Mohamed Abdolell

ABSTRACT To evaluate patient radiation dose from the most frequently performed CT examinations and to establish provincial diagnostic reference levels (DRLs) as a tool for protocol optimization. The study investigated the following CT examinations: head, chest, abdomen/pelvis, and chest/abdomen/pelvis (CAP). Dose data, volume CT dose index (CTDIvol) and dose-length product (DLP), were collected from 15 CT scanners installed during 2004-2014 in 11 hospital sites of Nova Scotia. All scanners had dose modulation options and multislice capability (16-128 detector rows). The sample for each protocol included 15 average size patients (70±20 kg). Provincial DRLs were calculated as the 75th percentile of patient dose distributions. The differences in dose between hospitals were evaluated with a single factor ANOVA statistical test. Generalized linear modeling was used to determine the factors associated with higher radiation dose. A sample of 36 abdominal studies performed on three different scanners was blinded and randomized for an assessment by an experienced radiologist who graded the imaging quality of anatomic structures. Data for 900 patients were collected. The DRLs were proposed using CTDIvol (mGy) and DLP (mGy*cm) values for CT head (67 and 1049, respectively), chest (12 and 393), abdomen/pelvis (16 and 717), and CAP (14 and 1034). These DRLs were lower than the published national data except for the head CTDIvol. The differences between the means of the dose distributions from each scanner were statistically significant (p<0.05) for all examinations. A very weak correlation was found between the dose and the scanner age or the number of slices with Pearson's correlation coefficients of 0.011-0.315. The blinded analysis of image quality demonstrated no clinically significant difference except for the noise category. Provincial DRLs were established for typical CT examinations. The variations in dose between the hospitals suggested a large potential for optimization of examinations. Radiology Research Foundation grant.

CONCLUSION We have developed a curriculum for an introductory course in biostatistics and epidemiology that is used to support in-service teaching with radiology residents. This course enables students to set their own pace, place and time of learning, while fostering critical appraisal skills via modern web-based applications. BACKGROUND Biostatistics and Epidemiology provide a rigorous framework for evaluating quantitative evidence in the health sciences and lend themselves to computer-based teaching, yet hospital-based research methods courses have traditionally achieved limited success, particularly when delivered as in-class teaching initiatives. The overwhelming clinical demands of radiology residency programs often supersede residents’ research activities and residents must therefore be engaged in and perceive the value of a research methods curriculum that relates to their clinical training. DISCUSSION The common requirement across radiology residency programs for residents ...

ABSTRACT The aim of this study was to quantify the impact of positron emission tomography-computed tomography (PET-CT) on clinical target volume (CTV) selection in non-small cell lung cancer (NSCLC) and head and neck squamous cell cancer (HNSCC) cancer patients.

We report our center's experience with the Toronto Western Hospital (TWH) catheter, and discuss our catheter survival and complication rates. Retrospective chart review of patients receiving peritoneal dialysis therapy via a TWH catheter. Catheter complication rates of peritonitis, exit site infection, obstruction, leak, and malfunction were assessed. A catheter was considered failed if removed because of exit site infection, obstruction, or malfunction. All other catheters, even if removed for other reasons, were considered censured. Survival was defined as the period from insertion to failure or censure date, and reported using Kaplan Meier analysis. 192 patients with a total of 208 TWH catheters (4,845.3 catheter months) were analyzed. Our overall 1- and 3-year catheter survival rates were identical at 0.9182. Our catheter complication rates (expressed as number of catheter months per event) were 31.3 for peritonitis, 42.9 for exit site infection, 72.3 for obstruction, 538.4 ...

The objective of the study was to determine possible differences in perception of quality of life (QoL) between patients with metastatic breast or prostate cancer, their partners, and the treating physician. Patients with metastatic breast cancer (n = 71), and metastatic prostate cancer (n = 29), a partner, and the physician each completed the same QoL questionnaire indicating how they perceived the patient's QoL. The European organization for research and treatment of cancer (EORTC) QLQ-C30 questionnaire was used to assess patients with breast cancer and the modified prostate cancer specific quality of life instrument (PROSQOLI) for patients with prostate cancer. There was reasonable agreement in mean scores between patients, and physicians or partners, for many domains of QoL; however, there was substantial discordance between scores when considering individual patients. For patients with metastatic breast cancer, physicians systematically underestimated overall QoL (p = 0.000...

Previous analyses of interval breast cancers have been limited because of a lack of control for screening interval length and patient age, failure to restrict the interval group to 'true' intervals, and incomplete descriptions of pathology, adjuvant therapies and clinical outcomes. A nested case-control study within the population-based Nova Scotia Breast Screening Program was performed. All true interval cases between 1991 and 2004 were identified, matched 1:2 to screen-detected cases (age, screening interval, time period), and compared in terms of pathologic characteristics and adjuvant therapies via logistic regression. Disease-free and overall survival was estimated, controlling for pathology and adjuvant chemotherapy receipt. A total of 241 true interval invasive cases were matched to 481 screen-detected cases. Interval cases were more likely to be > 1 cm (odds ratio [OR] = 1.76; 95% CI, 1.10-2.83), grade 3 (OR = 2.71; 95% CI, 1.49-4.92), and have lymphovascular invasion (OR = 3.06; 95% CI, 1.85-5.07). Interval cases received more adjuvant chemotherapy (OR = 4.37; 95% CI, 3.03-6.30) and radiation (OR = 1.43; 95% CI, 1.02-2.00). The 5-year Kaplan-Meier estimates of disease-free and overall survival rates for true intervals and screens were 0.830 (95% CI, 0.770-0.875) versus 0.926 (95% CI, 0.898-0.947) and 0.860 (95% CI, 0.804-0.901) versus 0.937 (95% CI, 0.910-0.956), respectively. True interval breast cancers have more adverse prognostic factors compared with screen-detected cases and, despite receiving more adjuvant chemotherapy, are associated with significantly poorer survival outcomes.

The objective of this study was to assess whether caregivers serve as valid and reliable proxies for the assessment of a patient's pain level using the Brief Pain Inventory (BPI) assessment tool. Twenty-seven patients with cancer with pain, referred for health care management to the Toronto Sunnybrook Regional Cancer Centre and the Sunnybrook and Women's College Health Sciences Centre, and their proxies participated in this study. Patient and proxy completed the BPI independently, with proxies given explicit instructions to rate and answer the questions from the perspective of the patient. Patient and proxy also completed a data extraction form to obtain patient/proxy characteristics such as sex, age, primary cancer site, and relationship. An intraclass correlation coefficient (ICC) was performed for each question of the BPI. In this study, we found poor to good agreement for questions that are subjective (eg, average pain), whereas the highest agreement was for worst pain scoring (ICC = 0.72). Poor to moderate agreement was found for questions that are objective (eg, walking ability), suggesting that the functional interference questions of the BPI may be too vague for proxies to respond reliably. The findings indicate that proxies and patients agree well on worst pain levels but agree much less on the effectiveness of pain medications and patients' average level of pain. Further investigations should examine other tools for pain assessment that may be more reliably used by a proxy.

An association has been reported between thrombotic events and the use of L-asparaginase (ASP) in children with acute lymphoblastic leukaemia (ALL). The mechanism for thrombosis is likely related to an acquired antithrombin deficiency. Since a primary prophylaxis using antithrombin concentrates may prevent thrombosis, the PARKAA (Prophylactic Antithrombin replacement in kids with ALL treated with L-asparaginase) study was performed. The objectives of PARKAA were to determine if there was a trend to efficacy and safety of antithrombin treatment as assessed by 1) incidence of thrombosis 2) incidence of bleeding and 3) plasma markers of endogenous thrombin generation as surrogate outcomes for thrombosis. The study was not powered to answer the question of efficacy and safety, but rather to detect a trend. PARKAA was an open, randomised, controlled study in children with ALL being treated with ASP. Children were randomised to receive antithrombin infusions or no antithrombin treatment. All thrombotic events were confirmed using bilateral venography, ultrasound, echocardiography and MRI. The incidence of thrombosis in patients treated with antithrombin was 28% (95% CI 10-46%), compared to 37% (95% CI 24-49%) in the non treated arm. Two minor bleeds occurred in patients in the treated arm, but were not considered to be related to antithrombin. No significant differences were seen in plasma markers by the treatment group. In conclusion, treatment with antithrombin concentrate shows a trend to efficacy and safety. In contrast, there was no difference in surrogate markers for thrombosis. Carefully designed clinical trials are needed to test the efficacy and safety of antithrombin in this population.

To describe the pharmacokinetic disposition of tobramycin in children undergoing stem cell transplantation (SCT) after intravenous administration either every 24 hours or every 8 hours, and to use this information to create initial dosing guidelines for administration every 24 hours in this patient population. Pharmacokinetic analysis of a randomized controlled trial. The Hospital for Sick Children, Toronto, Ontario, Canada. Sixty children (< 18 yrs) with febrile neutropenia undergoing stem cell transplantation. Patients were randomized to receive intravenous tobramycin either every 24 hours (29 patients) or every 8 hours (31 patients). Initially, they received either 2.5 mg/kg/dose every 8 hours or weight-based doses by age group (< 5 yrs, 9 mg/kg/dose; 5 to < 12 yrs, 8 mg/kg/dose;…

Our objectives were to describe the frequency and determine risk factors for hearing deterioration following pediatric stem cell transplantation (SCT). In this prospective cohort study, we performed pure tone audiometry and measured otoacoustic emissions (OAE) prior to and following SCT. Worse hearing was considered deterioration in either audiometry or OAE. Between October 2000 and November 2002, 45 informative audiometry or OAE results were obtained. Hearing deteriorated following SCT in 20/45 (44%) of these children. Those with worse hearing following SCT were more likely to have neuroblastoma (odds ratio [OR] 16.0 [95% CI 1.8, 143.2; P = 0.003]), receive carboplatin conditioning (OR 7.7 [95% CI 1.4, 41.9; P = 0.01]), have abnormal baseline hearing (OR 5.1 [95% CI 1.3, 19.5; P = 0.02]), and have higher baseline serum creatinine (OR for every increase of 5 micromol/L of serum creatinine of 1.5 [95% CI 1.03, 2.1; P = 0.03]). Many children who undergo SCT will have deterioration in hearing following SCT. A high-risk group of children can be delineated who may benefit from more intensive audiological monitoring following SCT.

Post-thrombotic syndrome (PTS) in adults, characterized by swelling, skin pigmentation, pain, and ulceration of the limb, is secondary to deep vein thrombosis (DVT). In contrast to the extensive documentation on PTS in adults, little is known about the risk of PTS in children. To determine the incidence, clinical characteristics, and predictors of PTS in children. A cross-sectional study in 153 nonselected children with objectively confirmed DVT. All children were assessed for PTS using a standardized score. As per the PTS score, severity was classified as: absent, mild, moderate, or severe. Post-thrombotic syndrome was present in 96/153 children (63%), in which 80 (83%) were mild and 16 (17%) were moderate. Swelling was the most frequently recorded subjective symptom (43%) while increased limb circumference (71%) and presence of collateral circulation (53%) were the most frequently recorded objective symptoms. Risk factors for development of PTS were: lack of resolution of the DVT by radiographic assessment (OR 3.96, 95% CI 1.68-9.30), number of vessels involved in the initial DVT (OR 2.05, 95% CI 1.52-2.77), and length of follow-up (OR 1.22, 95% CI 1.08-1.39). These findings demonstrate that PTS is a clinically significant disease in children with previous DVT.

ABSTRACT Purpose. To evaluate the effect of positioning errors from setup and from prostate motion during escalated dose conformal prostate irradiation for localised prostate cancer. Methods. Thirteen patients with localised prostate cancer had lateral portal films taken three times weekly during escalated dose conformal prostate radiotherapy. Field placement errors were measured by matching corresponding bony landmarks to the simulator film. Prostate motion at the base, mid-gland and apex was measured by imaging the displacement of 3 gold fiducial markers implanted into the prostate before therapy. From these measurements a planning target volume (PTV) was derived that allowed for both the prostate motion and field placement errors.

To derive and validate a simple predictive model for survival of patients with metastatic cancer attending a palliative radiotherapy clinic. We described previously a model predicting survival of patients referred for palliative radiotherapy using six prognostic factors: primary cancer site, site of metastases, Karnofsky performance score (KPS), and the fatigue, appetite, and shortness of breath subscales from the Edmonton Symptom Assessment Scale. Here we simplified the model to include only three factors: primary cancer site, site of metastases, and KPS. Each factor was assigned a value proportional to its prognostic weight, and the weighted scores for each patient were summed to obtain a survival prediction score (SPS). Patients were also grouped according to their number of risk factors (NRF): nonbreast cancer, metastases other than bone, and KPS < or = 60. The three- and six- variable models were evaluated for their ability to predict survival in patients referred during a different time period and of those referred to a different cancer center. A training set of 395 patients, a temporal validation set of 445 patients, and an external validation set of 467 patients were used. The ability of the three- and six-variable models to separate patients into three prognostic groups and to predict their survival was similar using both SPS and NRF methods in the training, temporal, and external validation data sets. There was no statistically significant difference in the performance of the models. The three-variable NRF model is preferred because of its relative simplicity.

Benchmark analysis of patients with chronic myeloid leukemia (CML) alive for more than 10 years after allogeneic bone marrow transplantation (BMT) including data on disease status, bone marrow reserve, long-term complications, and quality of life (QOL). Eighty-nine patients (46 in first chronic phase, 43 in advanced phase) received an allogeneic BMT for CML during the study period. Medical outcomes and QOL of patients were analyzed retrospectively. Twenty-eight (31.5%) of 89 patients were alive at 10 years and included in this analysis. Thirteen (46.4%) of 28 long-term survivors never relapsed. Fifteen patients relapsed between 0.5 and 16 years after transplantation. Ten patients showed a hematologic relapse and received salvage treatment. Five patients showed transient low levels of BCR-ABL-positive cells by Southern blot with no subsequent hematologic relapse. One of the 28 patients died in blast crisis at 12 years. The most frequent long-term complications were chronic graft-versus-host disease, osteoporosis, and cataracts. Frequency of clonogenic progenitors remained persistently decreased. QOL assessment yielded lower scores in physical performance as compared with an age-matched normative population, whereas social functioning was equivalent. A high degree of satisfaction was noted with interpersonal relationships. Patients with CML surviving their BMT long term do well in terms of medical outcomes. A constant rate of relapse was noted, with a high salvage rate of affected patients, suggesting the need for lifelong monitoring. QOL is perceived as good, particularly as related to social functioning; however, it is inferior to a normative population with regard to physical performance.

To review the association of an abnormal prenatal sonogram with most recent serum creatinine in patients with proven posterior urethral valves (PUV). Since 1992, all live-born patients between 1992-2004 with clinically proven PUV, with postnatally proven PUV, from 2 pediatric tertiary care centers, were reviewed for age at diagnosis, most recent serum creatinine, presence of chronic renal failure (CRF) (serum creatinine >2 standard deviations above normal for age), or end stage renal disease (dialysis or transplant). Available antenatal reports from the 2 centres and surrounding community hospitals were reviewed for gestational age (GA) at the time of ultrasound, volume of amniotic fluid, and urinary-tract abnormality. Thirty-four patients with proven PUV and prenatal sonograms were identified (1992-2004). Eighteen patients had abnormalities on their prenatal sonogram, with poor outcome in 5 (mean follow-up, 8 years [1-13 y]). No specific features were identified on prenatal sonogram. Sixteen patients had normal prenatal sonograms, with poor outcomes in 2 (mean follow-up, 8 years [3-13y]). There is an increased risk of an abnormal serum creatinine among those patients with an abnormal prenatal study, odds ratio (OR) 2.6 (95% confidence interval, 0.35-32). PUV represents a spectrum of disease severity. A normal prenatal ultrasound does not preclude PUV. The majority of patients with a normal prenatal examination have good outcomes. The OR suggests that there may be increased risk for poor outcome in those with an abnormal prenatal examination. A multicenter study is necessary to obtain a larger sample size and more precise ORs.

Various clinical risk factors, including high breast density, have been shown to be associated with breast cancer. The utility of using relative and absolute area-based breast density-related measures was evaluated as an alternative to clinical risk factors in cancer risk assessment at time of screening mammography. Contralateral mediolateral oblique digital mammography images from 392 women with unilateral breast cancer and 815 age-matched controls were analyzed. Information on clinical risk factors was obtained from the provincial breast imaging information system. Breast density-related measures were assessed using a fully-automated breast density measurement software. Multivariable logistic regression was conducted and Area Under the Receiver Operating Characteristic (AUROC) curve was used to evaluate the performance of three cancer risk models: the first using only clinical risk factors, the second using only density-related measures, and the third using both clinical risk fact...

Discussions of percent breast density (PD) and breast cancer risk implicitly assume that visual assessments of PD are comparable between vendors despite differences in technology and display algorithms. This study examines the extent to which visual assessments of PD differ between mammograms acquired from two vendors. Pairs of "for presentation" digital mammography images were obtained from two mammography units for 146 women who had a screening mammogram on one vendor unit followed by a diagnostic mammogram on a different vendor unit. Four radiologists independently visually assessed PD from single left mediolateral oblique view images from the two vendors. Analysis of variance, intra-class correlation coefficients (ICC), scatter plots, and Bland-Altman plots were used to evaluate PD assessments between vendors. The mean radiologist PD for each image was used as a consensus PD measure. Overall agreement of the PD assessments was excellent between the two vendors with an ICC of 0.95 (95% confidence interval: 0.93 to 0.97). Bland-Altman plots demonstrated narrow upper and lower limits of agreement between the vendors with only a small bias (2.3 percentage points). The results of this study support the assumption that visual assessment of PD is consistent across mammography vendors despite vendor-specific appearances of "for presentation" images.

The goal of this study was to evaluate speech understanding outcomes in prelingually deaf children who use a cochlear implant device. Specifically, we discuss investigations on 2 prognostic variables--age at implantation and degree of residual hearing--and use a novel method: binary partitioning analysis. Our outcome measures are standard speech perception evaluations, including the Word Intelligibility by Picture Identification (WIPI) test, the Phonetically Balanced-Kindergarten (PBK) test, and the Glendonald Auditory Screening Procedure (GASP). Regarding age at implantation, we definitely showed that growth rates of speech understanding do relate to age at implantation, but not in a simple fashion. We used binary partitioning in an attempt to find the age at implantation that best separates the performances of children with younger versus older ages at implantation. We found that there is no one "critical age"; much appears to depend on the nature and difficulty (eg, whether open- or closed-set) of the test used. Regarding residual hearing, binary partitioning analysis was unable to show that the amount of residual hearing (as shown by preimplantation audiometric data) has any significant bearing on speech outcome measures in congenitally or prelingually deaf children.

Understanding the predictors of a quick death following diagnosis may improve timely access to palliative care. The objective of this study was to explore whether factors in the 24 months prior to a colorectal cancer (CRC) diagnosis predict a quick death post-diagnosis. Data were from a longitudinal study of all adult persons diagnosed with CRC in Nova Scotia, Canada, from 01Jan2001-31Dec2005. This study included all persons who died of any cause by 31Dec2010, except those who died within 30 days of CRC surgery (n = 1885 decedents). Classification and regression tree models were used to explore predictors of time from diagnosis to death for the following time intervals: 2, 4, 6, 8, 12, and 26 weeks from diagnosis to death. All models were performed with and without stage at diagnosis as a predictor variable. Clinico-demographic and health service utilization data in the 24 months pre-diagnosis were provided via linked administrative databases. The strongest, most consistent predicto...

Abdolell et al. (2001)1 implemented a binary partitioning algorithm for the case of continuous repeated measures outcomes, using Mahalanobis distance as a deviance measure to evaluate goodness-of-split. The algorithm was implemented only for a single split at the root node ...

Prior research has demonstrated that people who die shortly after receiving a cancer diagnosis are at increased risk for not being referred to palliative care. No previous studies have empirically derived the length of time between diagnosis and death associated with increased risk. The objective of this study was to identify the length of time between diagnosis and death associated with increased risk for non-enrolment in a palliative care programme. Binary recursive partitioning was employed to derive the cut-point for the number of days from colorectal cancer diagnosis to death predictive of a high risk for non-enrolment in a palliative care programme in two health districts in Nova Scotia, Canada. The study included all adults (≥20 years) who were diagnosed with colorectal cancer in the two districts between 1 January 2001 and 31 December 2005 and who died between 1 January 2001 and 31 December 2008 (n=894). Individuals who died within 18.5 days following diagnosis were at highest risk for non-enrolment in palliative care. Of the 60 adults who died in <18.5 days, 16.7% were enrolled in a palliative care programme; of the 835 adults who died ≥18.5 days after diagnosis, 65.9% were enrolled. This data-driven approach may be used to define the short diagnosis-to-death time frame at which individuals are at increased risk for non-enrolment in palliative care programmes. This approach allows researchers to further investigate and compare empirically-derived cut-points that identify those who die quickly and are at risk of not receiving palliative care.

To evaluate the impact of cone-beam computed tomography (CT) during sclerotherapy of low-flow vascular malformations. Eighty-seven cone-beam CT examinations were acquired during 81 sclerotherapy treatments of low-flow malformations in 48 patients: 81 were performed to evaluate sclerosing agent diffusion and six were performed to evaluate needle or catheter positioning before injection of therapeutic agent. Image quality was rated by two observers. Clinical impact of cone-beam CT in the assessment of therapeutic agent diffusion, needle or catheter positioning, subsequent treatment planning, and complication detection was evaluated. The κ-statistic was used to assess interobserver reliability and proportions, with associated 95% confidence intervals (CIs). All cone-beam CT images were successfully acquired. Image quality was rated as excellent or good for the majority of studies, with substantial interobserver reliability (κ = 0.648). Cone-beam CT studies improved assessment of therapeutic agent diffusion in 83% of cases (67 of 81; 95% CI, 75%-91%) for observer 1, who had access to ultrasound, fluoroscopic, and digital subtraction angiographic (DSA) imaging, and in 95% of cases (77 of 81; 95% CI, 90%-100%) for observer 2, who had access to only stored fluoroscopic spot radiographs and DSA images. Cone-beam CT impacted planning of the next treatment session in 49% of cases (40 of 81; 95% CI, 38%-60%). In 7% of cases (six of 81; 95% CI, 1%-13%), complications such as migration of therapeutic agent or compression of upper airways were detected that were not seen with other imaging. Cone-beam CT can be a useful adjunctive imaging tool, providing information to help decision-making during percutaneous sclerotherapy and ongoing management of low-flow vascular malformations.

We have made a retrospective analysis on 70 prelingually deaf children (78% congenital; age range 2-15 years) followed for up to 5 years post-implant during which both closed set speech perception tests (TAC, WIPI) and open set tasks (PBK, GASP) were administered. We used a binary partitioning algorithm to optimally divide our dataset on the basis of age at implantation This technique achieves an optimal split when the heterogeneity of the data is most reduced (maximal drop in deviance). For the closed set speech perception tests (TAC and WIPI) partitioning best divided-out data at age 4.4 years. For the open set tests optimal division was at a higher age of implantation (GASP word, 5.6 years; PBK word, 8.4 years). Using these partitioning values, we have found statistically significant differences between rate of improvement of scores in the younger implanted children compared with those implanted later.

This study aims to determine whether a binocular occlusion test for distance in intermittent exotropia could reliably show the most frequently deviating eye. With the reliable demonstration of such an event this test can be very helpful in selecting the appropriate eye for surgery when a unilateral lateral rectus recession is planned for the treatment of an intermittent exotropia. 24 patients (11 males, 13 females) aged 2.4 to 21 years (mean 5.7, SD ± 4) with intermittent exotropia were recruited. Subjects were excluded if they had had previous strabismus surgery, uncorrectable poor vision in one or both eyes, neurological disorders, other ocular motility disorders (e.g., Duane syndrome, Brown syndrome), or inability to cooperate during the test. The dominant eye was first determined using the sighting test and the near point of convergence test. The binocular cover test, while fixating a distant target, was then performed and ocular divergence was noted. The binocular occluder test consistently demonstrated the diverging eye to be the non-dominant eye. The bilateral occlusion test for distance in intermittent exotropia can reliably show the most frequently deviating eye. Clinicians may find this test helpful in selecting the eye for surgery in patients where unilateral lateral rectus recession for the treatment of an intermittent exotropia is being considered.

Scholarly research is a key component of Canadian urology residency. Through comparison of scholarly performance of urology residents before residency with that achieved during residency, we aimed to elicit predictive factors for completion of research activities. Electronic surveys were sent to 152 urology residents of 11 accredited Canadian programs. Survey questions pertained to post-graduate training year (PGY), formal education, scholarly activity completed before and after the start of residency, protected/dedicated research time, structured research curriculum and pursuit of fellowship training. Surveys were completed by 42 residents from 10 programs. Only 26% of residents had a structured research curriculum, 38% a dedicated research rotation and 43% protected research time. We found that 45% of residents published at least 1 manuscript so far during residency (mean 1.14 ± 0.32), and 43% submitted at least 1 manuscript (mean 0.86 ± 0.25). During residency, 62% of residents completed ≥1 formal research presentation (median number 1.5; range: 0-≥10). Only the level of PGY significantly affected the number of manuscripts published (p < 0.001) and number of formal research presentations (p < 0.001) completed during residency. In total, 86% of residents planning to pursue fellowship training had a mean number of publications and presentations during residency of 1.25 ± 0.37 and 2.25 ± 0.54, respectively. Level of PGY significantly affected quantitative scholarly activity, but the numbers and types of presentations performed prior to residency, completion of an honours or graduate degree and plans to pursue fellowship training did not.

We investigate a binary partitioning algorithm in the case of a continuous repeated measures outcome. The procedure is based on the use of the likelihood ratio statistic to evaluate the performance of individual splits. The procedure partitions a set of longitudinal data into two mutually exclusive groups based on an optimal split of a continuous prognostic variable. A permutation test is used to assess the level of significance associated with the optimal split, and a bootstrap confidence interval is obtained for the optimal split.