State of the Art
When Tuberculosis Treatment Fails
A Social Behavioral Account of Patient Adherence
ESTHER SUMARTOJO
Centers for Disease Control and Prevention, National Center for Prevention Services, Division of Tuberculosis Elimination, Atlanta, Georgia
CONTENTS
Introduction
Research on the Scope of Nonadherence
Research on Methods of Measuring Adherence
Research on Factors Predicting Adherence
Research on Cultural Factors Influencing Adherence
Research on Interventions to Improve Adherence
Treatment strategies
Incentives
Social incentives
Educational interventions
Directly observed therapy
Community health workers
Comprehensive services
Summary and Conclusions
Why has tuberculosis, a treatable and curable disease, proved
so resistant to elimination? Why, in fact, are the numbers of cases,
particularly drug-resistant cases, increasing? Almost without ex-
ception, experts acknowledge the central role of patient adher-
ence in this dilemma, and increasingly they identify poor adher-
ence as a behavioral problem. This report provides an account
of the research on patient adherence as it relates to the treatment
and prevention of tuberculosis. It summarizes the literature on so-
cial and behavioral factors that relates to whether patients take
antituberculosis medicines and complete treatment, and it sug-
gests issues that require the attention of researchers who are in-
terested in behavioral questions relative to tuberculosis.
Behavioral and social research on patient adherence is par-
ticularly important at this time because of changes in the epidemi-
ology of the disease. Since the early 1980s, the number of cases
of tuberculosis has increased in the United States because of the
emergence of HIV/AIDS (1); large numbers of immigrants from
countries where tuberculosis rates are high (2); adverse social
circumstances including poverty, homelessness, and substance
abuse (3, 4); and deterioration of the health-care infrastructure
(5). Drug-resistant and multidrug-resistant strains of tuberculosis
have emerged (6) and are attributed in large part to poor adher-
ence to treatment. Research on adherence is also important be-
cause it should assist in the collection of more accurate data in
clinical trials of antituberculosis drugs.
Treatment adherence is a particular problem in tuberculosis.
Although tuberculosis is not a chronic disease, its prevention and
(Received in original form August 15, 1992 and in revised form January 13, 1993)
Correspondence and requestsfor reprints should be addressed to EstherSumar-
tojo, Ph.D., Centers for Disease Control and Prevention, National Center for
Prevention Services, Division of Tuberculosis Elimination, (E-10), Atlanta, GA
30333.
Am Rev Respir Dis Vol 147. pp 1311-1320, 1993
cure require the use of medications for long periods, 6 to 24
months. Multiple medications are needed, the regimens may be
complicated, and patients may experience unpleasant side effects.
Patients for whom preventive therapy is prescribed have no symp-
toms to convince them of the importance of taking their medica-
tion. Although medications are free in many public health clinics,
patients often have to endure long waits in crowded facilities be-
fore being seen. Patients may experience other barriers to adher-
ence: competing demands on their time, the contradictory norms
or expectations of their families or cultural groups, difficulties in
communication between patient and health care providers, or the
poor quality of the tuberculosis control infrastructure in their areas.
Some patients with tuberculosis also have problems with drug
abuse or mental illness that impair their ability to adhere to a med-
ical regimen. Generally, the reasons for poor adherence are not
only multifaceted and complex but range from characteristics of
the individual patients to qualities of the social and economic en-
vironment that influence all tuberculosis prevention and control
activities.
The word compliant, typically used to refer to a patient who
completes treatment, has the unfortunate connotation that the pa-
tient is docile and subservient to the provider. Given that patients
decide independently about taking medications, away from the
presence of the provider, and that effective completion of treat-
ment demands their independent action, compliance might seem
an undesirable characteristic. A better word, which reflects the
active role of the patient in self-management of treatment and the
importance of cooperation between patient and provider, is ad-
herence. This word is used as an alternative in many publications,
and it will be used in this report.
Several books and articles provide excellent reviews of the re-
search literature on patient adherence with medical treatment
(7-12). This literature includes descriptive studies about the scope
of the problem of poor adherence and methods for measuring
it, as well as empirical studies that attempt to explain, predict,
or change adherence behaviors. Research on adherence and
tuberculosis falls into similar categories; the following sections
review the tuberculosis literature on the scope of nonadherence,
methods of measuring adherence, research on predicting poor
adherence, studies of cultural factors that influence adherence,
and evaluations of interventions designed to increase patient ad-
herence. A few of the studies do not deal specifically with tuber-
culosis, but they have particular relevance for tuberculosis preven-
tion and control.
RESEARCH ON THE SCOPE OF NONADHERENCE
Nonadherence is cited as a major barrier to the control and elimi-
nation of tuberculosis at the level of public health (13, 14), as a
1312
serious problem in the treatment of individual patients (15), and
in the development of drug-resistant strains of the disease (6).
Data collected by the Centers for Disease Control and Prevention
indicate that approximately one fourth of patients with active tuber-
culosis fail to complete the usual six-months course of treatment
within 12 months, and that approximately one-third fail to com-
plete preventive therapy (CDC, Tuberculosis Program Manage-
ment Report for 1990). According to some accounts, nonadher-
ence approaches 90% among active patients who are also
homeless and alcoholic (5, 16). Many published reports on tuber-
culosis treatment programs refer to the problem of patient nonad-
herence.
RESEARCH ON METHODS OF MEASURING ADHERENCE
We have no gold standard by which to measure adherence to an-
tituberculosis medications. Techniques include providers' esti-
mates, patient self-report, measures of appointments kept, pill
counts, records of drugs picked up by patients, and assays of the
presence of drugs in urine. In addition, some researchers have
used pill boxes or bottle caps that incorporate microelectronic
devices to record when and how frequently pills are removed from
the containers.
Research in a general medical clinic indicated that physicians'
predictions of nonadherence are accurate in fewer than 50% of
cases, but they are more accurate when the patients are known
to be alcoholic or when physicians consider their relationship with
the patient unsatisfactory (17). In a study of patients with tubercu-
losis, physicians identified only 32% of nonadherent patients and
incorrectly identified 8% of adherent patients as nonadherent (18).
These data indicate that staff in tuberculosis clinics cannot pre-
dict adherence well, except when they have previous experience
with a patient on which to base a prediction.
Patient self-report would not seem to be a reliable measure
of adherence because of problems such as forgetfulness, unwill-
ingness to admit not taking medication, or fear of the medical pro-
vider. One researcher showed that 28% of patients with tubercu-
losis incorrectly reported taking medication as prescribed (19).
Nevertheless, there is some evidence that patients with a variety
of medical conditions who have had some experience with a regi-
men can predict their own level of adherence (20) and that care-
ful questioning by providers may yield correct information about
adherence (21, 22). In a comparison study of adherence and
nonadherence in patients with tuberculosis in India (23), par-
ticipants reported if and how often they forgot to take medications.
Nonadherent patients were significantly more likely to report for-
getting medications. The investigators concluded that self-report
is an acceptable measure of adherence, particularly when patients
have agreed to share responsibility for their treatment. The ac-
curacy of self-report remains controversial; structured and cul-
turally appropriate interview techniques to elicit patient self-report
deserve further study.
Several antituberculosis medications can be identified through
urine assays (24), although the most commonly used assays are
for rifampin (RIF) and isoniazid (INH). RIF is easily identified be-
cause it turns the urine and other body fluids red-orange. INH
can be identified through simple chemical analysis or with the
use of paper test strips (25, 26). However, each of these tests has
disadvantages: they are influenced by the patient's rate of metab-
olism of the drug, they may show a positive result even if the medi-
cation has not been taken regularly, and patients occasionally re-
fuse to provide urine samples for testing.
Several experimental microelectronic devices are available that
record when and how often a pill is extracted from a box or bottle,
AMERICAN REVIEW OF RESPIRATORY DISEASE VOL 147 1993
and they have been used to measure adherence in patients with
tuberculosis (27, 28). Obviously, these devices do not verify
whether medication is actually ingested, so they can estimate ad-
herence only. When effective, they measure both the daily fre-
quency and the long-term duration of box or bottle openings. The
devices are subject to being dismantled or lost, left open by pa-
tients who find them difficult to operate, or opened frequently be-
cause of curiosity (CDC: unpublished data).
None of these measures of adherence is completely accurate,
and researchers need to use multiple measures to estimate true
adherence. Each measures a different aspect of behavior. For this
reason, researchers should consider that patients may be adher-
ent on one measure but not on another (e.g., a patient fails to at-
tend clinic appointments, but does have drug metabolites in his
or her urine).
RESEARCH ON FACTORS PREDICTING ADHERENCE
The ultimate goal of predictor studies is to give providers a strategy
for distinguishing those patients who may be nonadherent in the
future or for assisting patients whose poor clinical progress is due
to nonadherence. Knowledge about factors associated with poor
adherence may also suggest how interventions can be planned
to improve adherence. Studies of adherence predictors are im-
portant for at least one additional reason. Good research on predic-
tors will challenge or support the conventional wisdom about the
causes of nonadherence. For example, providers of tuberculosis
services report that the following patient characteristics predict
poor adherence: homelessness, substance abuse, history of poor
adherence, emotional disturbance, lack of transportation, be-
havioral problems, dissatisfaction with clinic scheduling, forget-
fulness, mental retardation, lack of family or social support, mi-
grant status, illiteracy, unemployment, low income, minority status
(29). This list, developed from the reports of providers (and al-
most all patients with tuberculosis probably have at least one of
these characteristics), reflects their observations and experience,
but it excludes environmental, structural, and operational factors
that are beyond the patient's control. It is important that research-
ers challenge assumptions that attribute adherence problems ex-
clusively to patients; such attributions may bias providers against
certain kinds of patients and limit the providers' willingness to make
needed changes in health care services.
Demographic factors such as age, sex, race, ethnicity, occu-
pation, income, and education are often included in studies of
successful treatment of tuberculosis. However, according to
studies of other diseases, demographic variables are inconsis-
tent or unreliable predictors of patient adherence (7, 10, 11). In
addition to the poor predictive power of demographic factors, their
use presents two important problems. First, variables such as age,
income, or occupation are not inherently causal. Rather, they rep-
resent other factors about the patient that may be the real causes
of poor adherence such as lack of access to good information
about health and illness, lack of financial resources to support
medical treatment, or mistrust of the culture of the health care
system. Second, intervention by a tuberculosis treatment program
will not alter demographic characteristics. That is, the provider
cannot change the patient's age or sex to improve adherence. In
fact, studies intended to associate demographic variables with
adherence to antituberculosis medication give inconsistent results
that are difficult to interpret as causal explanations or to develop
into intervention strategies. Older age has been associated with
better adherence in at least one study (30), but it has shown no
association in others (23,31,32). Men have been shown to de-
fault from treatment more often than women (33), but sex has not
State of the Art: When Tuberculosis Treatment Fails
been associated with adherence in other studies (23, 30, 32). Sev-
eral socioeconomic variables seem to have more consistent rela-
tionships with poor adherence, including low family income (23,
34, 35) and low educational level (31, 32, 35). As noted earlier,
these socioeconomic variables may be markers for other factors
that may in fact lend themselves to interventions. For example,
if lack of education relates to poor knowledge of health care, one
intervention would be effective exposure to health education. Re-
search evaluations of educational strategies would be needed.
Several studies have dealt specifically with adherence predic-
tors among patients with tuberculosis. Corcoran (35), who stud-
ied 90 patients in DUblin, Ireland, concluded that low income, low
education, and excessive use of alcohol (more than 21 drinks per
week) were significantly associated with nonadherence. Patients
who were married and living with their families were significantly
more adherent than patients who were single, separated, or
widowed.
Nazar-Stewart and Nolan (36) studied the characteristics of
nonadherent men receiving preventive therapy in a homeless shel-
ter in Seattle. The men were asked to name persons who could
be contacted in case of emergency, and those with contacts in
other states were four times more likely to be nonadherent than
those with in-state contacts. Adherent men were more likely to have
been born outside the United States and to have other medical
or psychiatric conditions than were nonadherent men. Men with
records of alcoholism were slightly less likely to be adherent. Ad-
herent men also used the shelter more frequently. They were prob-
ably those who, for a variety of reasons, needed to stay in the shel-
ter, and while there they received supervised preventive therapy.
Alcabes and coworkers (31)studied adherence to INH preven-
tive therapy among 63 adolescent inmates at Rikers Island jail
in New York City. Inmates picked up daily medications at the jail's
pharmacy, and adherence measures were taken from pharmacy
records. The investigators found that having an eleventh-grade
education or higher predicted better adherence. Patients were in-
terviewed to determine their knowledge of tuberculosis and preven-
tive therapy. Inmates who knew that tuberculosis is preventable,
that their tuberculin test results were positive, and that INH must
be taken daily for a year, were significantly more adherent than
were inmates without this knowledge. Those who correctly as-
sociated their positive skin test results with infection were less
adherent than were those who believed incorrectly that the reac-
tion indicated active disease.
Two studies are notable because they describe complex sets
of predictor variables. Dubanoski's study (37) of adherence predic-
tors among patients receiving preventive therapy in Hawaii draws
on the theory of reasoned action of Fishbein and Ajzen (38).
Dubanoski posited that adherence behavior is influenced by the
patient's initial intention to take antituberculosis medication. This
behavioral intention is in turn influenced by personal and social
factors. The researcher concluded that patients' reported inten-
tions to take medicines were significantly associated with a posi-
tive attitude about the effectiveness of INH, strong feelings of be-
ing able to cope with the demands of the regimen, and positive
beliefs about taking preventive therapy that were in turn influenced
by the support of participants' social reference groups. However,
the strongest predictor of actual adherence after 6 months of ther-
apy was adherence after 1 month of therapy. This outcome sug-
gests the need for early and careful assessment of adherence
and early intervention with nonadherent patients. It also calls for
additional study of the influence that personal and social variables
have an adherence.
Barnhoorn and Adriaanse (23) reported on interviews with 52
adherent and 50 nonadherent patients with tuberculosis in India.
1313
Adherent patients reported higher incomes and were more likely
to self-report forgetting to take medication. In terms of knowledge
about tuberculosis, 79% of adherent compared with 460/0 of nonad-
herent patients knew the required length of treatment. Adherent
patients complained more about fever and loss of weight and appe-
tite, and they had stronger beliefs in the severity of their disease
and in the importance of healthy behaviors such as good nutri-
tion and visiting the health center. Strong social support differen-
tiated adherent patients: 56% compared with 28% of nonadher-
ent patients reported having families who were positive about their
taking medication, and 83% compared with 58% felt supported
by their doctors.
More systematic, theory-based work needs to be done by tuber-
culosis researchers. In addition, more work is needed on adher-
ence predictors in specific settings such as prisons and home-
less shelters so that the effects of the particular characteristics
of various settings can be known. Finally, predictor studies should
lead directly to research on the effectiveness of interventions to
improve adherence.
RESEARCH ON CULTURAL FACTORS
INFLUENCING ADHERENCE
In 1966,Jenkins (39) asked insightful questions about the influence
of culture on a person's perceptions of tuberculosis. That is, do
the various classes or ethnic groups have different views of tuber-
culosis and, if so, are these differences related to each group's
own experience with tuberculosis? Also, in the 1960s, Fox (14)
noted that local or regional beliefs affect patient adherence in im-
portant ways.
Rubel and Garro (40), who reviewed anthropologic research
on culture and tuberculosis, reported that many Hispanics attrib-
ute tuberculosis to folk illnesses such as "wasting sickness" or
to grippe or bronchitis, for which over-the-counter medicines are
taken. After hospitalization for tuberculosis, Hispanic patients risk
rejection by their families and may not be allowed back into the
household. Menegoni (41), who studied utilization of health care
services for tuberculosis by Indian groups in southern Mexico,
concluded that a person's perceptions of illness and the quality
of medical services influence their response to tuberculosis con-
trol programs.
Some researchers have described the impact of culture or lo-
cal beliefs on adherence to treatment of tuberculosis. Thompson
and coworkers (42) found that rates of adherence differed more
among seven European countries than among clinics within each
country, and they attributed this result to the differential influence
of national cultures. According to research done in a refugee camp
in Somalia (43), 20% of patients lost to treatment had been par-
ticipating in competing cultural activities, such as weddings or
festivals that precluded taking medicines or attending the clinic.
Teklu (44) found that social events and cultural beliefs sometimes
made it difficult for patients with tuberculosis in Ethiopia to take
medications. Housewives in Capetown, who were otherwise well-
informed about tuberculosis, listed cold weather, smoking, and
poor nutrition as main causes of the disease (45). Westawayand
Wolmarans (46) reported that South Africans aware of the high
incidence of tuberculosis in their area did not believe their own
family members could be affected by the disease. Farmer and
coworkers (47) found that approximately 85°A> ~f patients enrolled
in a study in Haiti believed sorcery to be a possible cause of tuber-
culosis. However, this belief was not associated with adherence.
Several studies have shown that persons from Central or South
American cultures fear the social implications of tuberculosis. Mata
(48) conducted surveys and focus groups with healthy adults and
1314
patients with tuberculosis in Honduras. He discovered strong
stigma associated with the disease; high percentages of healthy
adults feared that tuberculosis would lead to family rejection and
the loss of friends. Patients believed they were a danger to others
and that their disease was caused by aberrant behaviors such
as exposing themselves to "night air:' They reported trying over-
the-counter and home remedies for some time before the diag-
nosis of tuberculosis, and they suspected that the health care
providers did not really know what disease they had. Patients
reported rejection by others, and some said they would prefer death
to the social rejection associated with tuberculosis. In 1982, Robin-
son and Eiseman (Hispanic perceptions and beliefs toward tuber-
culosis: A pilot study; unpublished manuscript) surveyed English-
and Spanish-speaking Hispanics in shopping malls and churches
in San Fernando, California. Differentiating the two groups ac-
cording to the degree of their acculturation to mainstream Ameri-
can culture, the researchers found significant and compelling
differences. The less acculturated Spanish-speaking respondents
were significantly more likely to agree that most people who have
tuberculosis die of it, that tuberculosis cannot be cured, that peo-
ple who get tuberculosis live in dirty surroundings, that people
who live to please God will not get tuberculosis, that diagnosing
of tuberculosis is very expensive, that friends would think less
of them if the friends thought they had tuberculosis, and that they
should not get tested for tuberculosis because the medical authori-
ties do not respect them.
The influence of culture on patient behavior is integrated into
patient services by some tuberculosis treatment centers. For ex-
ample, the Utah Department of Health (49) reported the use of
foreign language videotapes with refugees who attend tuberculo-
sis clinics in Salt Lake City. The data, although primarily anecdo-
tal, indicate that patients who have seen the videotapes understand
the disease better and seem more adherent. Other tuberculosis
programs have attempted to use culturally and lingUistically ap-
propriate media to educate or influence patients. A number of
tuberculosis treatment programs employ foreign-born staff and
community outreach workers who have cultural backgrounds simi-
lar to those of patients in the area. Anecdotal reports are that these
providers work very successfully with patients from their own coun-
tries, and indeed they may bring persons into treatment or preven-
tive therapy who would not otherwise be willing to seek treatment.
The effects of these interventions require evaluation, but this sen-
sitivity to cultural factors shows an important trend in tuberculo-
sis services.
These studies and anecdotal reports suggest the value of eth-
nographic work that addresses the beliefs and attitudes of cul-
tural groups at highest risk for tuberculosis. They indicate that
tuberculosis services should be tailored to address the beliefs and
needs of various groups of patients. Needed are educational
materials that are culturally appropriate, that are innovative in the
use of media and messages, that have been designed using em-
pirical communication research techniques, and that have been
evaluated for effectiveness.
Some caution is warranted as researchers and program ad-
ministrators take greater interest in cultural factors. In designing
culturally appropriate interventions, they need to avoid stereotypic
or simplistic explanations for exceedingly complex cultural reali-
ties. The health care provider needs to try to relate to the patient
as a unique and respected person. The most appropriate use of
research findings on cultural factors may be a combination of ac-
curate cultural knowledge and skillful listening, the effective com-
munication of information, and a determination to interact with
the patient despite cultural differences. Finally, as Farmer and
AMERICAN REVIEW OF RESPIRATORY DISEASE VOL 147 1993
coworkers (47) suggest, based on observations of tuberculosis
services in Haiti, economic realities and the quality of programs
may override the influence of patient culture on treatment out-
comes.
RESEARCH ON INTERVENTIONS TO
IMPROVE ADHERENCE
Tuberculosis treatment centers use a variety of strategies to im-
prove patient adherence, and at least two sources recommend
specific strategies (50,51).A number of articles report on the evalu-
ations of interventions designed to improve adherence to antituber-
culosis medications. The interventions used in these studies fall
into seven categories: treatment strategies, incentives, social in-
centives, patient education, supervised therapy, community health
or outreach workers, and comprehensive services. Table 1 sum-
marizes major studies of these strategies.
Treatment Strategies
Effective treatment strategies include the use of short-course reg-
imens (52). For some patients, drug injections enhance the sig-
nificance of treatment, and they have been used successfully in
conjunction with supervised therapy (53). Special packaging has
helped patients remember to take their medications. Valeza and
McDougall (54) dispensed antituberculosis drugs weekly in blis-
ter calendar packs to more than 200,000 patients in the Philip-
pines from 1986 through 1989.They reported 80% adherence dur-
ing the study period compared with 41% adherence before the
intervention. Lange and coworkers (55) dispensed calendars to
patients receiving preventive therapy and asked them to record
the time at which they took their medication. The intervention did
not improve long-term adherence: most patients achieved good
adherence for a few weeks or months, although most then failed
to complete the prescribed year of therapy.
Incentives
Incentives such as money, bus tokens, or food are recommended
to improve adherence (51,56). However, little systematic research
has been carried out to assess the effects of these interventions.
Morisky and coworkers (57) combined education with a mone-
tary incentive for patients being treated for disease and patients
receiving preventive therapy. Of the patients receiving preventive
therapy, the intervention group was significantly more adherent
than a control group; the intervention had no significant effects
for the patients being treated for disease. The study did not
differentiate the effects of the monetary incentive and the educa-
tional intervention, so it is not possible to assess how each part
of the intervention influenced adherence.
Social Incentives
Social incentives involve activities that are socially rewarding for
patients. For example, in the study described next, patients could
be thought of as making a social contract with a provider: the pa-
tient's reward was derived from completing the agreement. Wur-
tele and coworkers (58) studied the effect of various levels of ver-
bal or written commitment by patients to return to have their skin
tests read. Compared with a control group, significantly more pa-
tients in the experimental groups returned for the reading. The
highest returns were among patients who gave both verbal and
written commitments. Edwards (59) reported on a successful
"breakfast club" for alcoholic patients. The investigator considered
a nutritional breakfast drink, along with the kinship that developed
among patients who came together for treatment, as incentives
State of the Art: When Tuberculosis Treatment Fails 1315

TABLE 1
STUDIES OF INTERVENTIONS TO IMPROVE ADHERENCE

Reference
Citation No. Intervention Adherence'

Valeza & McDougall, 1990 54 Treatment: blister packs and weekly > 80% vs 41%
appointments
Morisky et al., 1990 57 Health education and incentives: Prevention: 68%
Monthly counseling and monetary vs 38% (p < 0.001)
rewards Treatment: 93%
vs 90% (NS)
Wurtele et al., 1980 58 Social incentive: agreement to return Verbal and written: 78.2%
for skin test reading vs verbal: 76.9%
vs none: 69.8%
(p < 0.001)
Seetha et al., 1981 60 Education and social incentive: 59.9% vs 27.8%
patient and family education
Wobeser et al., 1989 61 1. Education and social incentive: 81% vs 25%
patient and family education (p < 0.001)
2. Education, social incentive, DOT: 52% vs 25%
patient and family education, daily (p < 0.01)
DOT
Hopewell et al., 1988 29 DOT 95.1% vs 89.7%
Nazar-Stewart & Nolan, 1992 36 DOT: homeless patients 48.9%
Westaway, 1988 66 Community health workers and DOT 91%
Manalo et al., 1990 67 Community health workers and DOT 90%
Curry, 1968 69 Comprehensive services: local 94.4% vs 66.3%
clinic and enhanced services
McDonald et al., 1982 71 Comprehensive services and DOT 80%
Werhane et al., 1989 16 Comprehensive services 86% vs 12%
(p < 0.0001)
Miles & Matt, 1984 72 Comprehensive services and DOT 93%
Brudney & Dobkin, 1991 65 Comprehensive services and DOT: 39%
homeless patients
Farmer et al., 1991 47 Comprehensive services 100% vs 56.7%

Definition of abbreviations: DOT = directly observed therapy.

, Outcomes for intervention versus comparison conditions and probability values are shown if reported in the research article. Studies differed
in methods used to measure adherence.

for adherence. Seetha and coworkers (60) attempted to draw on
the social incentive of familial involvement in adherence activi-
ties among patients with tuberculosis in India. Home health visi-
tors met with patients and family members to provide information
about tuberculosis and to encourage adherence. Compared with
a control group, patients receiving this social motivation collected
more monthly doses of antituberculosis drugs. An important ad-
ditional social incentive may have been the special attention and
encouragement provided by the health visitors.
When incentives succeed, the real source of motivation is dif-
ficult to identify. A health care provider typically delivers a mate-
rial incentive in a positive way so that it is difficult to differentiate
the tangible incentive from the social reward of positive attention
from and communication with the provider. The social aspect of
the intervention may be a much more powerful motivator than the
material object.
Rather than rewarding patients for adherence, some incentives
may serve as enablers, helping patients to access services or take
medication (personal communication: Betty Gore, R.N., M.S.N.,
South Carolina Department of Mental Health). For example, trans-
portation tokens or fruit juice to accompany medication may en-
able the patient to attend clinic or take medication rather than sim-
ply rewarding the patient for doing so. Incentives, enablers, and
social rewards influence patients in different ways; research is
needed that will identify which is most effective under different
circumstances.
Educational Interventions
Educational interventions are commonly used by tuberculosis
treatment centers, but evaluations typically examine education
in combination with other interventions (49,57,60,61) so that the
independent effect of education cannot be measured. Wobeser
and coworkers (61) used an educational intervention that included
personal and family counseling about information on tuberculo-
sis and the importance of adherence, and they daily supervised
patients receiving preventive therapy. The intervention produced
significantly higher adherence than did a control condition, but
the effect dissipated after the intervention ended. Moriskyand
coworkers (57) combined education with a monetary incentive,
and they produce higher adherence among patients receiving
preventive therapy, but the independent effect of education was
not studied.
Research on adherence to treatment for other diseases has
shown that education alone is not a highly successful interven-
tion. The best educational interventions instruct patients in ways
of changing behavior rather than simply providing information
about their diagnosis or disease (7, 62). One theoretical model
of behavior change (63) that may be relevant for tuberculosis re-
search predicts that educational interventions will be effective only
when patients are convinced that they need to alter their behaVior
and are open to new information on what they need to do. Re-
search on the effects of education on adherence might also in-
clude studies that use linguistically and culturally appropriate mes-
1316
sages and innovative media approaches such as video materials
or fotonovelas that are easy to read and show pictures of people
who are similar to the patient. This research needs to build on
accumulated knowledge about the effectiveness of education on
adherence to treatment for other diseases.
Directly Observed Therapy
Supervised, or directly observed therapy (DOT) clearly improves
adherence in patients with tuberculosis. In their research among
Canadian Plains Indians, Wobeser and coworkers (61) found that
daily observed therapy for preventive therapy achieved 81% ad-
herence, compared with 25% of a control group and 52% of pa-
tients receiving only an educational intervention. However, 10 wk
after the end of the intervention no differences remained among
the three groups. Not surprisingly, the effects of supervised ther-
apy do not hold after supervision stops. Williamson and cowork-
ers (64) found that 87% of patients receiving DOT, compared with
61% of patients receiving standard care, achieved sputum con-
version after 3 months of treatment. In a retrospective study of
patients with tuberculosis in 10 cities in the United States, investi-
gators compared treatment completion in patients who received
DOT and patients who were responsible for their own medication
(29). Approximately 5% of patients administered DOTfailed to com-
plete treatment, compared with approximately 10% of patients who
were responsible for their own medication. INH preventive ther-
apy was directly administered to 47 men in a homeless shelter
in Seattle (36). Nearly half the men completed the 6- to 12-month
regimen.
In addition to research findings, anecdotal reports from tuber-
culosis treatment personnel strongly favor supervised therapy, and
federal funding provides for the employment of outreach workers
to administer DOT. Responding to the increase in cases of tuber-
culosis and the threat of drug-resistant tuberculosis, new treat-
ment recommendations from the Centers for Disease Control and
Prevention emphasize the use of DOT for all patients with tuber-
culosis.
Additional studies of supervised therapy are needed. Many
different strategies are used by local tuberculosis programs to ad-
minister supervised therapy, including supervision in the clinic
or in the patient's home, or administration by another person who
is considered responsible such as a family member or an em-
ployer. No researcher has systematically described these various
approaches or evaluated their effectiveness. Supervised therapy
is costly and labor-intensive, and it is therefore important to learn
which of the strategies in use are the most effective. Increasingly,
researchers and program administrators will be asked to ensure
that these strategies recognize the needs and dignity of patients.
Community Health Workers
Paid community health workers are an essential part of many
tuberculosis programs in the United States and overseas. They
provide supervised therapy and follow the course of treatment and
preventive therapy. When delivering directly observed therapy, they
have an important impact on adherence, as described earlier. Out-
reach workers trained in social work are an effective part of the
activities of the Tuberculosis Research Centre in Madras, India,
and community workers are included in many other programs in-
ternationally. It is important to note that community workers may
be ineffective without careful supervision and coordination of their
activities (65).
In some tuberculosis programs, volunteer community health
workers assist in tuberculosis elimination activities. Typically, these
workers are from the patients' own communities. Approximately
AMERICAN REVIEW OF RESPIRATORY DISEASE VOL 147 1993
90% of the patients with tuberculosis who were part of a study
in South Africa completed treatment. These patients had been
visited 5 days each week for 6 months by volunteer workers. The
strong success of the program was attributed to the training and
support provided by the program to the volunteers and to the volun-
teers themselves, who were respected in the community, reliable,
literate, lived within walking distance of the patients, and had family
support for their volunteer activities (66). In Manila, church-based
community volunteers recruited patients and delivered more than
90% of twice-weekly doses of medication (67). The project's suc-
cess was attributed partly to the strong volunteer program already
in existence at the church. A project in Malawi trained community
workers to refer symptomatic persons for tuberculosis screening
(68). To date, community involvement with tuberculosis control
is not common in the United States, but the Centers for Disease
Control and Prevention is attempting to increase the involvement
of community-based organizations.
Comprehensive Services
The following studies assess the effect of comprehensive services
on adherence. The tuberculosis treatment programs that were
evaluated are unique because of their holistic view of the patient
and the patient's needs, and because each incorporated an as-
sortment of different methods to address adherence. An impor-
tant early study was conducted in three decentralized neighbor-
hood-based clinics in San Francisco (69). The clinics were opened
in 1962, and the study was a comparison of appointments that
had been missed in 1961 with appointments missed in 1967. Each
clinic was composed of a medical team with a physician, public
health nurses, and a clerk. Clinics were held at times and places
that were convenient based on surveys of patients and commu-
nity leaders, appointment schedules reduced waiting time for pa-
tients, persons needing social services were appropriately referred,
and nurses made home visits to patients who missed appoint-
ments. In establishing the program, community groups and health
department personnel were educated about the need for the
specialized services, and efforts were made to develop medical
teams that worked well together and held positive views of the
patients. After 5 yr of the program percentages of missed appoint-
ments fell from 34 to 6%. The success of the program was attrib-
uted to the specialized medical teams, an improved attitude on
the part of the professional staff, and greater convenience for pa-
tients. The tone of the program is reflected in the concluding state-
ments of the report:
If tuberculosis is to be effectively controlled, patient coopera-
tion is essential. If existing services are to be more effectively
utilized, they must be made more acceptable. This may require
modifications of services and changes in attitudes of person-
nel. (p. 1267)
A second project for difficult-to-reach, homeless patients was
described by McAdam and coworkers (70). Homeless men resid-
ing in a shelter achieved high rates of treatment completion in
a project in which an on-site medical team, consisting of a physi-
cian, nurse, and social worker, maintained close communication
with shelter staff. Similarly, McDonald and coworkers (71) described
a program for patients who had drug-resistant disease and histo-
ries of failed treatment because of poor adherence and drug abuse.
Patients received supervised therapy with drug injections 5 days
each week. The program also included case management with
the same physician, food during clinic visits, transportation and
home visits as required, flexible appointments, short waiting
periods in the clinic, and free medical care for other problems.
 State of the Art: When Tuberculosis Treatment Fails
It was reported that patients were treated with courtesy and re-
spect by their physicians. Adherence to clinic visits averaged 80%
during the first 6 months of intensive treatment. Although adher-
ence declined after the first 6 months, all but two of 21 patients
achieved sputum conversion.
A comprehensive clinic, managed by nurses trained, was es-
tablished in Chicago (16) to deal with nonadherent patients. The
intervention program included a pulmonary nurse specialist and
a nurse epidemiologist as well as careful case management. The
nurses attempted to establish strong relationships with patients
and to understand their beliefs about tuberculosis. Scheduled ap-
pointments reduced waiting times, patients were reminded of ap-
pointments for the following day and received close follow-up, pa-
tients were given money for transportation, and they were referred
for assistance with social, financial, and other medical problems
as needed. Treatment was simplified by the use of combination
capsules containing two drugs. The comprehensive clinic achieved
86% completion of treatment compared with 12% in a general
medical clinic. The poor success of the general clinic was attrib-
uted partly to inadequate patient records; often, data on medica-
tions or adherence were missing, laboratory results were not
recorded, and charts were frequently lost. It was concluded that
poor performance by health care teams is often responsible for
poor patient adherence.
In a refugee camp on the Thai-Cambodian border, 51 of 55
patients with active disease completed treatment, and only four
defaulted (72). The treatment program included daily supervised
therapy with 2 months of injections, the assistance of a bilingual
physician, assistance with food and housing for patients and their
families, discussions with the patients' families about the disease
and treatment, meals during clinic visits, and patient contracts
to stay in treatment for 6 months.
Residential shelters for homeless patients with tuberculosis
represent another form of comprehensive programming. For ex-
ample, the Bellevue Men's Shelter, established in New York City
in 1988,provides shelter, meals, medical assistance for other health
problems, assistance with job skills, and housing assistance for
residents, as well as supervised therapy for tuberculosis. At the
end of the first year of the program, 60% of initial residents either
completed treatment or remained on medication at the shelter (73),
and after 26 months 39% had completed treatment (65). Although
very low, this outcome compares favorably with data from a study
of similar patients in which all but 11% of patients were lost to
follow-up after initial hospital treatment for tuberculosis (5).
In rural Haiti, Farmer and coworkers (47) implemented a pro-
gram including daily visits from a community health worker, finan-
cial and nutritional assistance, travel expenses (e.g., for hiring a
donkey) to attend clinic monthly, and home visits when patients
failed to attend clinic. The comprehensive program included 30
patients, and 56 patients from a conventional program provided
a comparison group. Approximately 30 months after the start of
the program, 100% of patients in the comprehensive program were
considered cured, compared with 57% from the comparison group.
Overall, comprehensive services include health teams of per-
sonnel who assume responsibility for continuity of care, careful
case management and follow-up, attempts to make the clinic ac-
cessible and acceptable to patients, some system for provision
of social services to patients, and short-course treatment regimens
that include supervised therapy. These kinds of services have been
effective with patients receiving treatment for active disease, but
they could also be assessed for the delivery of preventive therapy.
McDonald and coworkers (71) emphasized that comprehen-
sive efforts can ensure treatment success with patients who would
1317
typically be treatment failures. Etkind and coworkers (74), concur-
ring with the importance of carefully designed services and pro-
grams, stated that when patients fail treatment, the responsibility
must be attributed in part to inadequate programs. Generally, com-
prehensive approaches emphasize tailoring treatment services
to the specific needs of patients, even to the extent of providing
assistance for problems other than tuberculosis such as social
and financial difficulties. Because of the services provided, com-
prehensive programs go to the patients; they actively seek out
the participation of patients by offering needed assistance. By con-
trast, many tuberculosis programs put "the onus on the patient
to get treated" (65). Many patients with tuberculosis suffer from
severe problems: homelessness, poverty, substance abuse. In suc-
cessful comprehensive services, the wide diversity of difficulties
faced by patients is recognized, and they are provided the addi-
tional assistance that makes the medical cure and control of tuber-
culosis possible.
This review indicates the need for further evaluative research
on strategies to improve adherence. In particular, studies should
focus on comprehensive service programs that address patients'
social as well as medical needs. Research is needed to identify
effective educational strategies and materials for patients and the
wider community. Incentives and enablers should be evaluated,
includlnq the social incentives that are implicit in the interaction
between patient and provider. Researchers should attempt to de-
fine the different methods used to deliver directly observed ther-
apy and to determine ways to supervise therapy that are cost-
effective as well as convenient and acceptable to patients.
SUMMARY AND CONCLUSIONS
Several conclusions about measuring adherence can be drawn.
Probably the best approach is to use multiple measures, includ-
ing some combination of urine assays, pill counts, and detailed
patient interviews. Careful monitoring of patient behavior early in
the regimen will help predict whether adherence is likely to be
a problem. Microelectronic devices in pill boxes or bottle caps have
been used for measuring adherence among patients with tuber-
culosis, but their effectiveness has not been established. The use
of these devices may be particularly troublesome for some groups
such as the elderly, or precluded for those whose life styles might
interfere with their use such as the homeless or migrant farm work-
ers. Carefully designed patient interviews should be tested to de-
termine whether they can be used to predict adherence.
Probably the best predictor of adherence is the patient's previ-
ous history of adherence. However, adherence is not a personal-
ity trait, but a task-specific behavior. For example, someone who
misses many doses of antituberculosis medication may success-
fully use prescribed eye drops or follow dietary recommendations.
Providers need to monitor adherence to antituberculosis medi-
cations early in treatment in order to anticipate future problems
and to ask patients about specific adherence tasks. Ongoing
monitoring is essential for patients taking medicine for active tuber-
culosis. These patients typically feel well after a few weeks and
either may believe that the drugs are no longer necessary or may
forget to take medication because there are no longer physical
cues of illness. Demographic factors, though easy to measure,
do not predict adherence well. Tending to be surrogates for other
causal factors, they are not amenable to interventions for behavior
change. Placing emphasis on demographic characteristics may
lead to discriminatory practices. Patients with social support net-
works have been more adherent in some studies, and patients
who believe in the seriousness of their problems with tuberculo-
1318
sis are more likely to be adherent. Additional research on adher-
ence predictors is needed, but it should reflect the complexity of
the problem. This research requires a theory-based approach,
which has been essentially missing from studies on adherence
and tuberculosis. Research also needs to target predictors for spe-
cific groups of patients.
There is clear evidence of the effect on adherence of cultur-
ally influenced beliefs and attitudes about tuberculosis and its treat-
ment. Cultural factors are associated with misinformation about
the medical aspects of the disease and the stigmatization of per-
sons with tuberculosis. Culturally sensitive, targeted information
is needed, and some has been developed by local tuberculosis
programs. A taxonomy of groups and their beliefs would assist
in the development of educational materials. Formative commu-
nications research is needed to identify the most effective educa-
tional messages for different groups. It is essential that this work
avoid oversimplified, stereotypic interpretations of cultural infor-
mation.
Tuberculosis programs use multiple strategies to improve pa-
tient adherence. The most successful of these interventions com-
bine outreach workers, supervised therapy, thorough case man-
agement, and additional social service assistance to patients. As
an alternative to paid outreach workers, some overseas programs
have drawn on paid or volunteer community health workers. Ma-
terial incentives for patients are difficult to assess because they
have been combined with other strategies or because they involve
the social reward of interaction with the provider.According to some
evidence, interaction with the provider, other patients, or family
members is an effective social reward for patients. Educational
interventions should emphasize specific adherence behaviors
rather than general information about tuberculosis or treatment,
and they should be appropriate for specific target groups. Directly
observed therapy is an effective but labor-intensive intervention;
research is needed to determine the specific strategies that are
most effective.
Additional aspects of the health care environment deserve the
attention of researchers. One of these is the quality of patient-
provider communication. It is essential that researchers describe
the quality of communication between various providers and pa-
tients with tuberculosis and evaluate the communication styles
and types of educational messages that lead to improved adher-
ence (for example, see references 75 and 76). There is anecdotal
evidence that some tuberculosis care providers have patients who
consistently complete treatment or preventive therapy. Research
might describe the communication strategies used by these suc-
cessful providers. Similarly, descriptive accounts of particularly
successful tuberculosis programs may reveal models to be emu-
lated. Unfortunately, many studies describe poor rather than ex-
emplary programs.
A second category for study has to do with the operation or
the infrastructure of tuberculosis services. Recent articles have
emphasized the relationship between characteristics of tubercu-
losis treatment centers and health care infrastructure and patient
adherence (24, 47, 65, 68, 77). Chaulet (77) states that the patient
should not carry primary responsibility for adherence, but rather
that adherence is "nothing more or less than the outcome of a
process involving a long chain of responsibilities, extending from
the decision-makers at the Health Ministry to the treating physi-
cians:' Nuyangulu and coworkers (68) attribute the high cure rate
of 87% for tuberculosis in Malawi to political stability, favorable
government policies on health, abundant food, standardized na-
tional tuberculosis practices, strong supervision and training of
tuberculosis program staff, community participation in case find-
AMERICAN REVIEW OF RESPIRATORY DISEASE VOL 147 1993
ing, and public education about tuberculosis. Farmer and cowork-
ers (47) conclude that patient adherence can be understood only
in terms of social and economic context.
In the context of the tuberculosis clinic, systematic operational
research is needed to assess the impact on adherence of case
management, data and records management, staff selection and
training, staff supervision and incentives, physician training,
management and organizational structure, and clinic communi-
cation with community organizations and other agencies. Gener-
ally, operational researchers need to ask how clinic policies and
practices affect individual patient behavior and how individual be-
havior reflects the quality of clinic operations. The challenge to
researchers is to acknowledge that adherence is influenced by
a complex array of factors, many of which are beyond the patient's
control, and to begin identifying and describing these factors.
At least three problems dominate the research on interventions
to improve adherence to antituberculosis medications. The ma-
jorityof articles fail to draw on relevant behavioral theory. The value
of a theoretical orientation is that it guides systematic and program-
matic research. Research that is not theory-based tends toward
trial and error, one-shot studies that describe behavior in very spe-
cific situations but provide no basis for generalization to other sit-
uations. A review of theoretical approaches is beyond the scope
of this review. However, Becker (62) has reviewed 10 relevant the-
oretical models of adherence, and Cummings and coworkers (78)
have written an excellent summary of 14 models of health behavior.
The second major problem is that most of the research has failed
to account for the multifaceted and complex quality of adherence.
For example, much of the work on adherence has been an at-
tempt to find a single, powerful solution to the problem. However,
the reasons for poor adherence are many, and many interven-
tions will be needed to address it. Finally, there is very little use
of randomized designs with control and experimental groups. As
researchers become interested in evaluating programs and inter-
ventions to improve adherence, more experimental approaches
will be used.
This review has two main goals. The first is to provide back-
ground for adherence researchers interested in tuberculosis and
to stimulate both tuberculosis specialists and behavioral scien-
tists to conduct studies in this area. The second goal is to chal-
lenge those who work directly with patients to take an innovative
and comprehensive approach to programming. Such an approach
may help program managers recognize and revise parts of their
programs that are not under the patient's control such as the
communication styles of the providers, the environment of the treat-
ment center, and the operational factors that may lead to ineffi-
cient or ineffective clinic practices. It may be helpful to acknowl-
edge that adherence requires behavior change, and most people
find it difficult to remain motivated to make even minor changes
in daily habits for long periods. This is particularly true for those
whose lives are severely stressed by competing difficulties such
as poverty, substance abuse, or homelessness.
Because of increasing numbers of cases, particularly multidrug-
resistant cases, tuberculosis has entered a dangerous new era.
Patient adherence is integral to both the source and the solution
of the tuberculosis problem. Further research is needed to de-
fine the social and behavioral dimensions of effective treatment
and control, and creative programming must take advantage of
the latest research.
Acknowledgment: The writer thanks Marie Morgan for editorial suggestions;
Dr.Dixie Snider, Dr.Samuel Dooley,Dr.Alan Hinman. LawrenceGeiter,Diana
Weil, Dr.John Sbarbaro, Henry Montes,and an anonymousreviewer for review-
ing the manuscript; and Dr.Barry Zimmerman for both his review and his en-
couragement.
State of the Art: When Tuberculosis Treatment Fails
References
1. CDC. Update: tuberculosis elimination: United States. MMWR 1990; 39:
153-6.
2. CDC. Tuberculosis among foreign-born persons entering the United States.
MMWR 1990; 39:1-21.
3. CDC. Prevention and control of tuberculosis among homeless persons.
MMWR 1992; 41:13-23.
4. CDC. Prevention and control of tuberculosis in U.S.communities with at-
risk minority populations. MMWR 1992; 41:1-11.
5. Brudney K, Dobkin J. Resurgent tuberculosis in New York City. Am Rev
Respir Dis 1991; 144:745-9.
6. CDC. Nosocomial transmission of multidrug-resistant tuberculosis among
HIV-infected persons: Florida and New York, 1988-1991. MMWR 1991;
40:585-91.
7. Meichenbaum D, Turk DC. Facilitating treatment adherence: a practitioner's
guidebook. New York: Plenum Press, 1987.
8. Gerber KE, Nehemkis AM, eds. Compliance: the dilemma of the chroni-
cally ill. New York: Springer, 1986.
9. DiMatteo MR, DiNicola DO. Achieving patient compliance: the psychol-
ogy of the medical practitioner's role. New York: Pergamon Press, 1982.
10. Haynes RB, Taylor DW, Sackett DL, eds. Compliance in health care. Balti-
more: The Johns Hopkins University Press, 1979.
11. Kirscht JP, Rosenstock 1M.Patients' problems in following recommenda-
tions of health experts. In: Stone Ge, Cohen, F. Adler NE, eds. Health
psychology: a handbook. San Francisco: Jossey-Bass, 1979; 189-215.
12. Shumaker SA, Schron EB, Ockene JK, Parker CT, Probstfield JL, Wolle
JM, eds. The handbook of health behavior change. New York: Springer,
1990.
13. Addington Ww. Patient compliance: the most serious remaining problem
in the control of tuberculosis in the United States. Chest 1979;76(Suppl:
741-3).
14. Fox W. Self-administration of medicaments: a review of published work
and a study of the problems. Bull Int Union Tuberc 1962; 32:307-31.
15. CDC. A strategic plan for the elimination of tuberculosis in the United States.
MMWR 1989; 38(S-3).
16. Werhane MJ, Snukst.:rorbeckG, Schraufnagel DE. The tuberculosis clinic.
Chest 1989; 96:815-8.
17. Mushlin AI, Appel FA. Diagnosing potential noncompliance. Arch Intern
Med 1977; 137:318-21.
18. Wardman AG, Knox AJ, Muers MF, Page RL. Profiles of non-compliance
with antituberculous therapy. Br J Dis Chest 1988; 82:285-9.
19. Preston OF, Miller FL. The tuberculosis outpatient's defection from ther-
apy. Am J Med Sci 1964; 247:55-8.
20. Kaplan RM, Simon HJ. Compliance in medical care: reconsideration of
self-predictions. Ann Behav Med 1990; 12:66-71.
21. Stewart M. The validity of an interview to assess a patient's drug taking.
Am J Prev Med 1987; 3:95-100.
22. Steele DJ, Jackson TC, Gutmann MC. Have you been taking your pills?
The adherence-monitoring sequence in the medical interview. J Fam
Pract 1990; 30:294-9.
23. Barnhoorn F, Adriaanse H. In search of factors responsible for noncom-
pliance among tuberculosis patients in Wardha District, India. Soc Sci
Med 1992; 34:291-306.
24. Burkhardt KR, Nel EE. Monitoring regularity of drug intake in tuberculous
patients by means of simple urine tests. S Afr Med J 1980; 57:981-5.
25. Henderson WT. The development and use of the Potts-Cozart tube test
for the detection of isoniazid (INH) metabolites in urine. J Arkansas Med
Soc 1986; 82:445-6.
26. Schraufnagel DE, Stoner R, Whiting E, SnuksHorbeck G, Werhane MJ.
Testing for isoniazid: an evaluation of the Arkansas method. Chest 1990;
98:314-6.
27. Moulding T, Onstad GD, Sbarbaro JA. Supervision of outpatient drug therapy
with the medication monitor. Ann Intern Med 1970; 73:559-64.
28. Cheung R, DickinsJ, Nicholson PW, Thomas ASC, Smith HH, Larson HE,
Deshmukh AA, Dobbs RJ, Dobbs SM. Compliance with anti-tuberculous
therapy: a field trial of a pill-box with a concealed electronic recording
device. Eur J Clin Pharmacol 1988; 35:401-7.
29. Hopewell PC, Judd K, Miller R, Luft H. Strategies for treating tuberculo-
sis: costs and outcomes. An unpublished report prepared for the Divi-
sion of Tuberculosis Elimination, Centers for Disease Control and Preven-
tion, 1988.
30. Bell J, Yach D. Tuberculosis patient compliance in the western Cape, 1984.
S Afr Med J 1988; 73:31-3.
31. Alcabes P, Vossenas P, Cohen R, Braslow C, Michaels D, Zoloth S. Com-
pliance with isoniazid prophylaxis in jail. Am Rev Respir Dis 1989;
140:1194-7.
32. Armstrong RH, Pringle D. Compliance with anti-tuberculous chemother-
apy in Harare City. Cent Afr J Med 1984; 30:144-8.
33. Henderson MA. Tuberculous in a developing country: experience of the
TB service at BMH, Dharan. J R Army Med Corps 1984; 130:22-30.
1319
34. Buri PS, Vathesatogkit P, Charoenpan P, Kiatboonsri S, Buranaratchada
S. A clinic model for a better tuberculosis treatment outcome and fac-
tors influencing compliance. J Med Assoc Thai 1985; 68:356-60.
35. Corcoran R. Compliance with chemotherapy for tuberculosis. Ir Med J
1986; 79:87-90.
36. Nazar-Stewart V, Nolan CM. Results of a directly observed intermittent
isoniazid preventive therapy program in a shelter for homeless men.
Am Rev Respir Dis 1992; 146:57-60.
37. Dubanoski JP. Preventive health behavior: a model of adherence predic-
tion (abstract). Dissertation Abstracts International 1988; 48(10-B):3152.
38. Fishbein M, Ajzen I. Belief, attitude, intention and behavior: an introduc-
tion to theory and research. Reading, MA: Addison-Wesley, 1975.
39. Jenkins CD. Group differences in perception: a study of community be-
liefs and feelings about tuberculosis. Am J Socio11966; LXXI:417-29.
40. Rubel AJ, Garro LC. Social and cultural factors in the successful control
of tuberculosis. Public Health Rep 1992; 107:626-36.
41. Menegoni L. Tuberculosis and health care in highland Chiapas, Mexico:
an ethnographic study (abstract). Dissertation Abstracts International
1991; 52(03-B):1379.
42. Thompson NJ, Snider DE, Farer LS. Variations in national health care prac-
tices and behaviours and their influence on international research. Int
J Epidemiol 1985; 14:457-62.
43. Shears P. Tuberculosis control in Somali refugee camps. Tubercle 1984;
65:111-6.
44. Teklu B. Reasons for failure in treatment of pulmonary tuberculosis in Ethio-
pians. Tubercle 1984; 65:17-21.
45. Metcalf CA, Bradshaw D, Stindt WW. Knowledge and beliefs about tuber-
culosis among non-working women in Ravensmead, Cape Town. S Afr
Med J 1990; 77:408-11.
46. Westaway MS, Wolmarans L. Knowledge and attitudes about tuberculo-
sis of black hospitalized TB patients (abstract). Am Rev Respir Dis 1990;
141:A204.
47. Farmer P,Robin S, Ramilus SL, Kim JY. Tuberculosis, poverty, and "com-
pliance": lessons from rural Haiti. Semin Respir Infect 1991; 6:254-60.
48. Mata JI. Integrating the client's perspective in planning a tuberculosis edu-
cation and treatment program in Honduras. Med Anthropol 1985;
Winter:57-64.
49. Sower P, Breckenridge-Potterf S. Utah Department of Health: refugee tuber-
culosis education program. TB Notes 1991, Division ofTuberculosis Elimi-
nation, Centers for Disease Control and Prevention, Spring:7.
50. Cuneo WD, Snider DE. Enhancing patient compliance with tuberculosis
therapy. Clin Chest Med 1989; 10:375-80.
51. Snider DE, Hutton MD. Improving patient compliance in tuberculosis treat-
ment programs. Atlanta: CDC, Public Health Service, U.S. Department
of Health and Human Services, 1989.
52. American Thoracic Society, CDC. Treatment of tuberculosis and tubercu-
losis infection in adults and children. Am Rev Respir Dis 1986;
134:355-63.
53. Sbarbaro JA. Compliance: inducements and enforcements. Chest 1979;
76(Suppl:750-6).
54. Valeza FS, McDougall AC. Blister calendar packs for treatment of tuber-
culosis. Lancet 1990; 335:473.
55. Lange RA, Ulmer RA, Weiss DJ. An intervention to improve compliance
to year-long isoniazid (INH) therapy for tuberculosis. Journal of Com-
pliance in Health Care 1986; 1:47-54.
56. ALA South Carolina. Enablers and incentives.: ALA of Columbia, South
Carolina, 1989.
57. Morisky DE, Malotte CK, Choi P, Davidson P, Rigler S, Sugland B, Langer
M. A patient education program to improve adherence rates with an-
tituberculosis drug regimens. Health Educ Q 1990; 17:253-67.
58. Wurtele SK, Galanos AN, Roberts MC. Increasing return compliance in
a tuberculosis detection drive. J Behav Med 1980; 3:311-318.
59. Edwards BB. The breakfast club. Florida Nurse 1988; 36:9.
60. Seetha MA, Srikantaramu N, Aneja KS, Singh H. Influence of motivation
of patients with their family members on the drug collection by patients.
Indian J Tuberc 1981; XXVIIJ:182-90.
61. Wobeser W, To T, Hoeppner VH. The outcome of chemoprophylaxis on
tuberculosis prevention in the Canadian Plains Indian. Clin Invest Med
1989; 12:149-53.
62. Becker MH. Theoretical models of adherence and strategies for improv-
ing adherence. In: Shumaker SA, Schron EB, Ockene JK, Parker CT,
Probstfield JL, Wolle JM, eds. The handbook of health behavior change.
New York: Springer, 1990; 5-43.
63. Prochaska JO, DiClemente CC. Towarda comprehensive model of change.
In: Miller W, Heather N, eds. Treating addictive behaviors. New York:
Plenum Press, 1986; 3-27.
64. Williamson DE, Hammack C, Burks G. Effect of directly observed therapy
with isoniazid and rifampin on 3 month sputum conversion in a Missis-
sippi rural health district (abstract). Am Rev Respir Dis 1986; 133:A205.
65. Brudney K, Dobkin J. A tale of two cities: tuberculosis control in Nicara-
gua and New York City. Semin Respir Infect 1991; 6:261-72.
1320
66. Westaway MS. Compliance with tuberculosis treatment the role of the
voluntary health worker. Report for the Tuberculosis Research Institute
of the Medical Research Council, Pretoria, Republic of South Africa. 1988.
07. Manalo F,Tan F,SbarbaroJA, Iseman MD. Community-based short-course
treatment of pulmonary tuberculosis in a developing nation. Am Rev
Respir Dis 1990; 142:1301-5.
68. Nuyangulu DS, Nkhoma WN, Salaniponi FML. Factors contributing to a
successful tuberculosis control programme in Malawi. Bull Int Union
Tuberc 1990; 66(Suppl:45-46).
69. Curry FJ. Neighborhood clinics for more effective outpatient treatment of
tuberculosis. N Engl J Med 1968; 279:1262-7.
70. McAdam JM, Brickner PW, Scharer LL, Crocco JA, Duff AE. The spec-
trum of tuberculosis in a New York City men's shelter clinic (1982-1988).
Chest 1990; 97:798-805.
71. McDonald RJ, Memon AM, Reichman LB. Successful supervised ambula-
tory management of tuberculosis treatment failures. Ann Intern Med 1982;
96:297-302.
72. Miles SH, Maat RB. A successful supervised outpatient short-course tuber-
culosis treatment program in an open refugee camp on the Thai-
AMERICAN REVIEW OF RESPIRATORY DISEASE VOL 147 1993
Cambodian border. Am Rev Respir Dis 1984; 130:827-30.
73. Adler JJ, Ruggiero D, Langhorne W, Heetderks A, Nivin B. Residential fa-
cility for homeless tuberculosis patients (abstract). Am Rev Respir Dis
1990; 141:A458.
74. Etkind S, Boutotte J, Ford J, Singleton L, Nardell EA. Treating hard to treat
tuberculosis patients in Massachusetts. Semin Respir Infect 1991;
6:273-82.
75. Stone GC. Patient compliance and the role of the expert. J Soc Issues
1979; 35:34-59.
76. West C, Frankel RM. Miscommunication in medicine. In: Coupland N, Giles
H, Wiemann JM, eds. Miscommunication and problematic talk. Beverly
Hills: Sage, 1991; 166-94.
77. Chaulet P. Compliance with chemotherapy for tuberculosis: responsibili-
ties of the Health Ministry and of physicians. Bull Int Union Tuberc 1990;
66(Suppl:33-35).
78. Cummings KM, Becker MH, Maile MC. Bringing the models together: an
empirical approach to combining variables used to explain health ac-
tions. J Behav Med 1980; 3:123-45.