Objectives: To demonstrate an important limitation of empirical tests of th
e proposition that medical care has contributed to longterm declines in mor
tality. Quasi-experiments cannot detect the effect of care that sustain, ra
ther than change, a downward trajectory. We demonstrate this limitation by
testing two hypotheses. One is that isolation of patients and antibiotic tr
eatment coincided with declines in tuberculosis (TB) mortality in Massachus
etts between 1850 and 1950. Another is that the introduction of Medicare in
the 1960s increased life expectancy at age 65. Results: The first hypothes
is is supported, for both patient isolation and streptomycin. The second is
not. The circumstances that could yield such results are explored. Conclus
ions: Epidemiologists and historians should cooperate to devise methods tha
t can resolve the issue of whether medical care has sustained the steady do
wnward trend in mortality witnessed over the last century-and-a-half. (C) 2
001 Elsevier Science Inc. All rights reserved.