Russell Steele

McGill University, Montréal, Quebec, Canada

Are you Russell Steele?

Claim your profile

Publications (112)351.25 Total impact

  • [Show abstract] [Hide abstract]
    ABSTRACT: Robustness of an existing meta-analysis can justify decisions on whether or not to conduct an additional study addressing the same research question. We illustrate the graphical assessment of the potential impact of an additional study on an existing meta-analysis using published data on statin use and the risk of acute kidney injury. A previously proposed graphical augmentation approach is used to assess sensitivity of current test- and heterogeneity statistics extracted from existing meta-analysis data. In addition, we extended the graphical augmentation approach to assess potential changes in the pooled effect estimate after updating a current meta-analysis and applied the three graphical contour definitions to data from meta-analyses on statins use and acute kidney injury risk. In the considered example data, the pooled effect estimates and heterogeneity indices demonstrated to be considerably robust to the addition of a future study. Supportingly, for some previously inconclusive meta-analyses, a study update might yield statistically significant kidney injury risk increase associated with higher statins exposure. The illustrated contour approach should become a standard tool for the assessment of the robustness of meta-analyses. It can guide decisions on whether to conduct additional studies addressing a relevant research question. Copyright © 2015 Elsevier Inc. All rights reserved.
    Journal of clinical epidemiology 06/2015; DOI:10.1016/j.jclinepi.2015.05.030 · 5.48 Impact Factor
  • Mireille E. Schnitzer · Russell J. Steele · Michèle Bally · Ian Shrier
    [Show abstract] [Hide abstract]
    ABSTRACT: While standard meta-analysis pools the results from randomized trials that compare two specific treatments, network meta-analysis aggregates the results of randomized trials comparing a wider variety of treatment options. However, it is unclear whether the aggregation of effect estimates across heterogeneous populations will be consistent for a meaningful parameter. Drawing from counterfactual theory and the causal inference framework, we define the population of interest in a network meta-analysis and define the target parameter under a series of a priori nonparametric structural assumptions. This allows us to determine the requirements for identifiability of this parameter, enabling a description of the conditions under which network meta-analyses are feasible. We then propose several modeling strategies, including a novel Bayesian hierarchical random effects model, for consistent estimation of the intervention-specific mean outcome and model-independent contrasts between treatments.
  • Source
    I Shrier · R J Steele · M Zhao · A Naimi · E Verhagen · S D Stovitz · M J Rauh · T E Hewett
    [Show abstract] [Hide abstract]
    ABSTRACT: Physical activity is beneficial for many aspects of health but is associated with a risk of injury. Studies that assess causal risk factors of injury and reinjury provide valuable information to help develop and improve injury prevention programs. However, the underlying assumptions of analytical approaches often used to estimate causal factors in injury and subsequent injury research are often violated. This means that ineffective or even harmful interventions could be proposed because the underlying analyses produced unreliable or invalid causal effect estimates. We describe an adapted version of the multistate framework [multistate framework for the analysis of subsequent injury in sport (M-FASIS)] that makes investigator choices more transparent with respect to outcome and healing time. In addition, M-FASIS incorporates all previous sport injury analytical frameworks and accounts for injuries or conditions that heal or do not heal to 100%, acute and overuse injuries, illnesses, and competing event outcomes. © 2015 John Wiley & Sons A/S. Published by John Wiley & Sons Ltd.
    Scandinavian Journal of Medicine and Science in Sports 06/2015; DOI:10.1111/sms.12493 · 3.17 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: The aim of this study was to compare oral radiologic abnormalities associated with systemic sclerosis (SSc) against abnormalities in the general population. Patients with SSc and healthy controls were enrolled in a multi-site cross-sectional study. Included in the radiology examination were a panoramic radiograph, four bitewings, and an anterior mandibular periapical radiograph. Radiographs were evaluated by two oral and maxillofacial radiologists tested for interobserver and intraobserver reliability. Chi-squared tests, Fisher exact tests, and Mann Whitney U tests were used to summarize the radiologic manifestations of patients and controls. We assessed 163 SSc patients and 231 controls. Widening of the periodontal ligament space (PLS) (P < .001), with higher percentage of teeth with PLS widening (P < .001), was significantly more frequent in patients with SSc than in controls. The most significant differences between the two groups were found in the molars and premolars (P < .001). Moreover, 26% of the patients with SSc had a periapical PLS greater than 0.19 mm compared with 13% of the controls (P = .003). Patients with SSc had significantly more erosions compared with controls (14.5% vs. 3.6%; P < .001), mostly in the condyles (P = .022), coronoid processes (P = .005) and other locations (P = .012). Patients with SSc had more teeth with PLS widening and erosions of the mandible compared with controls. Copyright © 2015 Elsevier Inc. All rights reserved.
    Oral Surgery, Oral Medicine, Oral Pathology, Oral Radiology, and Endodontology 03/2015; DOI:10.1016/j.oooo.2015.03.002 · 1.46 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: Background Major depressive disorder (MDD) may be present in 10%–20% of patients in medical settings. Routine depression screening is sometimes recommended to improve depression management. However, studies of the diagnostic accuracy of depression screening tools have typically used data-driven, exploratory methods to select optimal cutoffs. Often, these studies report results from a small range of cutoff points around whatever cutoff score is most accurate in that given study. When published data are combined in meta-analyses, estimates of accuracy for different cutoff points may be based on data from different studies, rather than data from all studies for each possible cutoff point. As a result, traditional meta-analyses may generate exaggerated estimates of accuracy. Individual patient data (IPD) meta-analyses can address this problem by synthesizing data from all studies for each cutoff score to obtain diagnostic accuracy estimates. The nine-item Patient Health Questionnaire-9 (PHQ-9) and the shorter PHQ-2 and PHQ-8 are commonly recommended for depression screening. Thus, the primary objectives of our IPD meta-analyses are to determine the diagnostic accuracy of the PHQ-9, PHQ-8, and PHQ-2 to detect MDD among adults across all potentially relevant cutoff scores. Secondary analyses involve assessing accuracy accounting for patient factors that may influence accuracy (age, sex, medical comorbidity). Methods/design Data sources will include MEDLINE, MEDLINE In-Process & Other Non-Indexed Citations, PsycINFO, and Web of Science. We will include studies that included a Diagnostic and Statistical Manual or International Classification of Diseases diagnosis of MDD based on a validated structured or semi-structured clinical interview administered within 2 weeks of the administration of the PHQ. Two reviewers will independently screen titles and abstracts, perform full article review, and extract study data. Disagreements will be resolved by consensus. Risk of bias will be assessed with the Quality Assessment of Diagnostic Accuracy Studies-2 tool. Bivariate random-effects meta-analysis will be conducted for the full range of plausible cutoff values. Discussion The proposed IPD meta-analyses will allow us to obtain estimates of the diagnostic accuracy of the PHQ-9, PHQ-8, and PHQ-2. Systematic review registration PROSPERO CRD42014010673
    10/2014; 3(1):124. DOI:10.1186/2046-4053-3-124
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: Objective Systemic sclerosis (SSc; scleroderma) is associated with decreased saliva production and interincisal distance, more missing teeth, and periodontal disease. We undertook this study to determine the clinical correlates of SSc with these oral abnormalities.Methods Subjects were recruited from the Canadian Scleroderma Research Group cohort. Detailed dental and clinical examinations were performed according to standardized protocols. Associations between dental abnormalities and selected clinical and serologic manifestations of SSc were examined.ResultsOne hundred sixty-three SSc subjects were included: 90% women, mean ± SD age 56 ± 11 years, mean ± SD disease duration 14 ± 8 years, 72% with limited cutaneous disease, and 28% with diffuse cutaneous disease. Decreased saliva production was associated with Sjögren's syndrome-related autoantibodies (β = −43.32; 95% confidence interval [95% CI] −80.89, −5.75), but not with disease severity (β = −2.51; 95% CI −8.75, 3.73). Decreased interincisal distance was related to disease severity (β = −1.02; 95% CI −1.63, −0.42) and the modified Rodnan skin thickness score (β = −0.38; 95% CI −0.53, −0.23). The number of missing teeth was associated with decreased saliva production (relative risk [RR] 0.97; 95% CI 0.94, 0.99), worse hand function (RR 1.52; 95% CI 1.13, 2.02), and the presence of gastroesophageal reflux disease (GERD; RR 1.68 [95% CI 1.14, 2.46]). No clinical or serologic variables were correlated with periodontal disease.Conclusion In SSc, diminished interincisal distance is related to overall disease severity. Decreased saliva production is related to concomitant Sjögren's syndrome antibodies. Tooth loss is associated with poor upper extremity function, GERD, and decreased saliva. The etiology of excess periodontal disease is likely multifactorial and remains unclear.
    10/2014; 67(5). DOI:10.1002/acr.22490
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: Objective. Both oral and global health-related quality of life (HRQoL) are markedly impaired in SSc. In this study we aimed to determine the degree of association between oral HRQoL and global HRQoL in SSc. Methods. Subjects were recruited from the Canadian Scleroderma Research Group registry. Global HRQoL was measured using the Medical Outcomes Trust 36-item Short Form Health Survey (SF-36) and oral HRQoL with the Oral Health Impact Profile (OHIP). The Medsger Disease Severity Score was used to determine organ involvement. Multivariate regression models determined the independent association of the OHIP with the SF-36 after adjusting for confounders. Results. This study included 156 SSc subjects. The majority (90%) were women, with a mean age of 56 years, mean disease duration 13.8 years (s.d. 8.5) and 29% of the subjects had dcSSc. Mean total OHIP score was 40.8 (s.d. 32.4). Mean SF-36 mental component summary (MCS) score was 49.7 (s.d. 11.1) and physical component summary (PCS) score was 37.0 (s.d. 10.7). In adjusted analyses, the total OHIP score was significantly associated with the SF-36 MCS and PCS, accounting for 9.7% and 5.6% of their respective variances. Measures of disease severity were not related to OHIP score. Conclusion. Oral HRQoL in SSc is independently associated with global HRQoL. Oral HRQoL, however, is not related to physician-assessed disease severity. This suggests that physicians may be disregarding issues related to oral health. HRQoL is an additional dimension of HRQoL not captured by generic instruments such as the SF-36.
    Rheumatology (Oxford, England) 10/2014; 54(4). DOI:10.1093/rheumatology/keu396 · 4.44 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: Projection analyses can provide estimates of the future health burden of increasing BMI and represent a relevant and useful tool for public health planning. Our study presents long-term (2013-2030) projections of the prevalence and numbers of individuals by BMI category for adult men and women in Quebec. Three applications of projections to estimate outcomes more directly pertinent to public health planning as well as an in-depth discussion of limits are provided with the aim of encouraging greater use of projection analyses by public health officers.
    BMC Public Health 09/2014; 14(1):996. DOI:10.1186/1471-2458-14-996 · 2.32 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: The purpose of this study was to validate a recently proposed return-to-play (RTP) decision model that simplifies the complex process into three underlying constructs: injury type and severity, sport injury risk, and factors unrelated to injury risk (decision modifiers). We used a cross-over design and provided clinical vignettes to clinicians involved in RTP decision making through an online survey. Each vignette included examples changing injury severity, sport risk (e.g. different positions), and non-injury risk factors (e.g. financial considerations). As the three-step model suggests, clinicians increased restrictions as injury severity increased, and also changed RTP decisions when factors related to sport risk and factors unrelated to sport risk were changed. The effect was different for different injury severities and clinical cases, suggesting context dependency. The model was also consistent with recommendations made by subgroups of clinicians: sport medicine physicians, non-sport medicine physicians, and allied health care workers.
    Scandinavian Journal of Medicine and Science in Sports 09/2014; 25(2). DOI:10.1111/sms.12306 · 3.17 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: The aim of this cluster randomized trial was to evaluate the effectiveness of a school-based theatre intervention program for immigrant and refugee youth in special classes for improving mental health and academic outcomes. The primary hypothesis was that students in the theatre intervention group would report a greater reduction in impairment from symptoms compared to students in the control and tutoring groups.
    PLoS ONE 08/2014; 9(8):e104704. DOI:10.1371/journal.pone.0104704 · 3.23 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: Objectives. To compare paramedics’ ability to minimize cervical spine motion during patient transfer onto a vacuum mattress with two stabilization techniques (head squeeze vs. trap squeeze) and two transfer methods (log roll with one assistant (LR2) vs. 3 assistants (LR4)). Methods. We used a crossover design to minimize bias. Each lead paramedic performed 10 LR2 transfers and 10 LR4 transfers. For each of the 10 LR2 and 10 LR4 transfers, the lead paramedic stabilized the cervical spine using the head squeeze technique five times and the trap squeeze technique five times. We randomized the order of the stabilization techniques and LR2/LR4 across lead paramedics to avoid a practice or fatigue effect with repeated trials. We measured relative cervical spine motion between the head and trunk using inertial measurement units placed on the forehead and sternum. Results. On average, total motion was 3.9° less with three assistants compared to one assistant (p = 0.0002), and 2.8° less with the trap squeeze compared to the head squeeze (p = 0.002). There was no interaction between the transfer method and stabilization technique. When examining specific motions in the six directions, the trap squeeze generally produced less lateral flexion and rotation motion but allowed more extension. Examining within paramedic differences, some paramedics were clearly more proficient with the trap squeeze technique and others were clearly more proficient with the head squeeze technique. Conclusion. Paramedics performing a log roll with three assistants created less motion compared to a log roll with only one assistant, and using the trap squeeze stabilization technique resulted in less motion than the head squeeze technique but the clinical relevance of the magnitude remains unclear. However, large individual differences suggest future paramedic training should incorporate both best evidence practice as well as recognition that there may be individual differences between paramedics.
    Prehospital Emergency Care 07/2014; 19(1). DOI:10.3109/10903127.2014.936634 · 1.81 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: Objective To examine the association between anti-Ro antibodies, namely anti-Ro60/SS-A and anti-Ro52/TRIM21, together and separately, and a prolonged QT interval corrected for heart rate (QTc) in systemic sclerosis (SSc) patients. Methods 689 SSc patients enrolled in a multi-center cohort study underwent a 12-lead resting EKG at baseline. The QTc interval was measured and a QTc ≥ 440 ms was considered prolonged. Detailed clinical data and sera for these patients were collected and positivity for anti-Ro60/SS-A and anti-Ro52/TRIM21 antibodies was determined using an addressable laser-bead immunoassay (ALBIA). Results QTc prolongation was common in this SSc cohort (25%). In univariate analysis, Ro antibodies, together or separately, were not associated with prolongation of the QTc interval (mean difference in QTc in anti-Ro antibody positive versus negative subjects -2.2 ms (p=0.5748), in anti-Ro60/SS-A antibody positive versus negative subjects 1.3 ms (p=0.8616) and in anti-Ro52/TRIM21 antibody positive versus negative subjects -3.3 ms (p=0.4106)). In multivariate logistic regression analysis adjusting for possible confounders, there was no association between prolonged QTc and anti-Ro antibodies (odds ratio (OR) 0.74, 95% confidence interval (CI) 0.45, 1.22), anti-Ro60/SS-A antibodies (OR 1.57, 95% CI 0.72, 3.41) and anti-Ro52/TRIM21 antibodies (OR 0.76, 95% CI 0.46, 1.26). However, in both univariate and multivariate analyses, QTc prolongation was associated with longer disease duration, greater disease severity and the presence of anti-RNA polymerase III antibodies. Conclusions QTc prolongation is common in SSc, although anti-Ro antibodies do not seem to be associated with it as is the case in systemic lupus erythematosus. The reasons for this difference as well as the cause of abnormalities in cardiac repolarization in SSc will require additional studies.
    Seminars in Arthritis and Rheumatism 07/2014; 44(3). DOI:10.1016/j.semarthrit.2014.07.001 · 3.63 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: Background Research on psychological risk factors for injury has focused on stable traits. Our objective was to test the feasibility of a prospective longitudinal study designed to examine labile psychological states as risk factors of injury. Methods We measured psychological traits at baseline (mood, ways of coping and anxiety), and psychological states every day (1-item questions on anxiety, sleep, fatigue, soreness, self-confidence) before performances in Cirque du Soleil artists of the show “O”. Additional questions were added once per week to better assess anxiety (20-item) and mood. Questionnaires were provided in English, French, Russian and Japanese. Injury and exposure data were extracted from electronic records that are kept as part of routine business practices. Results The 43.9% (36/82) recruitment rate was more than expected. Most artists completed the baseline questionnaires in 15 min, a weekly questionnaire in <2 min and a daily questionnaire in <1 min. We improved the formatting of some questions during the study, and adapted the wording of other questions to improve clarity. There were no dropouts during the entire study, suggesting the questionnaires were appropriate in content and length. Results for sample size calculations depend on the number of artists followed and the minimal important difference in injury rates, but in general, preclude a purely prospective study with daily data collection because of the long follow-up required. However, a prospective nested case-crossover design with data collection bi-weekly and at the time of injury appears feasible. Conclusion A prospective study collecting psychological state data from subjects who train and work regularly together is feasible, but sample size calculations suggest that the optimal study design would use prospective nested case-crossover methodology.
    BMC Medical Research Methodology 06/2014; 14(1):77. DOI:10.1186/1471-2288-14-77 · 2.17 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: A recent return to play decision-making (RTP-DM) model for sport medicine has organized the underlying concepts into 3 steps but has not yet been validated. To examine the validity of the 3-step RTP-DM model recently proposed. Repeated measures cross-over survey design. World-wide. American College of Sports Medicine clinicians involved in RTP-DM. We provided clinical vignettes of injuries and illnesses in athletes to participants through an online survey. Each vignette included examples of 3 factor types: increasing injury severity, changing risk associated with sport (e.g. different positions), and changing non-injury risk factors (e.g. financial considerations). For each vignette, participants indicated the level of activity restriction they would recommend (6 options from No Restrictions to No Activity) in accordance with the risk they placed on continued participation. We analyzed the data using multiple regression, adjusting for the correlated participant outcomes, to measure how changes in factors affected individual participants. The estimated participation rate for those involved in RTP decisions was 24.7%. As expected, we found that clinicians increase restrictions as injury severity increases. We also found that changing factors related to sport risk, and changing factors that are unrelated to sport risk will affect RTP decisions, although the effect is context-dependent and does not occur equally across all injury severities and clinical cases. The effect was also observed in each subgroup examined that included sex, age, specialty, region of training, academic status, and years of experience making RTP decisions. Our findings that clinicians from a wide variety of backgrounds will change RTP recommendations based on clinical vignettes with changing injury severity, sport risk modifiers and decision modifiers provides evidentiary support for the 3-step model for RTP decision making recently proposed.
    British Journal of Sports Medicine 04/2014; 48(7):661. DOI:10.1136/bjsports-2014-093494.270 · 5.03 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: Objective. The aim of this study was to compare oral abnormalities and oral health-related quality of life (HRQoL) of patients with SSc with the general population.Methods. SSc patients and healthy controls were enrolled in a multisite cross-sectional study. A standardized oral examination was performed. Oral HRQoL was measured with the Oral Health Impact Profile (OHIP). Multivariate regression analyses were performed to identify associations between SSc, oral abnormalities and oral HRQoL.Results. We assessed 163 SSc patients and 231 controls. SSc patients had more decayed teeth (SSc 0.88, controls 0.59, P = 0.0465) and periodontal disease [number of teeth with pocket depth (PD) >3 mm or clinical attachment level (CAL) ≥5.5 mm; SSc 5.23, controls 2.94, P < 0.0001]. SSc patients produced less saliva (SSc 147.52 mg/min, controls 163.19 mg/min, P = 0.0259) and their interincisal distance was smaller (SSc 37.68 mm, controls 44.30 mm, P < 0.0001). SSc patients had significantly reduced oral HRQoL compared with controls (mean OHIP score: SSc 41.58, controls 26.67, P < 0.0001). Multivariate regression analyses confirmed that SSc was a significant independent predictor of missing teeth, periodontal disease, interincisal distance, saliva production and OHIP scores.Conclusion. Subjects with SSc have impaired oral health and oral HRQoL compared with the general population. These data can be used to develop targeted interventions to improve oral health and HRQoL in SSc.
    Rheumatology (Oxford, England) 01/2014; 53(8). DOI:10.1093/rheumatology/ket441 · 4.44 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: The present study involved an in vitro examination of spinal cord interstitial pressure (CIP) during distraction before and after durotomy in three spinal cord segments obtained from five pigs. To determine whether durotomy can be used to decrease the elevated CIP associated with spinal cord distraction. Spinal cord distraction is a known cause of spinal cord injury. Several articles describing the pathophysiology of cord distraction injuries suggest that the underlying mechanism of injury is a microvascular ischemic event. The authors have previously described an increase in CIP with spinal cord distraction, with average pressures of 23 mmHg at loads of 1,000 g. To date, there are no published studies that have evaluated the efficacy of intentional durotomies as a treatment for elevated CIP. A total of 15 spinal cord sections were harvested from pigs and distracted while immersed in saline, using a fixed 1,000 g distraction force. The CIP decay was then measured at 30-s intervals for 10 min. The distraction/relaxation maneuver was performed six times with continuous CIP monitoring and was subsequently followed by durotomy. The pressure-decay curves were similar for each specimen, but varied according to individual pigs and anatomical levels. CIP decayed over the first 4 min of distraction and remained constant for the final 6 min. Longitudinal durotomy led to a dramatic drop in CIP toward baseline and appeared to be as effective as transverse durotomy with regard to the normalization of pressure. Spinal cord distraction causes elevations in CIP. Durotomy lowers elevated CIP in vitro and may be effective at lowering CIP in vivo. Further study is required to evaluate the usefulness of durotomy in vivo.
    European Journal of Orthopaedic Surgery & Traumatology 01/2014; 24(S1). DOI:10.1007/s00590-013-1409-5 · 0.18 Impact Factor
  • Open Journal of Rheumatology and Autoimmune Diseases 01/2014; 04(01):52-57. DOI:10.4236/ojra.2014.41008
  • [Show abstract] [Hide abstract]
    ABSTRACT: The objectives of this study were to develop a standard classification of digital ulcers (DUs) in systemic sclerosis (SSc) for use in observational or therapeutic studies and to assess the reliability of these definitions as well as of the measurement of ulcer area. Ten North American rheumatologists with expertise in SSc reviewed multiple photos of DUs, examined four SSc subjects with DUs, and came to a consensus on the definitions for digital, active, healed, and indeterminate ulcers. These ten raters then examined the right hand of ten SSc subjects twice and the left hand once to classify ulcers and to measure ulcer area. Weighted and Fleiss kappa were used to calculate intra- and interrater agreement on classification of ulcers, and intraclass correlation coefficient (ICC) was used to assess agreement on ulcer area. Because the traditional ICC calculations relied on a small number of ulcers, ICCs were recalculated using the results of linear mixed models to evaluate the variance components of observations on all the data. Intrarater kappa for classifying DU as not an ulcer/healed ulcer versus active/indeterminate ulcer was substantial (0.76), and interrater kappa was moderate (0.53). The ICC for ulcer area using the linear mixed models was moderate both for intrarater (0.57) and interrater (0.48) measurements. A consensus for the classification of DUs in SSc was developed, and after a training session, rheumatologists with expertise in SSc are able to reliably classify DUs and to measure ulcer area.
    Clinical Rheumatology 12/2013; DOI:10.1007/s10067-013-2460-7 · 1.77 Impact Factor
  • Yeting Du · Abbas Khalili · Johanna G. Nešlehová · Russell J. Steele
    [Show abstract] [Hide abstract]
    ABSTRACT: In this article, we study finite mixtures of linear mixed-effects (FMLME) models that are useful for longitudinal regression modelling in the presence of heterogeneity in both fixed and random effects. These models are computationally challenging when the number of covariates is large, and traditional variable selection techniques become expensive to implement. We introduce a penalized likelihood approach, and propose a nested EM algorithm for efficient numerical computations. The resulting estimators are shown to possess consistency and sparsity properties, and to be asymptotically normally distributed. We illustrate the performance of our method through simulations and a real data example. The Canadian Journal of Statistics 41: 596–616; 2013 © 2013 Statistical Society of Canada Résumé Dans cet article, les auteurs étudient des mélanges finis de modèles linéaires à effets mixtes, qui sont utiles pour la régression de données longitudinales en présence d'hétérogénéité des composantes fixes et aléatoires. Lorsque le nombre de covariables est élevé, ces modèles sont exigeants sur le plan calculatoire et les techniques classiques de sélection de variable deviennent laborieuses. Les auteurs présentent une approche de vraisemblance pénalisée et proposent un algorithme EM emboî té afin de procéder aux calculs numériques de façon efficace. Ils démontrent les propriétés de convergence, d’éparpillement et de normalité asymptotique des estimateurs. Ils illustrent leur méthode à l'aide de données simulées et réelles. La revue canadienne de statistique 41: 596–616; 2013 © 2013 Société statistique du Canada
    Canadian Journal of Statistics 12/2013; 41(4). DOI:10.1002/cjs.11192 · 0.70 Impact Factor
  • Ian Shrier · Jay S Kaufman · Robert W Platt · Russell J Steele
    The International Journal of Biostatistics 11/2013; 9(2):307-313. DOI:10.1515/ijb-2013-0045 · 1.28 Impact Factor

Publication Stats

1k Citations
351.25 Total Impact Points

Institutions

  • 2005–2015
    • McGill University
      • Department of Mathematics and Statistics
      Montréal, Quebec, Canada
  • 2014
    • The University of Western Ontario
      • Department of Mathematics
      London, Ontario, Canada
  • 2009–2013
    • Lady Davis Institute for Medical Research
      • Centre for Clinical Epidemiology (CCE)
      Montréal, Quebec, Canada
  • 2007–2013
    • Jewish General Hospital
      Montréal, Quebec, Canada
  • 2011
    • Université de Sherbrooke
      • Department of Surgery
      Sherbrooke, Quebec, Canada
  • 2001
    • University of Washington Seattle
      Seattle, Washington, United States