Introduction

The following information was provided by Pat Nicol regarding the way deaths outside of a care facility (e.g. at home) are reported.

Approximately 30% of deaths occur out of hospital/facility (we can't tell if a person is admitted specifically for autopsy). Using the Manitoba Health Insurance Registry data as well, we can safely say we "know" about 98% of all deaths for a given year (some problems exist for December of the most recent year available due to delays in reporting). These are the ones we link to Vital Stats. There are more Manitoba Health deaths available so we probably see close to 99% of actual mortality even if we can't completely resolve them.

My understanding is that autopsies are typically only done for deaths under unknown circumstances. If an autopsy is ordered I assume the person shows up in hospital (admitted deceased and date of separation is not a good indicator of date of death). I have only seen such cases for persons transferred for autopsy - and there is some confusion as to whether the person died in the original facility or whether the hospital simply knew of the death when the abstract was coded.

Reporting Deaths

Deaths in hospital or long-term care are reported on the claims (again date of separation may not be date of death).

Other deaths are reported to the Manitoba health Insurance Registry system (eventually) either by the estate or via monthly cross-linkage of Manitoba Vital Statistics mortality data (by name, etc.) at Manitoba Health (see the Vital Statistics - Use of ICD-9 Codes Identifying Cause of Death concept for more information). Typically such reports take three or more months to process. Occasionally death is not linked and the registration stays open until someone points out the omission (such as on death of another family member) - in these cases the date may be arbitrary representing the date the registration was closed. A very few strange cases exist where the death of family head caused the Registration Number (REGNO) / (REGNO_CODE) to be transferred to the spouse - upon which event Manitoba Health was informed of the much earlier death of that spouse. This causes the registration to show a family "canceled" date more than two years old and causes the entire family (including the recent deceased family head) to be erased from the system (the recent death may not be captured in a snapshot).

Given the high agreement between Vital Statistics and Manitoba Health (our linkages are about 97% for recent years) there is no reason to believe that out of facility deaths are poorly represented. There must be a few deaths where Manitoba Health is not informed and the person's health insurance registration simply expires due to inactivity and inability to locate via mail etc.

Note: Our linkage to Vital Stats is based ONLY on deaths reported by Manitoba Health sources and those reported by Vital Stats.

We do NOT link to the overall registry so unreported deaths cannot be detected by linkage.

We do use the facility death reports to improve the registry reports (some die after registration is closed for some other reason, some deaths do not make it to registry due to discrepancies in name, etc., deaths before 1974/12 (and a few since) for members of families caused Manitoba Health to erase the individual registration rather than report the death in registry snapshots).