Concept: Location of Residence (Postal Code) Agreement in Administrative Data

Introduction

This concept contains a description of a method that investigates the level of agreement on "location of residence" using the postal code recorded in different administrative health data sets, and the findings of this investigation. The information in this concept comes from the MCHP deliverable, Cancer Data Linkage in Manitoba: Expanding the Infrastructure for Research, by Lix et al. (2016).

The accuracy of "location of residence" is important for conducting geographical analyses and how this relates to health and health care utilization, and to the validity of using different administrative health data sets to answer these types of research questions.

Methodology

As one aspect of data quality, Lix et al. (2016) examined the level of agreement between the six-digit postal code recorded in the Manitoba Cancer Registry and Treatment data and corresponding information in the Manitoba Health Insurance Registry data and Hospital Abstracts data. The Manitoba Health Insurance Registry contains dates of health insurance coverage and demographic information such as date of birth, sex, and postal code. Hospital abstracts, which capture data from all inpatient hospitalizations, also contain the six–digit postal code.

For Manitoba Cancer Registry data from 1984 to 2011, the cancer patient’s postal code at the time of diagnosis was compared to:


• the postal code in the Manitoba Health Insurance Registry, where the date of cancer diagnosis was within the start and end dates of valid health insurance coverage, ** and
  
• the postal code recorded on any hospital discharge abstract that had a date of hospital separation within 90 days of the cancer diagnosis date.

For patients with multiple cancers, only the first cancer diagnosis was used to conduct the analysis. This analysis was conducted for all cancer patients.

** - the Manitoba Health Insurance Registry data is updated every six months in the MCHP Data Repository.

Measuring the Level of Agreement

The level of agreement was estimated using the kappa (κ) statistic. The interpretation of κ adopted in this analysis was:


• κ < 0.20 is poor agreement,
  
• 0.20 <= κ <= 0.39 is fair agreement,
  
• 0.40 <= κ <= 0.59 is moderate agreement,
  
• 0.60 <= κ <= 0.79 is good agreement, and
  
• κ >= 0.80 is very good agreement (Altman, 1990).

Analyses were stratified by age group (< 40 years, 40–64 years, 65+ years) because geographic mobility tends to be higher amongst younger than older age groups.

Findings

The results of this investigation are reported below in Table 1. "Overall agreement was very good and was similar for both the Manitoba Health Insurance Registry and Hospital Abstracts data. When the results were stratified by age group, agreement was observed to be high and similar for the two oldest age groups, and lower for the youngest age group. For the youngest age group, agreement was higher for hospital discharge abstracts than for the Manitoba Health Insurance Registry, where only good agreement was observed."

Table 1: Estimates of Agreement (Kappa) for Six-Digit Postal Code Recorded in the Manitoba Cancer Registry and Other Administrative Data Sources
Using the Kappa Statistic and 95% Confidence Intervals



Age Group (years)	Manitoba Health Insurance Registry	Hospital Abstracts
Under 40	0.72 (0.72,0.73)	0.81 (0.80,0.81)
40-64	0.86 (0.86,0.87)	0.88 (0.88,0.89)
65+	0.87 (0.87,0.87)	0.89 (0.89,0.89)
Overall	0.86 (0.85,0.86)	0.88 (0.88,0.88)

This study demonstrates that location of residence at the time of cancer diagnosis, based on the six–digit postal code in the Manitoba Cancer Registry, compares favorably with location of residence captured in the Manitoba Health Insurance Registry and Hospital Abstracts data.

NOTE:

The information documented in this concept is available in the section of the deliverable titled Location of Residence Agreement.