLUSION

SES was significantly associated with survival in both AML and MM. Most conspicuously, a lower mortality was observed among the highest SES group during more recent calendar periods. Differences in management, comorbidity, and lifestyle, are likely factors to explain these findings.

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.

Classifying and predicting errors of inpatient medication reconciliation

BACKGROUND

Failure to reconcile medications across transitions in care is an important source of potential harm to patients. Little is known about the predictors of unintentional medication discrepancies and how, when, and where they occur.

OBJECTIVE

To determine the reasons, timing, and predictors of potentially harmful medication discrepancies.

DESIGN

Prospective observational study.

PATIENTS

Admitted general medical patients.

MEASUREMENTS

Study pharmacists took gold-standard medication histories and compared them with medical teams' medication histories, admission and discharge orders. Blinded teams of physicians adjudicated all unexplained discrepancies using a modification of an existing typology. The main outcome was the number of potentially harmful unintentional medication discrepancies per patient (potential adverse drug events or PADEs).

RESULTS

Among 180 patients, 2066 medication discrepancies were identified, and 257 (12%) were unintentional and had potential for harm (1.4 per patient). Of these, 186 (72%) were due to errors taking the preadmission medication history, while 68 (26%) were due to errors reconciling the medication history with discharge orders. Most PADEs occurred at discharge (75%). In multivariable analyses, low patient understanding of preadmission medications, number of medication changes from preadmission to discharge, and medication history taken by an intern were associated with PADEs.

CONCLUSIONS

Unintentional medication discrepancies are common and more often due to errors taking an accurate medication history than errors reconciling this history with patient orders. Focusing on accurate medication histories, on potential medication errors at discharge, and on identifying high-risk patients for more intensive interventions may improve medication safety during and after hospitalization.

Line region hypomethylation is associated with lifestyle and differs by human papillomavirus status in head and neck squamous cell carcinomas

Genomic hypomethylation is a hallmark of essentially all cancers, but the degree of this hypomethylation differs among individual tumors. Little work has explored what leads to these differences and or asked whether they are clinically meaningful. In this study of head and neck squamous cell carcinoma, we assessed hypomethylation in tumors using a semiquantitative fragment analysis approach to determine the relative methylation status of the line retroviral element LRE1 (Line-1.2). Because this is an established marker of genomic methylation status, we examined the relationship between the relative methylation, patient demographics, and other risk factors for head and neck squamous cell carcinoma. We determined relative methylation status for 303 patients, 193 of which had complete data for all variables of interest. Using a generalized linear model, we found that patient body mass index was significantly positively associated with tumor LRE1 methylation level. Smoking duration, particularly in tumors lacking human papillomavirus (HPV) DNA, was significantly negatively associated with relative methylation level. Having previously assessed relative methylation in blood-derived DNA, we compared tumor with the blood DNA methylation level and observed these to be independent. Finally, the lower LRE1 methylation in patients whose tumors were HPV DNA negative was associated with poorer patient survival (hazard ratio, 1.6; 95% confidence interval, 1.0-2.6). These findings suggest that HPV-associated tumors differ molecularly from those arising after heavy tobacco use and that this epigenetic alteration may affect survival in HPV-negative patients already exhibiting a more aggressive disease.

Characteristics of symptomatic reflux episodes on Acid suppressive therapy

BACKGROUND

Persistent symptoms on acid suppressive therapy are due to either acid or nonacid gastroesophageal reflux (GER) episodes or are not related to reflux.

AIM

To compare physical and chemical characteristics of GER episodes associated with symptoms in patients on acid suppressive therapy.

METHODS

Patients with persistent symptoms on acid suppressive therapy underwent combined impedance-pH monitoring. Reflux episodes were classified as acid if nadir pH was <4.0, and nonacid if it remained at >/=4.0, separated into liquid-only or mixed (liquid-gas), and considered to reach the proximal esophagus if liquid was present 15 cm above the lower esophageal sphincter (LES). Reflux episodes were considered symptomatic if patients recorded a symptom within 5 min after the reflux episode. Risk factors of symptomatic reflux episodes were identified using multivariable generalized estimating equations (GEEs).

RESULTS

One hundred twenty patients (85 women, median age 54 yr, range 18-85 yr) recorded 3,547 reflux episodes (84.3% nonacid, 50.6% mixed), of which 468 (13.2%) were symptomatic. Based on multivariable GEE analysis with episode-level symptom status (symptomatic vs nonsymptomatic) as the outcome variable, reflux episode acidity was not significantly associated with symptoms (P= 0.40). Mixed reflux episodes were significantly associated with symptoms relative to liquid-only episodes (odds ratio [OR] 1.49, 95% confidence interval [CI] 1.19-1.87, P= 0.0005), as were reflux episodes reaching the proximal esophagus compared with those reaching the distal esophagus only (OR 1.28, 95% CI 1.06-1.55, P= 0.012).

CONCLUSION

The majority of reflux episodes on acid suppressive therapy are asymptomatic. Reflux episodes extending proximally and having a mixed (liquid-gas) composition are significantly associated with symptoms, irrespective of whether pH is acid (<4) or nonacid (>/=4).

Residual-based diagnostics for structural equation models

Classical diagnostics for structural equation models are based on aggregate forms of the data and are ill suited for checking distributional or linearity assumptions. We extend recently developed goodness-of-fit tests for correlated data based on subject-specific residuals to structural equation models with latent variables. The proposed tests lend themselves to graphical displays and are designed to detect misspecified distributional or linearity assumptions. To complement graphical displays, test statistics are defined; the null distributions of the test statistics are approximated using computationally efficient simulation techniques. The properties of the proposed tests are examined via simulation studies. We illustrate the methods using data from a study of in utero lead exposure.

Integrating the predictiveness of a marker with its performance as a classifier

There are two popular statistical approaches to biomarker evaluation. One models the risk of disease (or disease outcome) with, for example, logistic regression. A marker is considered useful if it has a strong effect on risk. The second evaluates classification performance by use of measures such as sensitivity, specificity, predictive values, and receiver operating characteristic curves. There is controversy about which approach is more appropriate. Moreover, the two approaches can give contradictory results on the same data. The authors present a new graphic, the predictiveness curve, which complements the risk modeling approach. It assesses the usefulness of a risk model when applied to the population. Although the predictiveness curve relates to classification performance measures, it also displays essential information about risk that is not displayed by the receiver operating characteristic curve. The authors propose that the predictiveness and classification performance of a marker, displayed together in an integrated plot, provide a comprehensive and cohesive assessment of a risk marker or model. The methods are demonstrated with data on prostate-specific antigen and risk factors from the Prostate Cancer Prevention Trial, 1993-2003.

Testing polynomial covariate effects in linear and generalized linear mixed models

An important feature of linear mixed models and generalized linear mixed models is that the conditional mean of the response given the random effects, after transformed by a link function, is linearly related to the fixed covariate effects and random effects. Therefore, it is of practical importance to test the adequacy of this assumption, particularly the assumption of linear covariate effects. In this paper, we review procedures that can be used for testing polynomial covariate effects in these popular models. Specifically, four types of hypothesis testing approaches are reviewed, i.e. R tests, likelihood ratio tests, score tests and residual-based tests. Derivation and performance of each testing procedure will be discussed, including a small simulation study for comparing the likelihood ratio tests with the score tests.

Modeling left-turn crash occurrence at signalized intersections by conflicting patterns

In order to better understand the underlying crash mechanisms, left-turn crashes occurring at 197 four-legged signalized intersections over 6 years were classified into nine patterns based on vehicle maneuvers and then were assigned to intersection approaches. Crash frequency of each pattern was modeled at the approach level by mainly using Generalized Estimating Equations (GEE) with the Negative Binomial as the link function to account for the correlation among the crash data. GEE with a binomial logit link function was also applied for patterns with fewer crashes. The Cumulative Residuals test shows that, for correlated left-turn crashes, GEE models usually outperformed basic Negative Binomial models. The estimation results show that there are obvious differences in the factors that cause the occurrence of different left-turn collision patterns. For example, for each pattern, the traffic flows to which the colliding vehicles belong are identified to be significant. The width of the crossing distance (represented by the number of through lanes on the opposing approach of the left-turning traffic) is associated with more left-turn traffic colliding with opposing through traffic (Pattern 5), but with less left-turning traffic colliding with near-side crossing through traffic (Pattern 8). The safety effectiveness of the left-turning signal is not consistent for different crash patterns; "protected" phasing is correlated with fewer Pattern 5 crashes, but with more Pattern 8 crashes. The study indicates that in order to develop efficient countermeasures for left-turn crashes and improve safety at signalized intersections, left-turn crashes should be considered in different patterns.

Tests and diagnostic plots for detecting lack-of-fit for circular-linear regression models

Regression diagnostics and lack-of-fit tests mainly focus on linear-linear regression models. When the design points are distributed on the circumference of a circle, difficulties arise as there is no natural starting point or origin. Most classical lack-of-fit tests require an arbitrarily chosen origin, but different choices may result in different conclusions. We propose a graphical diagnostic tool and a closely related lack-of-fit test, which does not require a natural starting point. The method is based on regional residuals which are defined on arcs of the circle. The graphical method formally locates and visualizes subsets of poorly fitting observations on the circle. A data example from the food technology is used to point out the before-mentioned problems with conventional lack-of-fit tests and to illustrate the strength of the methodology based on regional residuals in detecting and localizing departures from the no-effect hypothesis. A small simulation study shows a good performance of the regional residual test in case of both global and local deviations from the null model. Finally, the ideas are extended to the case of more than one predictor variable.

Reproductive history, infertility treatment, and the risk of acute leukemia in children with down syndrome: a report from the Children's Oncology Group

BACKGROUND

Children with Down syndrome (DS) have from 10 to 20 times the risk of developing acute leukemia than the general pediatric population. There is mixed evidence for associations between reproductive history or infertility and acute leukemia among children without DS.

METHODS

The authors conducted a case-control study of acute leukemia among children with DS to investigate possible risk factors in this population. From 1997 to 2002, 158 children aged <20 years with DS who had a diagnosis of acute leukemia (97 children with acute lymphoblastic leukemia [ALL] and 61 children with acute myeloid leukemia [AML]) were enrolled at Children's Oncology Group (COG) institutions. Controls with DS (n = 173) were selected from the cases' primary care clinic and frequency matched to cases on age. Telephone interviews were conducted with mothers of cases and controls assessing reproductive history, infertility, and infertility treatment.

RESULTS

Null results were observed overall and by subtype for reproductive factors, including previous pregnancy outcomes and contraceptive use, and for most infertility outcomes. There was an increased risk of AML among children with DS whose parents had ever tried for >/=1 year to become pregnant (odds ratio [OR], 2.22; 95% confidence interval [95% CI], 1.14-4.33). A 1-year increase in maternal age also was associated with AML (OR, 1.06; 95% CI, 1.01-1.12).

CONCLUSIONS

Although the questionnaire was limited in this area, the results suggested that the risk for AML may be raised in children with DS because of infertility. In that the risk of infertility, along with having a child with DS, increase with age, these results warrant more research.

A two-part model for censored medical cost data

The two-part model is often used to analyse medical cost data which contain a large proportion of zero cost and are highly skewed with some large costs. The total medical costs over a period of time are often censored due to incomplete follow-up, making the analysis difficult as the censoring can be informative. We propose to apply the inverse probability weighting method on a two-part model to analyse right-censored cumulative medical costs with informative censoring. We also introduce a set of simple functionals based on the intermediate cost history to be applied with the efficiency augmentation technique. In addition, we propose a practical model-checking technique based on the cumulative residuals. Simulation studies are conducted to evaluate the finite sample performance of the proposed method. We use a data set on the cardiovascular disease (CVD)-related Medicare costs to illustrate our proposed method.