Table 3

Proposed methods for primary analysis and sensitivity analysis

Objectives	Outcome variable	Predictor variables	Method of analysis
Primary analysis
To identify the independent factors associated with being acute care HCU and potential effect modifiers	The classification of being HCUs or non-HCUs (ie, being HSUs or non-HSUs defined by acute care cost)	Clinical factors: Admission category, the Elixhauser comorbidity score.	Mixed effects logistic regression
		Sociodemographic factors: Patient’s age, sex, rurality of residence, marital status, immigrant status and visible minority.
		Socioeconomic factors: Work activity, occupation classification, the after-tax low-income status of a family, income adequacy deciles among Canadians, and the highest level of education.
		Interaction terms: Comorbidity scores and age, comorbidity scores and sex, comorbidity scores and income level
Sensitivity analyses
To analyse the robustness of results when HSUs are defined using other metrics	The classification of being HSUs or non-HSUs defined by the total length of stay, frequency of hospitalisations and frequency of ED visits	Clinical factors: Admission category, the Elixhauser comorbidity score.	Mixed effects logistic regression
		Sociodemographic factors: Patient’s age, sex, rurality of residence, marital status, immigrant status and visible minority.
To analyse the robustness of results when missing data is handled using multiple imputation	The classification of being acute care HCUs or non-HCUs (ie, being HSUs or non-HSUs defined by acute care cost)	Socioeconomic factors: Work activity, occupation classification, the after-tax low-income status of a family, income adequacy deciles among Canadians, and the highest level of education.
		Interaction terms: Comorbidity scores and age, comorbidity scores and sex, comorbidity scores and income level

ED, emergency department; HCU, high-cost user; HSU, high system user.