OBJECTIVES: To create a profile of individuals nonadherent to their medicat
ions in an age-stratified sample (ages 34-84) of community-dwelling rheumat
oid arthritis patients. The relative contributions of age, cognitive functi
on, disability, emotional state, lifestyle, and beliefs about illness to no
nadherence were assessed.
DESIGN: A direct-observation approach was used in conjunction with structur
al equation modeling. All participants were administered a preliminary asse
ssment battery. Medications were then transferred to vials with microelectr
onic caps that recorded medication events for all medications for the next
4 weeks.
PARTICIPANTS AND SETTING: A volunteer sample of 121 community-dwelling rheu
matoid arthritis (RA) patients were recruited from newspaper ads, posters,
and via informal physician contact from private rheumatology practices in A
tlanta and Athens, Georgia. Written verification of the RA diagnosis and a
disease severity rating were obtained from personal physicians before patie
nts were enrolled in the study. Patients were tested in a private physician
's office, and their medication adherence was monitored electronically for
a month in their every-day work and home settings.
MEASUREMENTS AND RESULTS: Structural equation modeling techniques were used
to develop a model of adherence behavior. Cognitive and psychosocial measu
res were used to construct latent variables to predict adherence errors. Th
e model of medication adherence explained 39% of the variance in adherence
errors. The model demonstrated that older adults made the fewest adherence
errors, and middle-aged adults made the most. A busy lifestyle, age, and co
gnitive deficits predicted nonadherence, whereas coping with arthritis-rela
ted moods predicted adherence. Illness severity, medication load, and physi
cal function did not predict adherence errors. Omission of medication accou
nted for nearly all errors.
CONCLUSION: Despite strong evidence for normal, age-related cognitive decli
ne in this sample, older adults had sufficient cognitive function to manage
medications. A busy lifestyle and middle age were more determinant of who
was at risk of nonadherence than beliefs about medication or illness. Thus,
practicing physicians should not assume that older adults have insufficien
t cognitive resources to manage medications and that they will be the most
likely to make adherence errors. Very busy middle-aged adults seem to be at
the greatest risk of managing medications improperly.