De-identifying a public use microdata file from the Canadian national discharge abstract database

Table 2 The Quasi-identifiers

Quasi-identifiers	Coding	# Categories
PROV_XXX	Province/region. Quebec data is not included in the DAD.
PROV_ALL	The territories are grouped into one category + 9 provinces	10
PROV_REGION	The country is divided into three regions (West, Central, and East), where Central consists of Ontario.	4
TOTAL_LOS_XXX	Total Length of stay
TOTAL_LOS_DAYS	Days up to 1 week, then in weeks up to 6 months, and top coded at 6 months +	31
TOTAL_LOS_WEEKS	Weeks up to 6 months everything longer than that is top-coded into a single category	25
AGE_GROUP	Five year intervals and top coded at 90 years	20
GENDER_CODE	unchanged	5¹
MRDx DIAG3 DIAG_BLOCK DIAG_CHAPTER	Different levels of coding detail of the most responsible diagnosis code.	8967 1435 195 23
CMG_CODE	These identify Case Mix Groups (CMGs), which are groups of patients with similar clinical and cost characteristics. They are based on most responsible diagnosis (MRDx) and other diagnosis and intervention information.	545
CCI_CODE SHORT_CCI	Different levels of coding detail of the principle intervention. Approximately 46% of the records had no interventions.	8780 569

This table displays the quasi-identifiers that were being analyzed and the number of response categories. There are two versions of the province and total length of stay fields as they represent different levels of detail.
¹ Five gender codes are used in the DAD: F-female, M-male, O-other (trans-sexual or hermaphrodite, U-undifferentiated(stillbirths only), and Z-missing.

ISSN: 1472-6947