Objective: To search for major gaps in medication-taking behavior pred
isposing patients to cardiovascular morbidity and mortality. Design: C
onvenience sample; cohort prospectively followed for less-than-or-equa
l-to 5 months. Setting: General internal medicine and cardiology clini
cs in a university medical center. Patients: From among 893 patients,
the authors identified 132 eligible individuals and entered 33 subject
s (25%) with chronic cardiovascular conditions, 1 - 3 chronic oral med
ications for these conditions, overall regimen of less-than-or-equal-t
o 6 drugs, regular visits at 1 -3-month intervals, literacy in English
, willingness to use electronic monitors, and physician permission to
participate. Outcome measures: Medication compliance rates and pattern
s by patient self-report, physician estimates, pill count, and electro
nic monitoring of pill vial opening. Results: Despite moderately compl
ex regimens (5.4 +/- 0.5 pills daily; range 1 - 11), most subjects too
k most medications according to the prescription: median intervals bet
ween pill vial openings were 1.00, 0.50, and 0.43 days for once, twice
, and three times daily dosing, respectively. Medication-taking gaps o
f greater-than-or-equal-to 2 times the prescribed interdosing interval
occurred for 48% of the patients. Patients' dosing patterns often pro
duced ''uncovered'' intervals (mean duration 3.7 days, range 0 - 2 5)
with doubtful pharmacologic effectiveness. These lapses were underesti
mated by patients and poorly perceived by their treating physicians, d
espite familiarity with their care. Baseline sociodemographic, psychos
ocial, medical system, or clinical characteristics did not predict the
patterns or degrees of medication noncompliance. Conclusions: Major t
reatment gaps occur frequently, even in carefully selected ambulatory
populations, and generally escape detection. The compliance patterns a
nd gaps may contribute to reported excesses of cardiovascular morbidit
y and mortality.