Menu
In medical research, to find out whether a treatment works for a disease typically depends on comparing the results of two groups of people: those who get the treatment, versus those who do not, ideally in a clinical trial.
To avoid bias in results, researchers who design clinical trials make sure that the people in both groups are very similar (e.g., same age, seriousness of the disease, equal length of time with the disease, so on). Unfortunately, this type of research design often does not include patients who are the sickest, of older age, or are from different ethnic groups, and thus it is impossible to know whether the drug will actually work on these types of patients.
Pragmatic trials are a new kind of trial design, which aims to include these more vulnerable groups of patients. However, because these patients are less similar, it is difficult to analyze the data.
Our study focused on cases of “incomplete treatment adherence,” “partial adherence,” and “non-adherence” within a sample. For example, within a study, often some patients are not able to continue with the treatment, need to take less of the drug, or have to drop out of the study.
The current ways to analyze the data often ignore most of these details, and therefore the results are not very useful to a patient or a doctor in making treatment decisions. Sophisticated statistical methods are currently being developed, but often these methods are not well understood or accessible to the analysts.
So, we studied emerging methods of accounting for this variety within the data.
Watch a brief overview by the team summarizing the proposal for this project.
Runtime: 12:07
We evaluated different statistical methods to account for incomplete treatment adherence, and contrasted the performances of these methods to some of the commonly used methods, under different realistic clinical settings where patients were supposed to follow a sustained treatment strategy. We paid particular attention to the challenging setting for data wherein patients’ lab tests are done infrequently, evaluating various missing data analysis techniques to address such challenges.
There is some analytical guidance on estimating treatment effects when some patients are fully adherent, and some patients are not adherent at all (i.e., two categories of adherence). However, most patients are partially adherent in the real world—they start to take the treatment and then decide to discontinue it for various reasons.
Our research has extended the existing analytic approach to accommodate for this (i.e., considering a third category of adherence).
For dealing with medication non-adherence, a few methods are proposed in the economic literature (popularly known as “instrumental variable analysis”). However, it is currently unknown how good these economic methods are compared to statistical methods if we apply them to the same context, such as pragmatic trials.
In our project, we explored the characteristics of both these methods and determined how practical these methods are in various clinical scenarios.
1. Hossain MB, Mosquera L, Karim ME. Performance of statistical methods to address treatment non-adherence in pragmatic clinical trials with point treatment settings: a simulation study. University of Toronto Journal of Public Health. 2021; 2(2). doi: https://doi.org/10.33137/utjph.v2i2.36762 (Objective 3)
2. Mosquera, L., & Karim, M. E. (2021, February). Evaluating Adjusted Per-Protocol Effect Estimators in Pragmatic Trials to Address Treatment Non-Adherence. In International Journal Of Clinical Pharmacy (Vol. 43, No. 1, pp. 298-298). Netherlands: Springer. https://doi.org/10.1007/s11096-020-01213-y (Objective 1)
3. Hossain MB, Karim ME. (2021, February). ESPACOMP-20-011: Comparison of statistical methods to address treatment nonadherence in pragmatic trials with only baseline covariate-measurements. In International Journal Of Clinical Pharmacy (Vol. 43, No. 1, pp. 298-298). Netherlands: Springer. https://doi.org/10.1007/s11096-020-01213-y (Objective 3)
1. Sanders E. Incorporating Partial Adherence into the Principal Stratification Analysis Framework, Statistics, UBC, MSc thesis, 2019. (Objective 2)
2. Mosquera,L. Exploring inverse probability weighted per-protocol estimates to adjust for non-adherence using post-randomization covariates : a simulation study, Statistics, UBC, MSc thesis, 2020. (Objectives 1)
1. Karim ME (joint work with Hossain MB) Implications of choosing different imputation methods while inferring about per-protocol effects of sustained treatment strategies, ESPACOMP Conference (Virtual conference), Seraing, Belgium; 21 Oct 2021. https://www.youtube.com/watch?v=pm8OOlPh3MU (Objective 1)
2. Hossain MB (joint work with Karim ME) Addressing differential medication non-adherence in pragmatic trials with point-treatment settings: a simulation study. 25th ESPACOMP Conference (Virtual conference), Seraing, Belgium; 21 Oct 2021. (Objective 3)
3. Hossain MB (joint work with Karim ME) Comparison of statistical methods to address treatment nonadherence in pragmatic trials with only baseline covariate-measurements. 24th ESPACOMP: International Society for Medication Adherence Conference (Virtual conference), Seraing, Belgium; 10 Nov 2020. (Objective 3)
4. Mosquera LK (joint work with Karim ME) Evaluating Adjusted Per-Protocol Effect Estimators in Pragmatic Trials to Address Treatment Non-Adherence. 24th ESPACOMP: International Society for Medication Adherence Conference (Virtual conference), Seraing, Belgium; 10 Nov 2020. (Objective 1)
5. Hossain MB (joint work with Karim ME) Review of statistical methods to address treatment nonadherence in the pragmatic trial context. 41st Annual Conference of the International Society for Clinical Biostatistics (ISCB), Kraków, Poland, August 18, 2020 [RP3.28] (Objective 3)
6. Mosquera LK (joint work with Karim ME) Properties of Adjusted Per-Protocol Effect Estimators to Address Treatment Non-Adherence in Pragmatic Trials. 41st Annual Conference of the International Society for Clinical Biostatistics (ISCB), Kraków, Poland, August 18, 2020 [RP3.26] (Objective 1)
7. Mosquera LK (joint work with Karim ME) Comparing instrumental variable and naive methods for estimating the causal effect of treatment in pragmatic trials with non-compliance, The 2019 Atlantic Causal Inference Conference (ACIC), Montreal, May, 2019 (Objective 1)
1. Sanders E (joint work with Gustafson P, Karim ME) Incorporating Partial Adherence into the Principal Stratification Analysis Framework, Annual Meeting of the Statistical Society of Canada, Calgary, May, 2019 (Objective 2)
2. Hossain MB (joint work with Karim ME) Comparing methods to address sparse follow-up issues in estimating per-protocol effects in pragmatic clinical trials: a simulation study. The ninth annual Canadian Statistics Student Conference (Virtual conference), Ottawa, Canada; 26 May 2021 https://drive.google.com/file/d/15Cz9Dwup6fc407AbeEJ5TNUxZ6IY_AK4/view (Objective 1)
3. Hossain MB (joint work with Karim ME) Statistical approaches to deal with treatment nonadherence in the pragmatic trial context. Canadian Statistics Student Conference 2020 (Virtual conference), Ottawa, Canada; 30 May 2020. https://www.youtube.com/watch?v=SZYp0aSz8Y4 (Objective 3)
4. Hossain MB (joint work with Karim ME) Comparing statistical methods in estimating per-protocol effects to address sparse follow-up issue in pragmatic clinical trials with treatment non-adherence. 6th Canadian Conference in Applied Statistics (Virtual conference), Montreal, Canada; 15 May 2021 https://www.youtube.com/watch?v=fZL5mRhvLqs (Objective 1)
Workshop and Seminar Presentations:
1. Hossain MB (joint work with Karim ME) Performance of statistical methods to address treatment non-adherence in pragmatic clinical trials with point-treatment settings: a simulation study. 2021 SORA-TABA Annual Workshop & DLSPH Biostatistics Research Day, May 27-28, 2021, Online. (Objective 3)
2. Sanders E (joint work with Gustafson P, Karim ME) Incorporating Partial Adherence into the Principal Stratification Analysis Framework, Statistics Seminar, Department of Statistics, University of British Columbia, August 15, 2019. (Objective 2)
Sanders, E, Gustafson, P, Karim, ME. Incorporating partial adherence into the principal stratification analysis framework. Statistics in Medicine. 2021; 40: 3625– 3644. https://doi.org/10.1002/sim.8986
Hossain, M.B., Mosquera, L. & Karim, M.E. Analysis approaches to address treatment nonadherence in pragmatic trials with point-treatment settings: a simulation study. BMC Med Res Methodol 22, 46 (2022). https://doi.org/10.1186/s12874-022-01518-8
Ehsan Karim, PI ✉︎
Paul Gustafson
Joan Hu
Hubert Wong
Samar Hejazi
Sharon Roman
Derek Ouyang
Md Belal Hossain
Lucy Mosquera
Eric Sanders
