Introduction

This concept describes how Time to First Healthcare Contact is operationalized for immigration-related and interprovincial migration research at MCHP. The concept includes the following sections:


• a brief definition,
  
• a list of Data Sources where the data comes from,
  
• a description of the Methods employed,
  
• links to Research Findings reported in MCHP publications,
  
• any Cautions / Limitations found during development of this concept, and
  
• a relevant SAS® code example in the SAS code and formats section.

Definition: Time to first healthcare contact was measured in days from Coverage Start Date to first inpatient hospitalization or ambulatory / primary care provider visit. Individuals were censored if they died, moved out of province, were lost to follow-up (i.e., Cannot Locate), or reached the end of the study period without a healthcare contact (whichever came first).

Data Sources

The data used to operationalize this concept came from the following Repository databases:


• Immigration Refugees and Citizenship Canada (IRCC) – Permanent Residence Database
  
• Manitoba Health Insurance Registry data
  
• Hospital Abstracts data
  
• Medical Services / Medical Claims data

Methods

In the deliverable The Diversity of Immigrants to Manitoba, Migration Dynamics and Basic Healthcare Service Use by Urquia et al. (2020), Time to First Healthcare Contact was investigated for immigrants to Manitoba. The data range for this project included IRCC records from 1985 to 2017 and hospital abstracts and medical services / medical claims from January 1, 1985 to March 31, 2019. A time to event analysis is performed on the healthcare contact dates, in the same manner as a survival analysis, using both proportional hazards regression and Kaplan-Meier curves.

The following describes the basic algorithm in developing the methodology for this concept. It summarizes the SAS® code example that is provided in the SAS code and formats section below.

1. create the immigrant cohort file from the IRCC crosswalk file for all immigrants found in Manitoba;
   
2. create a landing file containing the arrival and landing dates from the two IRCC legacy and current landing files;
   
3. sort the landing file, and then merge with the immigrant cohort file using the unique IRCC identification number (IDNO);
   
4. sort the landing file by scrambled PHIN and landing date to identify and retain the earliest landing record per scrambled PHIN;
   
5. create a registry_coverage file of all Manitobans (not just immigrant cohort) using the MCHP REGCOV macro and then sort this file by scrambled PHIN, coverage start date and coverage end date;
   
6. merge the landing file with the registry_coverage file to identify coverage episodes during or after landing for immigrants, and occurring during study period for all Manitobans;
   
7. keep the first coverage episode (on or after landing for immigrants), and then calculate the number of person-years of residency based on the later of start of coverage or landing date to the earlier of end of coverage date or end of study period date;
   
8. make a SAS format from the list of unique scrambled PHINs for all cohort members, for selection of health care records;
   
9. concatenate hospital abstracts datasets, selecting records matching scrambled PHIN format and a separation date in study period; omit abstract type 4 and non-acute care facilities, and limit to transaction code 1 (in-patient records); remove duplicates;
   
10. concatenate medical services claims datasets, selecting records matching scrambled PHIN format and a service date in study period; limit to prefix 7 for physician visits;
    
11. run claims through NET_MED macro to combine related tariff claims and flag in-hospital and ambulatory visits;
    
12. limit claims to ambulatory visits and remove duplicates;
    
13. merge claims with hospital abstracts to produce a combined set of health care usage (HCU) records;
    
14. merge HCU records with cohort to remove records outside of person's coverage period;
    
15. sort by service date and keep first HCU record per individual scrambled PHIN;
    
16. merge first contact HCU record with cohort and calculate time to first contact or end of follow-up;
    
17. run a Cox proportional hazard regression to calculate hazard ratios, comparing immigrants to other Manitobans, then calculate confidence intervals from HR estimate and standard error;
    
18. run a LIFETEST procedure to generate a Kaplan-Meier curve of time to first contact, contrasting the two cohort groups;
    
19. produce output tables summarizing these and other statistics on the two groups.

Research Findings

In the deliverable The Diversity of Immigrants to Manitoba, Migration Dynamics and Basic Healthcare Service Use by Urquia et al. (2020) they investigated time to first healthcare contact for two cohorts: immigrants to Manitoba (from the IRCC data) and interprovincial migrants to Manitoba, and reported on this information.

Please see the following information from the on-line deliverable:


• Healthcare Service Use
  
• Figure 4.1: Time From Start of Coverage to First Healthcare Contact Among Immigrants and Interprovincial Migrants, 1985-2019

Cautions / Limitations

The following cautions were identified during development of this concept:


• there are different ways of deciding what counts as a healthcare contact, based on inclusions and exclusions, and the use of different data sources (e.g. hospital abstracts, medical services / medical claims, emergency department visits, etc.). NOTE: While in-patient hospital admissions and ambulatory care visits are the common primary contacts with healthcare, and are captured well in the administrative data on a province-wide basis, they do omit other important contacts such as Emergency Department (ED) visits. The ED visits have been incomplete in our past data, and have only been available province-wide in recent years, so including these for a larger study period would lead to uneven results.
  
• the data is limited to contact with the Manitoba healthcare system. Data for healthcare contact in other provinces/territories is not generally available in the Repository.