PRISM :: Browsing by Author "Sajobi, Tolulope"

Browsing by Author "Sajobi, Tolulope"

Now showing 1 - 20 of 24

Embargo
A Novel Semi-Automated Approach for Trial Identification and Evaluation of the Certainty of Evidence from Network Meta-Analyses
(2024-09-09) Kamso, Mohammed Mujaab; Hazlewood, Glen; Deardon, Rob; Sajobi, Tolulope; Tomlinson, George
This thesis introduces an innovative approach for the rapid identification of randomized controlled trials (RCTs) and evaluation of the certainty of evidence within the context of a living systematic review and network meta-analysis. The first paper (Chapter 4) describes a living systematic review methodology that incorporates crowd-sourcing, machine learning and a web-based tool to streamline the identification and classification of RCTs, introducing a novel "studification" process to enhance review maintenance. The second paper (Chapter 5) presents a semi-automated method for evaluating the certainty of evidence derived from direct estimates within a Bayesian network meta-analysis framework, adhering to GRADE guidance. The study also addresses the assessment of indirectness at the study-specific level using online tools. The final paper (Chapter 6) extends this methodology to assess the certainty of evidence for indirect and mixed evidence separately. This is achieved through a semi-automated process that utilizes the concept of the contribution matrix to identify the first-order loop, highlighting the primary contributors to indirect estimate. Additionally, in accordance with GRADE recommendations, an automated approach for evaluating imprecision is developed. Overall, this thesis may enhance the efficiency of maintaining a living systematic review, offering a novel approach to semi-automate the evaluation of evidence certainty from Bayesian network meta-analysis models while adhering to GRADE guidelines. Applied to the context of early rheumatoid arthritis, the findings potentially have positive policy implications such as how fast reviews can be done for clinical practice guideline development.
Open Access
Administrative health data in Canada: lessons from history
(BioMed Central, 2015-08-19) Lucyk, Kelsey; Lu, Mingshan; Sajobi, Tolulope; Quan, Hude
BACKGROUND: Health decision-making requires evidence from high-quality data. As one example, the Discharge Abstract Database (DAD) compiles data from the majority of Canadian hospitals to form one of the most comprehensive and highly regarded administrative databases available for health research, internationally. However, despite the success of this and other administrative health data resources, little is known about their history or the factors that have led to their success. The purpose of this paper is to provide an historical overview of administrative data for health research in Canada to contribute to the institutional memory of this field. METHODS: We conducted a qualitative content analysis of approximately 20 key sources to construct an historical narrative of administrative health data in Canada. Specifically, we searched for content related to key events, individuals, challenges, and successes in this field over time. RESULTS AND DISCUSSION: In Canada, administrative data for health research has developed in tangent with provincial research centres. Interestingly, the lessons learned from this history align with the original recommendations of the 1964 Royal Commission on Health Services: (1) standardization, and (2) centralization of data resources, that is (3) facilitated through governmental financial support. CONCLUSIONS: The overview history provided here illustrates the need for longstanding partnerships between government and academia, for classification and standardization are time-consuming and ever-evolving processes. This paper will be of interest to those who work with administrative health data, and also for countries that are looking to build or improve upon their use of administrative health data for decision-making.
Open Access
Association between glycemic load and cognitive function in community-dwelling older adults: results from the Brain in Motion study
(2017) Garber, Anna; Poulin, Marc; Friedenreich, Christine; Csizmadi, Ilona; Longman, Richard S.; Sajobi, Tolulope; Shearer, Jane
Background: Impaired glucose tolerance is a risk factor for non-age-related cognitive decline and is also associated with measures of physical activity (PA) and cardiorespiratory fitness (CRF). A low glycemic load (GL) diet can aid in the management of blood glucose levels, but little is known about its effect on cognition with poor glucoregulation. Objective: The aim of this thesis was to assess the relation between GL and cognitive function by glucoregulation, and possible mediatory effects by CRF and PA, in older adults. Design: A cross-sectional analysis of 194 cognitively healthy adults aged ≥55 years (mean=65.7, SD=6.1) was conducted. GL was assessed using a quantitative food frequency questionnaire, and glucoregulation was characterized on the HOMA-IR index. Subjects also completed a cognitive assessment, CRF testing, a validated self-reported PA questionnaire, and a blood draw. Multiple linear regression models adjusted for significant covariates were used to evaluate the relation between GL and cognition, and mediation analysis was used to assess potential mediatory effects by CRF and PA. Results: GL was inversely associated with global cognition (β=-0.014; 95% CI -0.024, -0.0036) and figural memory (β =-0.035; 95% CI -0.052, -0.018) in subjects with poor glucoregulation. Neither CRF nor PA mediated these relations. In subjects with good glucoregulation, no association was found between GL and cognitive function (p>0.05). Conclusions: A low GL diet is associated with better cognitive function in older adults with poor glucoregulation. This study provides supportive evidence for the role of GL in maintaining better cognitive function during the aging process.
Open Access
Bayesian Variable Selection Model with Semicontinuous Response
(2022-01-14) Babatunde, Samuel; Chekouo, Thierry; Sajobi, Tolulope; Zhang, Qingrun; Deardon, Robert; Bezdek, Karoly
We propose a novel Bayesian variable selection approach that identifies a set of features associated with a semicontinuous response. We used a two-part model where one of the models is a logit model that estimates the probability of zero responses while the other model is a log-normal model that estimates responses greater than zero (positive values). Stochastic Search Variable Selection (SSVS) procedure is used to randomly sample the indicator variables for variable selection which in turn searches the space of feature subsets and identifies the most promising features in the model. For the logistic model, a data augmentation approach is used to sample from the posterior density. We impose a spike-and-slab prior for the regression effects where the unselected covariates take on a prior mass at zero while the selected covariates follow a normal distribution (including the intercept and clinical covariates). Since the joint posterior density had no closed form, we employed the techniques of the Markov Chain Monte Carlo (MCMC) to sample from the posterior distribution. Simulation studies are used to assess the performance of the proposed method. We computed the average area under the receiver operating characteristic curve (AUC) to assess variable selection and compared it with competing methods. We also assessed the convergence diagnosis of our MCMC algorithm by computing the potential scale reduction factor and correlations between the marginal posterior probabilities. We finally apply our method to the coronary artery disease (CAD) data where the aim is to select important genes associated with the CAD index. This data consists of clinical covariates and gene expressions.
Open Access
Binary and Ordinal Outcomes: Considerations for the Generalized Linear Model with the Log Link and with the Identity Link
(2017) Singh, Gurbakhshash; Fick, Gordon; Kopciuk, Karen; Sajobi, Tolulope; Lu, Xuewen; Horrocks, Julie
There are gaps in the current literature on Generalized Linear Models (GLM) for binary outcomes with the log link. This dissertation explores a number of these gaps and presents specific results: (1) Uniqueness considerations for the Maximum Likelihood Estimate (MLE) are established from the conditions for the strict concavity of the log-likelihood. The full column rank of certain subsets of the covariate matrix is shown to be a condition for the strict concavity of the loglikelihood. (2) Conditions are established for the finiteness of components of the MLE. A method is proposed to address the possibility of non-finite components for the MLE, and it is based on determining directions of recession of the log-likelihood. In addition, it is established when the MLE will be in the interior of the parameter space and when the MLE will possibly be on a boundary of the parameter space. (3) Examples are presented of closed form expressions for the MLE. For a number of models with indicator variables and measured variables, closed form expressions for the MLE are presented. (4) There are considerations for the construction of confidence intervals when the MLE is close to a boundary of the parameter space. A new metric, called the “fraction within the parameter space”, is introduced for assessing intervals for MLEs close to a boundary. A simulation study is provided that offers support for Bootstrap Percentile Intervals having larger fractions when compared to Relative Likelihood Intervals and Normal Confidence Intervals. This dissertation continues by developing a proportional probability model using the log link for ordinal outcomes. For this model, similar results are presented for topics (1) and (3) above. In addition, there is the introduction of a score test to assess proportionality. The dissertation concludes with a discussion of future work. In particular, this discussion includes some preliminary work with the identity link GLM for binary and ordinal outcomes. Throughout this dissertation, there are many practical considerations and illustrations presented. The use of the log link and the identity link for binary and ordinal outcomes should now become a viable modeling option for researchers.
Open Access
Cardiac Rehabilitation and Secondary Prevention in Patients with Coronary Artery Disease and Atrial Fibrillation
(2021-08-26) Liu, Hongwei; Wilton, Stephen; Tian, Ye; James, Matthew; Sajobi, Tolulope; Arena, Ross; Shaheen, Abdel-Aziz; Oh, Paul
Background: Referral to and participation in cardiac rehabilitation (CR) in Canada and elsewhere remains suboptimal. The evidence for the benefits of CR in reducing incidence of atrial fibrillation (AF) in patients with coronary artery disease (CAD) is modest. Furthermore, whether multifactorial risk factor intervention is effective in improving prognosis in patients with AF remains unclear. Methods: We studied these questions by conducting two projects. Project 1 is a systematic review, which evaluated evidence on the effects of multifactorial risk factor intervention in patients with AF. Project 2 is a retrospective cohort study, which evaluated the relationships of CR completion status and cardiorespiratory fitness (CRF) across a CR program with the risk of incident AF. These analyses in Project 2 were performed by linking databases from an Alberta provincial cardiac catheterization registry, a city-wide CR program in Calgary, and Alberta provincial health administrative datasets. Results: In Project 1, the systematic review suggested that multifactorial risk factor intervention was positively associated with improved AF-related symptoms and health-related quality of life. In Project 2, we first used electrocardiography data to improve the diagnostic yield of administrative data-based AF identification algorithms. We further demonstrated that CR program completion was not associated with lower risk of incident AF after adjusting for baseline characteristics. However, both baseline CRF, 12-week CRF, and CRF changes following CR completion had inverse dose-dependent relationships with the risk of incident AF. Furthermore, we developed a risk prediction model for incident AF in patients completing a CR program, which showed good discrimination and was well calibrated in predicting the risk of AF at 5-years follow-up. Conclusions: These findings have enhanced the importance of multifactorial risk factor intervention in managing patients with AF, and added to the current state of knowledge of CR in improving the prognosis of patients with CAD, thereby providing further support for the promotion of CR. Furthermore, the risk prediction model will help prioritize resources for patients who are at high risk of developing AF and can benefit the most from screening for AF and participating in personalized CR services.
Open Access
Correlates of low resilience and physical and mental well-being among black youths in Canada
(2024-08-31) Oluwasina, Folajinmi; Henderson, Jo; McKenzie, Kwame; Mullings, Delores V.; Renzaho, Andre M.; Sajobi, Tolulope; Rousseau, Cecile; Senthilselvan, Ambikaipakan; Hamilton, Hayley; Salami, Bukola
Abstract Background Resilience has gained considerable attention in the mental health field as a protective factor that enables individuals to overcome mental health issues and achieve positive outcomes. A better understanding of resilience among Black youth is important for supporting the strengths and capacities within this population. This study seeks to investigate the correlates of resilience among Black youths in Canada. Methods The survey was conducted online through REDCap between November 2022 and March 2023. The Brief Resilience Scale (BRS) was utilized to measure the capacity of participants to recover from or bounce back from stress. The BRS comprises six five-point Likert scale items. Data were analyzed employing a bivariate analysis followed by a multivariable binary logistic regression. Results A total of 933 Black youths participated in the study across all Canadian provinces, of which 51.8% (483) identified as female and 46.7% (436) as male. Most respondents 51.3% (479) were between the ages of 16 and 20 years, with 28% (261) between the ages of 21 and 25 years, and 20.2% (188) between the ages of 26 and 30 years. In terms of employment, 62.0% (578) were working part-time, 23.7%, (220) were unemployed, and 9.8% (91) were working full-time. Over a third of participants (39.3%, 331) rated their mental health over the last month as good, with 34% (317) giving a rating of poor and 20.9% (195) giving a rating of fair. Black youths who were working part-time had four times greater odds of expressing low resilience (OR: 4.02; 95% CI: 1.82–11.29) than those who were not working. Black youth who ranked their mental health as poor were about nine times (OR: 8.65; 95% CI: 1.826–21.978) more likely to express low resilience. Conclusion In this study, the Black youth participants reported relatively low resilience scores. Employment, physical health, and mental health status were factors that contributed to low resilience. Further studies are needed to examine the causal link between resilience and its dynamic effect on health outcomes among Black youth. More interventions are needed to make mental health services accessible to Black youth in a more culturally sensitive way with cross-culturally trained professionals.
Open Access
Group Selection in Semiparametric and Nonparametric Accelerated Failure Time Models
(2017) Huang, Longlong; Lu, Xuewen; Kopciuk, Karen; Deardon, Rob; Sajobi, Tolulope; Yan, Ying; Hu, Joan
In survival analysis, a number of regression models can be used to estimate the effects of covariates on the censored survival outcome. When covariates can be naturally grouped, group selection is important in these models. Motivated by the group bridge approach for variable selection in a multiple linear regression model, we consider group selection in a semiparametric accelerated failure time (AFT) model using Stute's weighted least squares and a group bridge penalty. This method is able to simultaneously carry out feature selection at both the group and within-group individual variable levels and enjoys the powerful oracle group selection property. Although the group bridge penalized approach can effectively remove unimportant groups, it cannot effectively remove unimportant variables within the important groups. To overcome this limitation, the adaptive group bridge method is proposed. We show that the adaptive group bridge method obtains the oracle property. Simulation studies indicate that the group bridge and adaptive group bridge approaches for the AFT model can correctly identify important groups and variables even with high censoring rates. A real data analysis is provided to illustrate the application of the proposed methods. We further study a nonparametric accelerated failure time additive regression (NP-AFT-AR) model whose covariates have nonparametric effects on the survival time. The proposed model is more flexible than the linear model and can be fitted to high-dimensional censored data when some components are unknown non-linear functions. B-splines are used to approximate the nonparametric components. A group bridge penalized variable selection approach based on the inverse probability-of-censoring weighted least squares is developed to select nonparametric components. The proposed approach is able to distinguish the nonzero components from the zero components and estimate the nonzero components simultaneously. Computational algorithms and theoretical properties of the proposed method are established. Simulation studies indicate that the proposed method has satisfactory performance even with relatively high censoring rates. Two real data analyses are used to illustrate the application of the proposed method to survival data analysis.
Open Access
Immigration and Depression in Canada: Is there really a Healthy Immigrant Effect? What is the Pattern of Depression by Time since Immigration?
(2017) Diaz, Ruth; Patten, Scott; Bulloch, Andrew; Thomas, Bejoy; Sajobi, Tolulope
Objective: This study aimed to contribute to the understanding on inequalities in mental health in Canada by exploring whether or not immigrants have lower prevalence of past-year major depressive episode (MDE) than non-immigrants (i.e., the healthy immigrant effect HIE). Methods: Data were from ten cross-sectional Canadian population health surveys. Survey-specific log odd ratios were calculated, and then pooled using random effects meta-analytic techniques. Results: Evidence of the HIE on MDE was found; however, the HIE disappears with age. The pattern of the HIE by age was observed overall, and when the analysis was conducted by sex, country of birth, and time since immigration. Elder immigrants seem to be at similar or higher risk of MDE than elder Canadian-born. Conclusion: More research is needed to replicate this findings, and to understand why elder immigrants may be at higher risk of MDE than elder non-immigrants.
Open Access
Impact of a farmers’ market healthy food subsidy on the diet quality of adults with low incomes in British Columbia, Canada: a pragmatic randomized controlled trial
(Elsevier, 2023-02-01) Aktary, Michelle L.; Dunn, Sharlette; Sajobi, Tolulope; O'Hara, Heather; Leblanc, Peter; McCormack, Gavin R.; Caron-Roy, Stephanie; Ball, Kylie; Lee, Yun Yun; Nejatinamini, Sara; Reimer, Raylene A.; Pan, Bo; Minaker, Leia M.; Raine, Kim D.; Godley, Jenny; Downs, Shauna; Nykiforuk, Candace I.J.; Olstad, Dana Lee
Adults with low incomes have lower diet quality than their higher income counterparts. In Canada, the British Columbia Farmers’ Market Nutrition Coupon Program (FMNCP) provides coupons to low-income households to purchase healthy foods in farmers’ markets.
Open Access
Is the Prevalence of Major Depression Increasing in Canadian Adolescents? Assessing Trends from 2000 to 2014.
(2016) Wiens, Kathryn; Patten, Scott; Duffy, Anne; Pringsheim, Tamara; Sajobi, Tolulope
Objective: The aim of this thesis was to determine whether there is evidence of an epidemic of major depression in Canadian adolescents. Methods: Prevalence estimates for major depressive episodes (MDE) were derived from a series of Canadian Community Health Surveys. Meta regression and graphical analyses were used to evaluate trends over time. Results: The findings do not support an increase in MDE prevalence in Canadian adolescents from 2000 to 2014 (=0.0006; p=0.532). Age and sex groups did not modify the observance of trends. A post hoc analysis observed mood disorder diagnosis to increase from 2003 to 2014 (=0.0012; p=0.024). Conclusion: MDE prevalence in adolescents has remained relatively stable over the past 15 years. These results suggest mood disorder diagnosis is increasing, which may contribute to the popular belief of an epidemic in adolescents. Policy makers may need to incorporate increasing need of services into future planning.
Open Access
Machine learning using multimodal clinical, electroencephalographic, and magnetic resonance imaging data can predict incident depression in adults with epilepsy: A pilot study
(Wiley, 2023-07-08) Delgado-García, Guillermo; Engbers, Jordan D. T.; Wiebe, Samuel; Mouches, Pauline; Amador, Kimberly; Forkert, Nils D.; White, James; Sajobi, Tolulope; Klein, Karl Martin; Josephson, Colin B.; Calgary Comprehensive Epilepsy Program Collaborators
Objective: To develop a multi-modal machine learning (ML) approach for predicting incident depression in adults with epilepsy. Methods: We randomly selected 200 patients from the Calgary Comprehensive Epilepsy Program registry and linked their registry-based clinical data to their first-available clinical electroencephalogram (EEG) and magnetic resonance imaging (MRI) study. We excluded patients with a clinical or Neurological Disorders Depression Inventory for Epilepsy (NDDI-E)-based diagnosis of major depression at baseline. The NDDI-E was used to detect incident depression over a median of 2.4 years follow-up (IQR 1.5-3.3 years). A ReliefF algorithm was applied to clinical as well as quantitative EEG and MRI parameters for feature selection. Six ML algorithms were trained and tested using stratified three-fold cross-validation. Multiple metrics were used to assess model performances. Results: Of 200 patients, 150 had EEG and MRI data of sufficient quality for ML, of which 59 were excluded due to prevalent depression. Therefore, 91 patients (41 women) were included with a median age of 29 (IQR 22-44) years. A total of 42 features were selected by ReliefF, none of which were quantitative MRI or EEG variables. All models had a sensitivity >80% and 5 of 6 had an F1 score ≥0.72. Multilayer perceptron model had the highest F1 score (median 0.74; interquartile range [IQR] 0.71-0.78) and sensitivity (84.3%). Median area under the receiver operating characteristic curve and normalized Matthews correlation coefficient were 0.70 (IQR 0.64-0.78) and 0.57 (IQR 0.50-0.65), respectively. Significance: Multimodal machine learning using baseline features can predict incident depression in this population. Our pilot models demonstrated high accuracy for depression prediction. However, overall performance and calibration can be improved. This model has promise for identifying those at risk for incident depression during follow-up, though efforts to refine it in larger populations along with external validation are required.
Open Access
Patient preferences and individualized risk prediction for management of acute coronary syndrome in chronic kidney disease
(2021-09-22) Wilson, Todd Allen; James, Matthew; Sajobi, Tolulope; Wilton, Stephen; Hazlewood, Glen
Chronic kidney disease (CKD) affects over 10% of adult Canadians and is associated with high risks of morbidity and mortality following non-ST-elevation acute coronary syndrome (ACS). ACS is managed invasively or conservatively with medication, however, people with CKD are 20-50% less likely to receive invasive management than patients without CKD. There are important knowledge gaps to support shared decision-making for ACS in this patient population related to risk prediction of clinical outcomes and patient preferences for treatment options. This thesis’ objectives are to: 1) understand patient preferences towards invasive versus conservative management of ACS, 2) develop tools for predicting long-term adverse outcomes following ACS, and 3) estimate net effects of invasive management based on trade-offs in reducing risks of mortality or readmission for myocardial infarction (MI) versus increasing risk of end-stage kidney disease (ESKD). To achieve these objectives, we conducted a discrete choice experiment (DCE) to quantify patient preferences toward treatment decisions for ACS and developed risk prediction models for mortality, readmission for MI, the composite of mortality and readmission for MI, and ESKD. Further, we synthesized patient preference and absolute risk estimates, while incorporating treatment effects from randomized control trials, to conduct a net effect analysis. The DCE found most patients preferred treatment options that lowered their risk of mortality; however, a subgroup of patients was identified with strong preferences toward conservative treatment for ACS. Risk prediction model performance varied. Model calibration was very good; however, discrimination ranged from excellent for predicting ESKD to poor for predicting readmission for MI. The net effect analysis estimated 87% of all patients with CKD were expected to experience net benefit from invasive management with reductions in risks of mortality or readmission for MI outweighing increases in risk of progression to ESKD. This work has addressed knowledge gaps for understanding preferences of patients with CKD toward key attributes of treatment options for ACS and for providing individualized estimates of long-term outcomes according to treatment strategy for ACS for patients with CKD. This work provides knowledge to individualize benefit-harm information and support shared decision-making approaches for ACS treatment for patients with CKD.
Open Access
Predicting poor postoperative pain control after elective spine surgery
(2019-06-26) Yang, Min-Han Michael; Riva-Cambrin, Jay; Casha, Steven; Sajobi, Tolulope; Jetté, Nathalie
Background: Inadequate postoperative pain control after spine surgery is common and can lead to patient dissatisfaction and poor outcomes. Predictors for poorly controlled pain after spine surgery are unknown and preoperative prognostic tools are not available to aid in the identification of high-risk patients to help facilitate the development of personalized treatments. In this thesis, we performed (1) a systematic review on the predictors associated with poor pain control in surgical patients; (2) performed a retrospective cohort study evaluating predictors of poor postoperative pain control following spine surgery; and (3) developed and validated a clinical prediction score to identify patients at high-risk for developing poor pain control. Methods: (1) A random-effects model was used to meta-analyze the predictors for poor pain control after surgery in the systematic review. (2) Adults from the Canadian Spine Outcomes and Research Network registry who underwent elective cervical or thoracolumbar surgery were included. Preoperative predictors for poor pain control (mean numeric rating scale for pain>4 at rest during the first 24 hours after surgery) were identified using a multivariable logistic regression model. (3) The prediction score was developed and internally validated using a 70:30 split-sample method. Results: (1) Thirty-three studies representing 53,362 patients were included in the systematic review. Nine significant predictors for poor postoperative pain control were identified across surgical disciplines. (2) The retrospective cohort study included 1,300 patients, of which 56.7% had poor pain control after surgery. The multivariable model identified that younger age, female sex, preoperative daily opioid use, higher preoperative neck/back pain, higher depression scores on patient health questionnaire-9, ≥3 motion segment surgery, and fusion surgery were associated with poor pain control. (3) Patients identified as low-, high-, and extreme-risk by the score had 32.0%, 63.0%, and 85.0% probability of developing poor pain control, respectively. Conclusion: Seven significant predictors for poorly controlled pain after spine surgery were identified and incorporated into a prediction score. The score can discriminate patients at higher risk for, and accurately predict the probability of, developing poor pain control after surgery.
Open Access
Predicting the Risk of Intracerebral Haemorrhage in Patients with Acute Ischemic Stroke Receiving IV-alteplase with or Without Endovascular Thrombectomy
(2016) Batchelor, Connor; Menon, Bijoy K.; Demchuk, Andrew M.; Goyal, Mayank; Lee, Ting-Yim; Sajobi, Tolulope
Intracerebral Haemorrhage (ICH) is a devastating complication of acute ischemic stroke (AIS) treatment with no known effective management protocols. The need to identify patients at risk of developing this condition is becoming increasingly recognized among the stroke community. Computed tomography perfusion (CTP) is a powerful diagnostic imaging tool that measures blood flow in the brain. This tool can also be used to provide information regarding the integrity of the blood-brain barrier (BBB). Severe brain ischemia and consequent disruption of the BBB are probable mechanisms for why ICH occurs after AIS treatment. The goal of my research is to investigate the potential role of CTP primarily and other imaging and clinical parameters in predicting ICH secondary to AIS treatment in patients.
Open Access
Predicting the Side Effects of Antiseizure Medications Using Machine Learning Models
(2024-01-02) Lin, Chantelle Qing Yang; Josephson, Colin Bruce; Sajobi, Tolulope; Klein, Karl Martin; Forkert, Nils Daniel
With over 20 anti-seizure medications (ASMs), identifying the ideal drug is often imprecise and time-consuming. Developing predictive models to expedite optimal drug selection is challenging due to the minimal differences in efficacy among adult patients with epilepsy. However, side-effects vary considerably between medications, and are one of the main reasons for discontinuation of ASM treatment. The aim was to (1) assess the prognostic utility of high- dimensional data such as genetic features with clinical features to predict ASM discontinuation, and (2) determine the optimal regression/machine learning model for predicting ASM discontinuation. This retrospective cohort study included 4,853 exposures to any ASM, and 624 patients exposed to valproic acid (VPA) from the RAISE-GENIC study during the years 2006-2020. The predicted outcome was defined as ASM discontinuation due to any side-effect reported by the patient. Clinical features included age of onset, patient age, sex, comorbidities, seizure type, EEG variables, and imaging variables. Network analysis of mRNA expression data from VPA-exposed neurons derived from control induced pluripotent stem cells (iPSCs) was leveraged to extract exome sequencing and genome-wide single nucleotide polymorphism data. Features were selected for model inclusion based on relevance as determined by the ReliefF algorithm. Penalized logistic regression, support vector machine, random forest, and k-nearest neighbor models were trained on the normalized bootstrapped dataset and model quality was assessed using stratified 10-fold cross validation. Models with only clinical and combined clinical and genetic features were compared by quantitative as well as visual discrimination and calibration metrics. The results showed that the best performing model was the penalized logistic regression using the VPA dataset with genetic and clinical features. The accuracy was 0.75 [95% confidence interval 0.74-0.76], area under the receiver operating characteristic curve was 0.66 [0.66-0.67], Brier score was 0.20 [0.19-0.21], sensitivity was 0.42 [0.41-0.42], and specificity 0.82 [0.82-0.83]. Machine learning using clinical and genetic features can moderately predict treatment-ending side- effects to VPA with moderate performance, discrimination, and calibration. If these results can be validated and improved upon, decision tools can be incorporated into clinical routines, simplifying drug prescriptions, saving time, and improving patient quality of life.
Open Access
Psychometric evaluation of a Canadian version of the Seattle Angina Questionnaire (SAQ-CAN)
(2020-12-01) Lawal, Oluwaseyi A; Awosoga, Oluwagbohunmi; Santana, Maria J; James, Matthew T; Southern, Danielle A; Wilton, Stephen B; Graham, Michelle M; Knudtson, Merrill; Lu, Mingshan; Quan, Hude; Ghali, William A; Norris, Colleen M; Sajobi, Tolulope
Abstract Background The Seattle Angina Questionnaire (SAQ) is a widely-used patient-reported outcomes measure in patients with heart disease. This study assesses the validity and reliability of the SAQ in a Canadian cohort of individuals with stable angina. Methods and results Data are from the Alberta Provincial Project for Outcome Assessment in Coronary Heart Disease (APPROACH) registry, a population-based registry of patients who received cardiac catheterization in Alberta, Canada. The cohort consists of 4052 patients undergoing cardiac catheterization for stable angina and completed the SAQ within 2 weeks. Exploratory factor analysis and confirmatory factor analysis (CFA) were used to assess the factorial structure of the SAQ. Internal and test–retest reliabilities of a new measure (i.e., SAQ-CAN) was measured using Cronbach α and intraclass correlation coefficient, respectively. CFA model fit was assessed using the root mean square error of approximation (RMSEA) and comparative fit index (CFI). Construct validity of the SAQ-CAN was assessed in relation to Hospital Anxiety and Depression Scales (HADS), Euro Quality of life 5 dimension (EQ5D), and original SAQ. Of the 4052 patients included in this analysis, 3281 (80.97%) were younger than 75 years old, while 3239 (79.94%) were male. Both exploratory and confirmatory factor analyses revealed a four-factorial structure consisting of 16 items that provided a better fit to the data (RMSEA = 0.049 [90% CI = (0.047, 0.052)]; CFI = 0.975). The 16-item SAQ demonstrated good to excellent internal reliability (Cronbach’s α range from 0.77 to 0.90), moderate to strong correlation with the Original SAQ and EQ5D but negligible correlations with HADS. Conclusion The SAQ-CAN has acceptable psychometric properties that are comparable to the original SAQ. We recommend its use for assessing coronary health outcomes in Canadian patients with Coronary Artery Disease.
Open Access
Quality improvement interventions to prevent unplanned extubations in pediatric critical care: a systematic review
(2022-12-02) Wollny, Krista; Cui, Sara; McNeil, Deborah; Benzies, Karen; Parsons, Simon J.; Sajobi, Tolulope; Metcalfe, Amy
Abstract Background An unplanned extubation is the uncontrolled and accidental removal of a breathing tube and is an important quality indicator in pediatric critical care. The objective of this review was to comprehensively synthesize literature published on quality improvement (QI) practices implemented to reduce the rate of unplanned extubations in critically ill children. Methods We included original, primary research on quality improvement interventions to reduce the rate of unplanned extubations in pediatric critical care. A search was conducted in MEDLINE (Ovid), Embase, and CINAHL from inception through April 29, 2021. Two reviewers independently screened citations in duplicate using pre-determined eligibility criteria. Data from included studies were abstracted using a tool created by the authors, and QI interventions were categorized using the Behavior Change Wheel. Vote counting based on the direct of effect was used to describe the effectiveness of quality improvement interventions. Study quality was assessed using the Quality Improvement Minimum Quality Criteria Set (QI-MQCS). Results were presented as descriptive statistics and narrative syntheses. Results Thirteen studies were included in the final review. Eleven described primary QI projects; two were sustainability studies that followed up on previously described QI interventions. Under half of the included studies were rated as high-quality. The median number of QI interventions described by each study was 5 [IQR 4–5], with a focus on guidelines, environmental restructuring, education, training, and communication. Ten studies reported decreased unplanned extubation rates after the QI intervention; of these, seven had statistically significant reductions. Both sustainability studies observed increased rates that were not statistically significant. Conclusions This review provides a comprehensive synthesis of QI interventions to reduce unplanned extubation. With only half the studies achieving a high-quality rating, there is room for improvement when conducting and reporting research in this area. Findings from this review can be used to support clinical recommendations to prevent unplanned extubations, and support patient safety in pediatric critical care. Systematic review registration This review was registered on PROSPERO (CRD42021252233) prior to data extraction.
Open Access
Resilience in Women After Intimate Partner Violence
(2017) Fenton, Carol; Thurston, Wilfreda Enid; Sajobi, Tolulope; Radtke, Hazel Lorraine
Objective To describe the distribution of well-being in survivors of Intimate Partner Violence (IPV), and examine the relationships between the Resilience Portfolio Model and well-being. Method This study was a cross-sectional survey of 665 women IPV survivors from three Canadian provinces. Well-being was defined as a multi-domain score. Partial proportional ordinal regression was used to evaluate relationships between Resilience Portfolio Model elements and well-being. Results The median score for well-being was 3 out of 6 and the mode was 2. The factors that were associated with well-being included: income, education, focusing on the future, and receiving emotional support. Thinking that one’s partner would stop being violent if they stopped using alcohol or drugs was statistically significant for lower levels of well-being. Conclusions This study reveals that there is variable well-being in IPV survivors, and provides evidence for specific traits or behaviours that may support well-being in this population.
Embargo
Some Contributions to Understanding the Heterogeneity of Treatment Effects in Stroke Trials
(2024-06-20) Ademola, Ayoola; Sajobi, Tolulope; Hildebrand, Kevin A.; Hill, Michael D.; Thabane, Lehana
Background: Stroke is a neurological disease that is the third leading cause of death and the tenth-largest known cause of disability-adjusted life years in Canada. Fortunately, clinical trial evidence has identified a few treatments that improve patients’ outcomes, resulting in faster reperfusion, better functional outcomes, lower mortality rates, and improved quality of life. Despite the overall positive benefits of these interventions, there remain differences in the impact of the treatment at the individual level, with some patients experiencing positive benefits and others showing neutral or adverse effects of interventions. Such heterogeneity of treatment effects (HTE) could be attributed to differences in patients’ socio-demographic or clinical characteristics, study designs, inclusion/exclusion criteria, and geographic or regional healthcare systems. Conventional statistical approaches for accounting for within-study and between-study HTE have primarily relied on within-trial subgroup analysis and meta-analysis. However, these approaches are limited because they are based on restrictive distributional assumptions, which may only be tenable in some clinical trials. Methods: This dissertation investigates relevant methodologies for characterizing and accommodating treatment effects within- and between-study heterogeneity in stroke trials. The specific objectives of this dissertation are to: 1) assess the credibility of subgroup analyses reported in published stroke trials; 2) investigate the comparative performance of methods for subgroup identification in clinical trials with binary endpoints when there is no a priori knowledge of patients’ characteristics associated with HTE, and 3) examine the performance of random-effects models when synthesizing evidence from trials with different study design characteristics. This study uses a combination of knowledge synthesis methodology and computer simulations to address these objectives. For objective 1, we conducted a systematic review to examine the credibility of reported subgroups in stroke trials. We used the Instrument for Assessing the Credibility of Effect Modification Analyses (ICEMAN) checklist to evaluate the quality of the subgroup analyses conducted for each study. For Objectives 2 and 3, computer simulations were used to examine the comparative performance of subgroup identification methods for identifying relevant variables/biomarkers associated with HTE in clinical trials of binary endpoints and meta-analytic methods for synthesizing treatment effects obtained from explanatory and pragmatic trials, respectively. Results: The systematic review of reporting quality of subgroup analyses in stroke trials revealed that the credibility of reported subgroup analyses is poor, with most studies not providing a priori rationale for the type and number of subgroup analyses conducted. Among all the subgroup identification methods investigated, the model-based recursive partitioning (MOB) method had the best control of Type I and higher statistical power to detect HTE. The random-effects model based on t-distribution (robustRE) and the mixture random-effects model (mixRE) are more appropriate for meta-analysis data with substantial HTE. However, the conventional random-effects model (RE model) remains reliable for estimating pooled treatment effects in data with moderate HTE. Conclusion: Understanding and capturing treatment effect heterogeneity is critical for generating evidence about treatment effectiveness in clinical trials. More statistical methods that account for heterogeneity in the study population and design characteristics are recommended to analyze and synthesize evidence from clinical trials.