Faculty Bibliography

Pooled analyses integrate data from multiple studies and achieve a larger sample size for enhanced statistical power. When heterogeneity exists in variables' effects on the outcome across studies, the simple pooling strategy fails to present a fair and complete picture of the effects of heterogeneous variables. Thus, it is important to investigate the homogeneous and heterogeneous structure of variables in pooled studies. In this article, we consider the pooled cohort studies with time-to-event outcomes and propose a penalized Cox partial likelihood approach with adaptively weighted composite penalties on variables' homogeneous and heterogeneous effects. We show that our method can characterize the variables as having heterogeneous, homogeneous, or null effects, and estimate non-zero effects. The results are readily extended to high-dimensional applications where the number of parameters is larger than the sample size. The proposed selection and estimation procedure can be implemented using the iterative shooting algorithm. We conduct extensive numerical studies to evaluate the performance of our proposed method and demonstrate it using a pooled analysis of gene expression in patients with ovarian cancer.

BACKGROUND: Destruction of the World Trade Center (WTC) towers on September 11, 2001, released massive dust, gas, and fumes with environmental exposures for community members. Many community members have lower respiratory symptoms (LRSs) that began after September 11, 2001, and remain persistent. We evaluated whether systemic inflammation measured by C-reactive protein was associated with WTC dust exposures, persistent LRS, and lung function. METHODS: Community members self-referred for the treatment of symptoms related to September 11, 2001. C-reactive protein and lung function measurements, including spirometry and forced oscillation tests (impulse oscillometry system), were included as routine analyses in patients (2007 to 2012). RESULTS: Increased C-reactive protein levels were associated with the type of WTC dust exposure, LRS, reduced spirometry, and increased forced oscillation measurements (n = 724). CONCLUSIONS: Ongoing systemic inflammation measured years after the event was associated with WTC dust exposures, persistent LRS, and abnormal lung function in a community cohort. These findings have implications for treatment and surveillance.

IMPORTANCE/OBJECTIVE:The target for the oral erectile dysfunction drugs, phosphodiesterase type 5 (PDE5) inhibitors, is part of a pathway implicated in the development of malignant melanoma. An increased risk of melanoma in sildenafil users was recently reported. OBJECTIVE:To examine the association between use of PDE5 inhibitors and melanoma risk, including data on specific PDE5 inhibitors, number of prescriptions, and melanoma stage. DESIGN, SETTING, AND PARTICIPANTS/METHODS:Nationwide, population-based, nested case-control study in the Swedish Prescribed Drug Register, the Swedish Melanoma Register, and other health care registers and demographic databases in Sweden, including 4065 melanoma cases diagnosed from 2006 through 2012 and 5 randomly selected controls per case with matching year of birth. EXPOSURES/METHODS:Number of filled prescriptions for the PDE5 inhibitors sildenafil and vardenafil or tadalafil. MAIN OUTCOMES AND MEASURES/METHODS:Risk of melanoma; overall and by stage and risk of basal cell carcinoma in multivariable logistic regression analyses. RESULTS:Of 4065 melanoma cases, 435 men (11%) had filled prescriptions for PDE5 inhibitors, as did 1713 men of 20,325 controls (8%). In multivariable analysis, there was an increased risk of melanoma in men taking PDE5 inhibitors (OR, 1.21 [95% CI, 1.08-1.36]). The most pronounced increase in risk was observed in men who had filled a single prescription (OR, 1.32 [95% CI, 1.10-1.59]; exposure rate, 4% for cases vs 3% for controls), but was not significant among men with multiple filled prescriptions (for 2-5 prescriptions: OR, 1.14 [95% CI, 0.95-1.37], 4% for cases and 3% for controls; for ≥6 prescriptions: OR, 1.17 [95% CI, 0.95-1.44], 3% for cases vs 2% for controls). PDE5 inhibitors were significantly associated with melanoma stage 0 (OR, 1.49 [95% CI, 1.22-1.83], 13% for cases vs 8% for controls) and stage I (OR, 1.21 [95% CI, 1.02-1.43], 12% for cases vs 10% for controls), but not stage II through IV (OR, 0.83 [95% CI, 0.63-1.09], 6% for cases vs 7% for controls). The risk estimates were similar for sildenafil and vardenafil or tadalafil. PDE5 inhibitor use was also associated with an increased risk of basal cell carcinoma (OR, 1.19 [95% CI, 1.14-1.25], 9% for cases vs 8% for controls). Men taking PDE5 inhibitors had a higher educational level and annual income, factors that were also significantly associated with melanoma risk. CONCLUSIONS AND RELEVANCE/CONCLUSIONS:In a Swedish cohort of men, the use of PDE5 inhibitors was associated with a modest but statistically significant increased risk of malignant melanoma. However, the pattern of association (eg, the lack of association with multiple filled prescriptions) raises questions about whether this association is causal.

Few studies have reported on African American and Hispanic (AA and H) populations' informational needs when seeking cancer care at an institution that offers clinical trials. Moffitt Cancer Center (MCC) sought to identify and examine the decision making process, the perceptions, and the preferred channels of communication about cancer care services for AA and H communities in order to develop a list of marketing recommendations. Five focus groups (N = 45) consisting of two AA and three H were conducted in four counties of the MCC catchment area in Tampa, FL. Participants were asked about their perceptions, knowledge, attitudes, and beliefs about cancer care and MCC. Focus groups were audio-recorded and verbatim transcripts were analyzed using content analysis. Similarities in responses were found between AA and H participants. Participants received general health and cancer information from media sources and word of mouth and preferred to hear patient testimonials. There were concerns about costs, insurance coverage, and the actual geographic location of the cancer center. In general, H participants were not opposed to participating in cancer clinical trials/research, whereas, AA participants were more hesitant. A majority of participants highly favored an institution that offered standard care and clinical trials. AA and H participants shared similar concerns and preferences in communication channels, but each group had specific informational needs. The perceptions and preferences of AA and H must be explored in order to successfully and efficiently increase cancer clinical trial participation.

This study explored federally qualified health center (FQHC) patients' perceptions about colorectal cancer screening (CRCS) tests, including immunochemical fecal occult blood tests (iFOBT), as well as preferences for receiving in-clinic education about CRCS. Eight mixed gender focus groups were conducted with 53 patients. Findings centered on three thematic factors: (1) motivators and impediments to CRCS, (2) test-specific preferences and receptivity to iFOBTs, and (3) preferences for entertaining and engaging plain language materials. Results informed the development of educational priming materials to increase CRCS using iFOBT in FQHCs.

BACKGROUND: Guidelines for anticoagulant therapy in patients with atrial fibrillation (AF) conflict with each other. The American College of Chest Physicians (ACCP) guidelines suggest no anticoagulant therapy for patients with a CHADS2 score of 0. The European Society of Cardiology (ESC) prefer anticoagulant therapy for patients with a CHA2DS2-VASc of 1, which includes 65-74-year-olds with a CHADS2 score of 0. Resolving this conflicting advice is important, because these guidelines have potential to change anticoagulant therapy in 10 % of the AF population. METHODS: Using the National Registry of Atrial Fibrillation (NRAF) II data set, we compared these guidelines using stroke equivalents. Based on structured review of 23,657 patient records, we identified 65-74-year-old patients with a CHADS2 stroke score of 0 and no contraindication to warfarin. We used Medicare claims data to ascertain rates of ischemic stroke, intracranial hemorrhage, and other hemorrhage. We calculated net stroke equivalents for these (N = 478) patients using a weight of 1.5 for intracranial hemorrhages (ICH) and 1.0 for ischemic stroke. In a multivariate analysis, we used 14,466 records with documented atrial fibrillation and adjusted for CHADS2 and HEMORR2 HAGES score. RESULTS: In 65-74-year-old patients with a CHADS2 stroke score of 0, the stroke equivalents per 100 patient-years was 2.6 with warfarin and 2.9 without warfarin; the difference between these two strategies was not significant (0.3 stroke equivalents, 95 % CI -3.2 to 3.7). However, rates of hemorrhage per 100 patient-years were nearly tripled (hazard ratio 2.9; 95 % CI 1.5-5.4; p = 0.0011) with warfarin (21.1) versus without it (7.4). The most common site for major hemorrhage was gastrointestinal (ICD-9 code 578.9). CONCLUSIONS: By expanding warfarin use to 65--74-year-olds with a CHADS2 score of 0, rates of hemorrhages would rise without a significant reduction in stroke equivalents.

BACKGROUND: The 'New Forest Parenting Package' (NFPP), an 8-week home-based intervention for parents of preschoolers with attention-deficit/hyperactivity disorder (ADHD), fosters constructive parenting to target ADHD-related dysfunctions in attention and impulse control. Although NFPP has improved parent and laboratory measures of ADHD in community samples of children with ADHD-like problems, its efficacy in a clinical sample, and relative to an active treatment comparator, is unknown. The aims are to evaluate the short- and long-term efficacy and generalization effects of NFPP compared to an established clinic-based parenting intervention for treating noncompliant behavior ['Helping the Noncompliant Child' (HNC)] in young children with ADHD. METHODS: A randomized controlled trial with three parallel arms was the design for this study. A total of 164 3-4-year-olds, 73.8% male, meeting DSM-IV ADHD diagnostic criteria were randomized to NFPP (N = 67), HNC (N = 63), or wait-list control (WL, N = 34). All participants were assessed at post-treatment. NFPP and HNC participants were assessed at follow-up in the next school year. Primary outcomes were ADHD ratings by teachers blind to and uninvolved in treatment, and by parents. Secondary ADHD outcomes included clinician assessments, and laboratory measures of on-task behavior and delay of gratification. Other outcomes included parent and teacher ratings of oppositional behavior, and parenting measures. (Trial name: Home-Based Parent Training in ADHD Preschoolers; Registry: ClinicalTrials.gov Identifier: NCT01320098; URL: http://www/clinicaltrials.gov/ct2/show/NCT01320098). RESULTS: In both treatment groups, children's ADHD and ODD behaviors, as well as aspects of parenting, were rated improved by parents at the end of treatment compared to controls. Most of these gains in the children's behavior and in some parenting practices were sustained at follow-up. However, these parent-reported improvements were not corroborated by teacher ratings or objective observations. NFPP was not significantly better, and on a few outcomes significantly less effective, than HNC. CONCLUSIONS: The results do not support the claim that NFPP addresses putative dysfunctions underlying ADHD, bringing about generalized change in ADHD, and its underpinning self-regulatory processes. The findings support documented difficulties in achieving generalization across nontargeted settings, and the importance of using blinded measures to provide meaningful assessments of treatment effects.

An increasingly important goal of psychiatry is the use of brain imaging data to develop predictive models. Here we present two contributions to statistical methodology for this purpose. First, we propose and compare a set of wavelet-domain procedures for fitting generalized linear models with scalar responses and image predictors: sparse variants of principal component regression and of partial least squares, and the elastic net. Second, we consider assessing the contribution of image predictors over and above available scalar predictors, in particular via permutation tests and an extension of the idea of confounding to the case of functional or image predictors. Using the proposed methods, we assess whether maps of a spontaneous brain activity measure, derived from functional magnetic resonance imaging, can meaningfully predict presence or absence of attention deficit/hyperactivity disorder (ADHD). Our results shed light on the role of confounding in the surprising outcome of the recent ADHD-200 Global Competition, which challenged researchers to develop algorithms for automated image-based diagnosis of the disorder.