Alt LI
В КАК т
с.
2
I'iUr; I H UrtK^L. .nA
Ь
I
RALEIGH
PHSB STUDIES
N. C.
Doc.
OCT 2 3 1978
A Special Report Series by the N,C. Department of Human Resources, Division of
Health Services, Public Health Statistics Branch, P.0. Box 2091, Raleigh, N.C.
No. 12 October 1978
SMALL AREA STATISTICS!
PROBLEMS WITH VITAL STATISTICS RATES
For many years — since 1932 — the Division of Health Services has published
resident birth and death rates for North Carolina counties and major cities. While
aware of certain inconsistencies between recording of residence on birth and death
certificates and Census enumerations of resident populations, we regarded the
magnitude of these inconsistencies as minimal and have proceeded to do the job we
felt charged to do, that is, to generate and disseminate data presumably helpful to
health officials and researchers In identifying the particular health problems of
cities, counties and various regions of the state.
Beginning with 1975 vital statistics reports, however, we had come to recognize
that annexations generally preclude reliable intercensal city population estimates in
the required detail, and we discontinued the computation of birth and death rates.
At the same time, we felt reasonably secure in the use of census data and correspond¬
ing vital event data and early this year computed 1968-72 age-race-sex-adjusted death
rates (cause-specific) for 38 cities, those that were incorporated and exceeded 10,000
population in 1970. Findings were reported in a PHSB study entitled "Mortality in
North Carolina Cities" (1) after satisfying ourselves that data counts were correct
(insofar as original sources were correct) and that adjustment procedures were
accurately programmed. Also at that time, we investigated whether Morganton's low
death rate (lowest of the 38 cities) might be due to the inclusion of Broughton
Hospital residents in rate denominators. According to Rand McNally (2) and later,
the Bureau of the Census (personal communication), Morganton's 1970 population of
13,625 excluded Broughton Hospital which was reported "outside corporate limits."
Thus, we published these data, providing — for the first time ever — comparisons of
mortality levels among cities on a cause-specific basis. These data, which showed
wi de variation in age-race-sex-adjusted city mortality levels, were widely publicized
by the news media and generated considerable interest and concern among health
officials and researchers.
Detection of Problems: The Morganton Case
Contrary to earlier information, we are now informed by the Bureau of the Census
that Morganton's 1970 population count did include Broughton Hospital residents in
accordance with the city map for 1970. This map did not identify Broughton Hospital
as a "political island" as was true of the map used in the i960 Census. Hence, since
deaths to residents of long-term health and penal institutions are by state regulation
allocated to the decedent’s place of residence prior to admission (3), Morganton's
"population at risk" (denominator used in rate computation) was substantially inflated,
Broughton Hospital accounting for about 2,100 of Morganton's 13,625 residents in 1970.
This contributed greatlv to Morganton's having the lowest of the 38 city death rates.
All other things being equal, other cities with sizable inmate populations within
their corporate boundaries similarly would have artificially low death rates. Among
the 38 cities, Salisbury and Raleigh appear particularly at risk, based on the
