Upload
amir-khan
View
216
Download
1
Embed Size (px)
Citation preview
Tuberculosis in Pakistan: socio-cultural constraints andopportunities in treatment
Amir Khana, b, John Walleyb, James Newellb,*, Naghma Imdadc
aAssociation for Social Development, Islamabad, PakistanbNu�eld Institute for Health, 71±75 Clarendon Road, Leeds, LS2 9PL, UK
cDataline, Islamabad, Pakistan
Abstract
This study explores the extent to which factors related to individuals, the care provision process, and the cultural
context in¯uence the behaviour of tuberculosis patients attending TB clinics in rural Pakistan, and examines thee�ects of disease on their personal lives. Thirty-six patients attending three TB treatment clinics were interviewed indepth. These patients were strati®ed by stage of treatment (treatment proceeding, treatment completed, default), sex
and by rural/urban status. Results indicate that the majority of patients were very poor, but nonetheless initiallychose to attend private practitioners. Normally their disease was correctly diagnosed as tuberculosis only afterrepeated visits to a succession of health care providers. Patients' knowledge about their disease was limited, and
doctors gave incorrect or only very limited health education. Most patients reported dissatisfaction with careprovided. Almost all patients reported problems with access to treatment, both in terms of time and money; thiswas particularly true of women, whose freedom to travel in Pakistan is limited. Potential causes of default appearedto be more closely linked to de®ciencies in treatment provision rather than patients' unwillingness to comply.
Largely because of a perception that TB was incurable, respondents were generally unwilling to disclose that theywere undergoing or had undergone TB treatment. For reasons related to con®dential access to treatment, this couldlead to default, perpetuating the perception of incurability, and hence causing a vicious circle. For TB programmes
to be successful in Pakistan, it is essential that this circle is broken; and this can only be done through provision ofgood quality TB care and education to improve the population's understanding that TB can be cured. In addition,patients' unwillingness to disclose to health care providers that they had already received previous treatment meant
that many patients were prescribed incorrect treatment regimes, potentially leading to the emergence of drug-resistant TB. In common with other researchers' ®ndings, no clear di�erences were found between those who hadcompleted treatment and those who had defaulted from treatment. This study was performed to provideinformation to assist the researchers to design potential TB treatment delivery strategies, and has proved invaluable
for this purpose. Strategies based on ®ndings from the study are currently being assessed using a randomisedcontrolled trial. # 1999 Elsevier Science Ltd. All rights reserved.
Keywords: Tuberculosis; Developing countries; DOTS; Pakistan
Social Science & Medicine 50 (2000) 247±254
0277-9536/99/$ - see front matter # 1999 Elsevier Science Ltd. All rights reserved.
PII: S0277-9536(99 )00279-8
www.elsevier.com/locate/socscimed
* Corresponding author. Tel.: +44-113-233-6950; fax: +44-
113-233-6997.
E-mail address: [email protected] (J. Newell)
Introduction
Tuberculosis is a problem of global signi®cance, esti-
mated to cause about 8 million new cases of disease
and about 3 million deaths each year, more than half
of which are in Asia. About three-quarters of all tuber-
culosis cases and deaths in developing countries are
concentrated in the economically productive age group
of the population (15±59 years) (Kochi, 1991). The
link between tuberculosis and HIV/AIDS, and the
emergence of multi-drug resistant tuberculosis as a
public health concern, have further added to the sig-
ni®cance of e�ective measures for the control of tuber-
culosis.
Short-course chemotherapy (SCC) for tuberculosis
has been shown to be one of the most cost-e�ective
health interventions (De Jonghe et al., 1994; Murray,
1994; Murray et al., 1991), and is recommended by
WHO and WB even in countries with very low levels
of health expenditure (Murray, 1994). However, in
most developing countries, treatment completion rates
are generally poor (Monterrosa, 1991), posing a
serious risk not only for the individual patient but also
for the community, contributing to failure to eradicate
the disease globally (Addington, 1979). The problem of
poor completion rates has led to the use of newer
strategies such as directly observed treatment using
short-course drugs (DOTS).
Many factors may lead to poor completion rates in
developing countries. The high cost of drugs (relative
both to per-capita health expenditure and to individual
income), and the long duration of so-called SCC, are
likely to be major contributors. Despite WHO/WB rec-
ommendations, many developing countries choose not
to or are unable to provide free SCC, and to most
patients in developing countries, the SCC drugs are
extremely expensive. From the patient's point of view,
SCC is a misnomer, since the course of treatment lasts
a minimum of six months, and since most patients'
symptoms are cured after 3±4 weeks of treatment,
patients are often unwilling or unable to complete
their treatment. In addition to these two factors, there
may be many further contributors to poor completion
rates. It is di�cult to list further general contributory
factors, since they are often country-, region- or dis-
trict-speci®c. A complex interaction of a wide range of
personal, socio-economic and organisational factors
determine the patient behaviour in any speci®c context.
For example, in many Asian countries, there is major
stigma attached to TB in women, while in some areas
of Africa this is not the case. It is therefore important
for all tuberculosis control programmes to perform
locally-based studies to investigate possible impedi-
ments to treatment completion.
In Pakistan, TB continues to be one of the majorcauses of morbidity and mortality. There are estimated
to be 220,000 sputum positive cases (Government ofPakistan, 1994), with an incidence of 85±100 new casesper 100,000 population per year. Previous studies in
Pakistan have found default rates of 66% (Sloan andSloan, 1981) and 72% (Liefooghe et al., 1997). In arecent (as yet unpublished) small-scale retrospective
cohort study in Pakistan, the rate of con®rmed cure orcompletion of treatment (where cure can not be con-®rmed) was estimated at around only 41% among newsputum-positive tuberculosis cases. The government of
Pakistan is considering implementation of a multi-million dollar nation-wide programme for tuberculosiscontrol, using DOTS as the key strategy.
It is now widely recognised that the method of deli-vering DOTS needs to be tailored to the socio-culturalsetting in which it is to be implemented. Compliance in
tuberculosis treatment has been studied in India andvarious other Asian countries, but due to wide di�er-ences in context, these ®ndings are of limited generali-sability to the situation in Pakistan. We have therefore
initiated a randomised controlled trial of DOTS, thepurpose of which is to evaluate and compare types ofDOTS suitable for the situation in Pakistan. As a pre-
cursor to this trial, we felt it necessary and importantto design and conduct a study to help us understandthe extent to which factors related to individuals, the
care provision process, and the cultural context in¯u-ence the behaviour of tuberculosis patients in Pakistan.This was intended to inform the process of designing
DOTS interventions which were acceptable and feas-ible and therefore likely to be e�ective. There is clearlylittle point in going to the trouble and expense ofRCTs to test strategies which are by the nature of the
socio-cultural context, unlikely to succeed.
Sample and methods
Three TB treatment centres were selected from three
districts of Pakistan (a specialist TB centre inRawalpindi; a local NGO-run TB clinic in Sahiwal;and a government district hospital in Gujranwala) onthe basis of their ability to provide acceptable TB
treatment (ie they had a quali®ed doctor in charge,and an adequate laboratory for diagnosis and follow-up). As part of the selection of the study sites, diagno-
sis and treatment procedures were observed and infor-mal discussions were held with the sta�. From patientsattending these centres, it was intended to select a total
of 36 `new' cases (ie patients who had never receivedtreatment for TB before the treatment they were cur-rently undergoing) for an in-depth sociological investi-
A. Khan et al. / Social Science & Medicine 50 (2000) 247±254248
gation, purposively selected to give a balance overthree factors: sex; stage of treatment; and urban/rural
residence (see Table 1 for details). The intention wasto use treatment facility records to select patients, butthe addresses recorded were found to be inadequate
for tracing patients at their home. It was thereforenecessary to adopt a revised procedure, as follows. Ateach of the selected treatment centres, consecutive
`new' patients were approached and asked if theywould be prepared to be interviewed at home abouttheir TB treatment. This process was continued until a
list of 50 patients per site (ie 150 in all) had been cre-ated, each of whom had consented and given theiraddresses to our ®eld sta�. Attempts were made tointerview the patients at their homes, progressing
through the list in order until the purposive samplehad been achieved. If even after repeated attempts apatient was not traceable, the next patient of the rel-
evant sex, stage of treatment and urban/rural statuswas visited.Each patient was given an in-depth, semi-structured
interview at his or her place of residence. The surveyteam comprised a senior male sociologist and a femaleresearcher. A specially developed interview schedule
including both open and closed questions was used toelicit the required information. The schedule was ®eld-tested before the survey and modi®ed where necessary.Data were collected on the pro®le of the patient and
his/her family, the patient's knowledge, attitude andpractices regarding TB and its treatment, and theacceptability to the patient of various DOTS options.
Responses to open questions were reviewed and codedby the senior team member. Themes were collatedmanually from transcripts of interviews. SPSS-PC was
used for analysing quantitative data.
Results
Although patients who had initially been selected to
be interviewed had agreed to meet the research teamfor detailed interviews, more than half of them werenot traceable at the address they had given.
Sociodemographic pro®le of the respondents
Thirty of the 36 interviewees were in the economi-
cally active age range of 18±59 years. Nineteen weremarried. Most respondents lived in relatively largefamilies: seven with a household size of seven, 28 of
between eight and 18, and one of 23. Family size wasgenerally smaller among rural interviewees than amongthose living in urban areas. Most respondents lived in
a house having between one and three rooms, withthose from urban areas having fewer rooms.Defaulters came from the most overcrowded families,T
able
1
Variablesusedto
stratify
respondents
Variable
Stratum
name
Stratum
de®nition
Number
instratum
Stageoftreatm
ent
Treatm
entcontinuing
Treatm
entcontinued
foratleast
3monthsbutless
than7months
12
Treatm
entcompleted
Treatm
entcontinued
for7monthsormore
12
Defaulted
Treatm
entstarted
butnotcompleted,andpatienthasnotreceived
treatm
entforatleast
onemonth
12
Sex
Male
18
Fem
ale
18
Areaofresidence
Rural
18
Urban
18
A. Khan et al. / Social Science & Medicine 50 (2000) 247±254 249
with 3.8 persons/room, while completers were some-
what less overcrowded, with 3.1 persons/room.Thirteen of the 36 respondents had only one mem-
ber of the household who was wage earning: only one
respondent's household had no earning member andrelied on begging and charity. Six of the respondents'households had only male earners, ®ve had both male
and female earners, and two only female earners. Theaverage number of earning members per householdwas higher among households of respondents from
urban areas than those from rural areas (respectivelytwo and 1.2 earners per household).All respondents reported very low household
incomes, with all but ®ve having incomes below the
recognised poverty level of 3745 Pakistani rupees(World Bank, 1998). The average household incomereported by respondents from rural families was 1870
rupees: the corresponding income among urbanfamilies was 3210 rupees. Respondents who had com-pleted treatment had lower both average and per
capita monthly household incomes than other groups(see Table 2).Seventeen of the respondents were literate, the lit-
eracy rate being much higher among men (12/18) andamong urban respondents (12/18). Among completers,midway patients and defaulters respectively, the aver-age number of literates in respondents' families were
4.2, 2.0 and 2.3.
Respondents' knowledge about TB
The majority of respondents and their family mem-bers were not aware of causes of TB, symptoms, treat-
ment, duration of treatment, dangers of default orprecautions to minimise the risk of infection. Onlyfour respondents acknowledged any knowledge of TB
prior to their ®rst diagnosis. All four of these had TBpatients in their family prior to their diagnosis.
Respondents' health seeking behaviour and treatmentreceived
Twenty of the respondents reported that their ®rstattempt to seek treatment was with a private prac-titioner. Of the remainder, many initially attended a
public facility because their symptoms were severe
enough to require admission to hospital for at least 1month. Overall, only 11 respondents were diagnosed as
having TB on their ®rst contact with a medical prac-titioner: ®ve of these were with private practitionersand six with public practitioners. The remainder were
diagnosed as having fever (®ve respondents), coughand fever (6), pains (5), typhoid (4), pneumonia (2) or
other diagnoses (3). On the second attempt to seektreatment, 14 respondents were diagnosed as havingTB, four by a private practitioner and 10 by a public
practitioner. Of the remaining 11 patients, two werediagnosed as having TB on their third contact with a
medical practitioner, and nine on their fourth or sub-sequent contact.
Ten respondents reported that they `did not accept'the diagnosis of TB when it was ®rst made. Twenty-sixstated that they were frustrated, disappointed or wor-
ried by the diagnosis; of the remaining ten, only foursaid they were not worried because they knew TBcould be successfully cured, the remainder believing
that the disease was God-given and had to be acceptedpassively.
Only 13 respondents were advised to attend the TBCentre by medical personnel. The remainder went on
the suggestion of a close relative or friend, or becausethere were other TB patients in the household. Threerespondents attended the TB Centre directly, without
®rst seeing any other medical personnel: this wasbecause a family member was already attending the
Centre, or because the respondent was su�ering fromchest pains and coughing up blood.Only 18 respondents claimed to be `new' TB patients
(ie had not received treatment for TB before the cur-rent course of treatment). Nine of the 12 defaulters
acknowledged that they had received treatment in thepast, as compared with six of the 12 completers. These®gures, although high, are still likely to be underesti-
mates: only three respondents reported that they hadbeen su�ering from TB for less than a year before
starting this treatment course. Twenty reported havingsu�ered TB-related symptoms for between 1 and 2years, eight for between 2 and 5 years, and ®ve for
more than 5 years. Ten respondents reported otherfamily members su�ering from TB: in total, 17 family
Table 2
Household income (Rupees)
Respondents' stage of treatment
Treatment completed Treatment continuing Defaulters
Number of respondents 12 12 12
Average monthly household income 1980 2820 2910
Average household income per capita 270 310 310
A. Khan et al. / Social Science & Medicine 50 (2000) 247±254250
members in addition to the respondents were reported
to currently have TB, nine of whom were from thefamilies of defaulters.Most medical and paramedical sta� admitted in
informal discussions with research sta� that they didnot have time to give much time or information to
patients, due to high patient loads and low sta� num-bers. In one day our survey team observed more than40 new patients being registered and 30 patients under
treatment who required check-ups etc, all being dealtwith by a single doctor. In fact, in general patients didnot report receiving useful information. None of the
respondents reported being informed about the causesof TB. 25 respondents were advised not to eat sour
food (eg buttermilk, lemons, oranges, tomatoes etc),and to avoid `cool' food (eg yoghurt). Only fourrespondents were advised to eat nutritious food such
as meat, chicken, eggs, milk or fruit.Almost all respondents reported that they were
instructed by TB facility sta� to avoid meeting or hav-
ing close contact with other people, essentially recom-mending `voluntary social isolation'; to cover the
mouth when coughing; and to use separate eatingutensils. Moreover, no patient reported being told thatif the regime was correctly followed, he or she would
not be infectious after at most 8 weeks, and in mostcases much sooner. This suggests that health workersshare the same cultural beliefs about TB as the rest of
Pakistani society.Almost all respondents (32/36) reported that they
were informed about the duration of treatment,although none said they were told why such a longregime was necessary, nor the dangers of default.
Only six respondents thought the treatment o�eredat the TBC/hospital was satisfactory, and even these 6
complained of long waiting hours, unsympathetic atti-tude of the health personnel, and di�culty in seeingmedical and paramedical sta�. The remaining 30
patients were unhappy with the treatment o�ered.Many respondents (16/36) complained about unavail-ability of drugs at the facility. 2 respondents lived
within 15 min walk of a facility: the remaining 34respondents spent on average 30 rupees, 3 h travel,
and 3±4 h at the facility to collect drugs. They all com-plained of the fatigue and weakness induced by eachvisit to the facility. Many (20/36) were so apprehensive
that in addition their visit would be fruitless (sincedrugs were often not available) that they delayed oravoided the trip. They had many doubts about the use-
fulness of the whole treatment process.The three treatment facilities from which patients
were enrolled did not di�er substantially in the qualityof services provided. At every facility, provision fortreatment left much to be desired: drug supply was
irregular; access to treatment was expensive (forexample patients were expected to pay for unnecessary
X-rays); little personal attention was given by sta�;considerable travel was required to reach the facility;
no ¯exibility was available in allocating appointmentsto patients; etc.
E�ect of TB on respondents' personal lives
Thirty respondents stated that their relatives andfriends knew that they had TB, and that they them-selves had told them. On further enquiry, it transpired
that normally only very few relatives, and close ones atthat, had been informed; and this could not have beenavoided due to the closeness of Pakistan family life.The remaining 6 respondents were all young women
recently married or of marriageable age. They saidthey had not disclosed their infection for fear of socialaversion: they commented `they would hate me' and
`people would start avoiding me'.When asked whether TB patients generally avoid
informing others about their disease, 25 respondents
agreed, believing that the reason for this reticence wasfear of seclusion and social boycott.Almost half the respondents (17/36) stated they ex-
perienced no signi®cant change in the behaviour of
people towards them. The remainder felt that peopletried to avoid close contact and had reduced meetingthem. Some of the responses (`they try to avoid me';
`people do not even want to talk with me') reveal thesocial stigma associated with TB in Pakistan.However, when asked about precautions their close
relatives took to avoid infection, 15 respondentsadmitted being secluded from the rest of the family,reporting that the family reduced socialising with them
and insisted they use separate eating utensils. Fiverespondents stated they were made to sleep in a separ-ate room. The remaining respondents maintained thatno precautions were taken by other members of the
family, and social interaction continued as normal.A history of employment immediately prior to get-
ting TB was reported by 14 men (including two stu-
dents) and 16 women (including three students). By thetime of the study only six men (including one student)and seven women (all three female students had
stopped studying) were still working. Two men whowere still working had changed job to one that wasless strenuous and less skilled. Five of the ten womenwhose reported work was housework had stopped this
work. Only two completers stopped work, whereasseven midway patients and six defaulters stopped.Once respondents' families accepted the diagnosis,
they were supportive of the respondent. Fourteenrespondents reported that they were either ac-companied to the facility or someone went in their
place to collect medicines. This of course increases theburden on the family, and six respondents reportedthat the family had taken out loans to meet the ex-
A. Khan et al. / Social Science & Medicine 50 (2000) 247±254 251
penses. Only one of these loans was taken for a femalerespondent.
Twenty-seven respondents reported that they couldnot always take their medication themselves, butneeded help when they felt too weak. Six respondents
went further to say that when they refused to taketheir medication the helper tried to convince them.
Reasons for respondents' compliance and non-compliance with treatment
There were no obvious di�erences between defaultersand those who completed treatment. Literacy and edu-cational levels were very similar between the twogroups. A history of previous default does not appear
to be associated with increased risk of default from thecurrent treatment episode.However, defaulters had the highest rate (eight out
of 12) of respondents not satis®ed with treatmento�ered. Four of these stopped treatment because theycould not see any improvement; the remaining 4 actu-
ally felt worse (probably due to side-e�ects of drugs)and gave this as their reason for stopping treatmentprematurely. Most respondents felt increasingly demor-
alised as initially strong family support becameweaker.All defaulters identi®ed ®nancial constraints as the
major factor leading to discontinuation of treatment.
All defaulters also complained of the time unnecess-arily spent waiting at the TB facility. Rural defaulterscomplained of the costs of travel and physical fatigue
involved in visits to the TB facility as contributing todefault. Rural women were particularly disadvantagedby problems associated with travel: all 9 rural women
respondents named duration/cost of travel and beingunable to travel alone as factors contributing todefault.Four defaulters did not appreciate that they were
regarded as such by the health service. For thesepeople, irregular collection of medication (leading tobreaks in taking the medication) was assumed to be
acceptable and normal behaviour. As far as they wereconcerned, they were still continuing treatment. Themajor reason given to explain irregular collection of
medication was dissatisfaction with treatment due toside e�ects caused by medication.Five defaulters reported that they stopped treatment
because they believed they were cured.
Discussion
This was a sociological study involving in-depth
interviews. For this reason the sample size of our in-vestigations was relatively small, with only 36 respon-dents. It has provided useful information to allow
providers to ascertain tuberculosis treatment delivery
methods that might be more appropriate to patientsthan those currently available in Pakistan. We should,however, emphasise that the sample is not representa-
tive of tuberculosis patients in general, because of thedi�culties we encountered in tracing registeredpatients. Many patients gave false addresses to the
tuberculosis centre, and it became clear that therewould be ethical problems in approaching patients at
their homes without their prior consent. Even whenprior consent was obtained, it became clear during thecourse of data collection that some patients had no
intention of co-operating with our study, since theygave false addresses. A similar situation has beenobserved in India (Juvekar et al., 1995). It may be that
patients unwilling to co-operate are di�erent in kindfrom our respondents; if so, some selection bias will
have ensued. It may also be worth noting that forsome patients some recall bias may have occurred,since some information was collected several months
after the occurrences on which it was based.All patients included in our study were poor; many
were extremely poor. This clearly has major impli-cations regarding the cost of access to tuberculosistreatment. If patients are very poor they are unlikely
to be able to a�ord to travel regularly to the tubercu-losis treatment centre, and this may seriously a�ecttheir ability to comply with treatment, and particularly
with DOTS schemes that potentially place considerable®nancial burdens on patients.
The majority of patients had little or no knowledgeabout tuberculosis, and unfortunately the perceivedamount of health education received at the tuberculosis
treatment centre was very limited. This is likely tohave two causes: sta� at the treatment centres areheavily overloaded, and hence have little time for what
they perceive to be the less important parts of treat-ment; and a general culture of passive patients receiv-
ing instructions rather than being actively involved intheir treatment. Moreover, the health education thatwas delivered appears to have been incorrect. While
some of the advice given was based on Asian humore-tic systems (Pool, 1987) (eg advice on `cool' and `sour'foods), other advice had no basis and led only to a
further feeling of isolation amongst patients.An essential part of tuberculosis treatment is advice
regarding the long duration of treatment. Such advicewas given to patients, but the reasons for such a longregime were not. Similarly, no advice was given on the
dangers of default. Such advice is crucial in encoura-ging patients to continue with treatment for the full
duration of the regime. It is clear that Pakistanidoctors need training on the importance of health edu-cation to encourage treatment completion.
Observations similar to those found on fear of socialaversion have been made in other studies (Hudelson,
A. Khan et al. / Social Science & Medicine 50 (2000) 247±254252
1996). To young unmarried Pakistani women, having
or having had TB is perceived as a threat to the possi-bility of marriage: this perception is not unreasonablesince arranged marriages are often at risk if the poten-
tial in-laws learn of a history of TB (or other seriousillness) in the woman or even among the woman'sfamily (Liefooghe et al., 1997). Most patients are very
unwilling to disclose that they have tuberculosis toanyone other than the immediate family; and it is
likely that if it were possible, they would not even dis-close their illness to them. This appears to be closelylinked with a perception that tuberculosis is incurable,
which in the current circumstances in Pakistan is prob-ably a reasonable perception. Unfortunately this
unwillingness to disclose their illness makes treatmentmuch less likely to be successful. Thus we see a viciouscircle being perpetuated. Until patients have con®dence
in TB treatment they will continue to be unwilling todisclose their infection; and such lack of disclosureincreases the likelihood that the treatment will not be
successful, for reasons linked with problems of accessto treatment, and subsequent default. For TB pro-
grammes to be successful in Pakistan, it is essentialthat this circle is broken; and this can only be donethrough a combination of providing a good quality of
TB care, and health education to improve the popula-tion's understanding that TB can be cured.
Our ®nding that compliant patients had a somewhatlower average and per capita income than the othergroups was unexpected, but it should be remembered
that this is a gradation within a very poor group;nearly all respondents fell below the recognised povertylevel. We found that patients often stop working after
TB has been diagnosed, a ®nding supporting a reportfrom a study of Mumbai slums that the wage earning
capacity of both men and women was a�ected by TB(Nair et al., 1997). Moreover, respondents' familieswere supportive of the respondent, a ®nding also
reported by others (Liefooghe et al., 1995). It is clearthat a diagnosis of tuberculosis places a burden notonly on the patient him or her self, but also on the
patient's family. In situations where resources areavailable, it is clear that support needs to be given to
the family as well as the patient. Unfortunately, in thePakistani situation, resources are extremely limited. Itmay, however, be possible to organise family support
groups, through which information may be passedfrom family to family, and encouragement be given to
assist the patient to complete treatment. Such supportgroups will only be feasible when the community atti-tude to TB changes, when the need for con®dentiality
should be less crucial.One of the potential uses of the study results was to
identify potential predictors of default. For a predictor
of default in TB to be useful, it must have a high sensi-tivity, and such high sensitivity predictors should be
evident even from small studies such as this one. The
results we present here show no clear pattern to allowus to distinguish between those who are continuingwith treatment and those who have defaulted. Our
®ndings are consistent with those of others; as yet, noworkers in the ®eld of tuberculosis have been able toidentify reliable and persistent predictors of default
(Barnhoorn and Adriaanse, 1992; Reed et al., 1990).However, some pointers are apparent. Patients who
report that they are unsatis®ed with treatment may bemore likely to default; and if patients are asked abouttheir level of satisfaction at every contact with the TB
service, such potential di�culties may be identi®ed. Inaddition, it seems that some defaulters do not appreci-ate that they are defaulters, and this suggests that
tuberculosis treatment providers need to educatepatients as to what is expected of them, and make
greater attempts to contact them when appointmentsare missed. We intend to perform a second studynested within our RCT to further investigate factors
that may lead to default.Our results con®rm that it is that more di�cult for
women to receive adequate treatment than for men inPakistan. There appear to be two reasons for this:restrictions to women's movement, caused by the
closed society in Pakistan; and an unwillingness to payfor treatment for women. While there appears to bestigma associated with tuberculosis both for men and
women, this is considerably greater for women. Youngunmarried women are extremely unwilling to announce
the fact that they have tuberculosis, as it is their belief(which is almost certainly correct) that this will preju-dice their chances of getting married. Even younger
women who are already married are concerned toadmit that they have tuberculosis, because they feelthat there is considerable risk that they may be
expelled from their marriage home and returned totheir natural family.
It is clear from this study that many of the problemsencountered by patients trying to obtain good tubercu-losis treatment are not due to factors related to
patients, but are caused by poor service provision.Roughly half of patients with tuberculosis-like symp-
toms initially visited a private practitioner. Even fromsuch a small survey as this one it is clear that privatepractitioners in general are providing a poor TB treat-
ment service in Pakistan, as is also the case in India(Uplekar and Shepard, 1991). Results from studies inPakistan (Jameel et al., 1996; Qari et al., 1997) and
Nepal (as yet unpublished) suggest that the treatmentthey will receive from private medical practitioners is
likely to be of poorer quality than that from the publicservices, but the advantage of con®dentiality outweighsthis disadvantage of private medical practitioners' ser-
vices.Even when patients arrived at the public service the
A. Khan et al. / Social Science & Medicine 50 (2000) 247±254 253
treatment they received was unsatisfactory. Patientswere subjected to long waiting times, given little edu-
cation about tuberculosis, and often not provided withthe drugs they required for treatment.So long as health services do not provide drugs,
patients who have limited resources will ®nd it di�cultto comply with treatment. The e�ect both to individ-uals and to Pakistani society of such poor treatment
provision is catastrophic. Individuals do not receivethe treatment they require, and may either becomeincreasingly unwell, a situation leading eventually to
death, or may become chronic su�erers of tuberculosis,with all the consequences this has for the public healthof society.
Conclusion
This study has proved useful in the determination offactors that must be taken into account while design-ing DOTS interventions which are acceptable and feas-
ible, and therefore likely to be e�ective, in Pakistan.Most important of these are factors related to accessto services: cost to patients, both in money and time;distance of services from patients' homes; and the
necessity of accompanying `chaperones' for womenpatients. TB service providers should be aware thatmany potential causes of default remain their, rather
than their patients', responsibility.
Acknowledgements
We would like to thank the patients involved in thisstudy for their cooperation, sta� of the three health
centres involved for their assistance in facilitatingaccess to patients, and to employees of Dataline fordata collection. This study was funded by the
Department for International Development of theUnited Kingdom. However, the Department forInternational Development can accept no responsibility
for any information provided or views expressed.There are no con¯icts of interest.
References
Addington, W.W., 1979. Patient compliance: the most serious
remaining problem in the control of tuberculosis in the
United States. Chest 76, 741±743.
Barnhoorn, F., Adriaanse, H., 1992. In search of factors re-
sponsible for non compliance among tuberculosis patients
in Wardha District, India. Social Science and Medicine 34,
291±306.
De Jonghe, E., Murray, C.J.L., Chum, H.J., Nyangulu, D.S.,
Salomao, A., Styblo, K., 1994. Cost e�ectiveness of che-
motherapy for sputum smear positive pulmonary tubercu-
losis in Malawi, Mozambique and Tanzania. International
Journal of Health Planning and Management 9, 151±181.
Government of Pakistan, 1994. Eighth Five Year Plan (1993±
98). Islamabad.
Hudelson, P., 1996. Gender di�erential in tuberculosis: the
role of socio-economic and cultural factors. Tubercle and
Lung Disease 77, 391±400.
Jameel, M., Ahmad, M., Khan, M.A., Ishaq, S., Bokhari,
I.A., 1996. Satisfaction of pulmonary tuberculosis patients
attending private medical practitioners and government fa-
cilities in Gowalmandi Lahore. Pakistan Journal of Health
33, 34±38.
Juvekar, S.K., Morankar, S.N., Dalal, D.B., Rangan, S.G.,
Khanvilkar, S.S., Vadair, A.S., Uplekar, M.W.,
Deshpande, A., 1995. Social and operational determinants
of patient behaviour in lung tuberculosis. Indian Journal
of Tuberculosis 42, 87.
Kochi, A., 1991. The global tuberculosis situation and the
new control strategy of the World Health Organisation.
Tubercle 72, 1±6.
Liefooghe, R., Michiels, N., Habib, S., Moran, M.B., De
Muynck, A., 1995. Perception and social consequences of
tuberculosis: a focus group study of tuberculosis patients
in Sialkot, Pakistan. Social Science and Medicine 141,
1685±1692.
Liefooghe, R.G., Moran, M.B., Habib, S., De Muynck, A.O.,
1997. Treatment adherence of tuberculosis patients in
Bethonia Hospital, Sialkot. Journal of the College of
Physicians and Surgeons, Pakistan 7, 140±144.
Monterrosa, A., 1991. Epidemiology and control of tubercu-
losis in the world. Wellcome Trust, London.
Murray, C.J.L., 1994. Resource allocation priorities: value for
money in tuberculosis control. In: Porter, J., MacAdams,
K. (Eds.), Tuberculosis: Back to the Future. John Wiley &
Sons, UK.
Murray, C.J.L., Styblo, K., Rouillon, A., 1991. Tuberculosis.
In: Jamison, D.T., Mosley, H. (Eds.), Disease Control
Priorities in Developing Countries. Oxford University
Press for World Bank, New York.
Nair, D.M., George, A., Chacko, K.T., 1997. Tuberculosis in
Bombay: new insights for poor urban patients. Health
Policy and Planning 12, 77±85.
Pool, R., 1987. Hot and cold as an explanatory model: the
example of Bharuch district in Gujarat, India. Social
Science and Medicine 25, 389±399.
Qari, M.A., Ahmad, M., Ishaq, S., Ali, A., 1997. Pulmonary
tuberculosis case management by private medical practito-
ners in Gawalmandi Lahore. Pakistan Journal of Health
34, 53±56.
Reed, J.B., McCausland, R., Elwood, J.M., 1990. Default in
the outpatient treatment of tuberculosis in two hospitals in
Northern India. Journal of Epidemiology and Community
Health 44, 20±23.
Sloan, J.P., Sloan, M.C., 1981. An assessment of default and
non-compliance in tuberculosis control in Pakistan.
Transactions of the Royal Society of Tropical Medicine
and Hygiene 75, 717±718.
Uplekar, M.W., Shepard, D.S., 1991. Treatment of tuberculo-
sis by private general practitioners in India. Tubercle 72,
284±290.
World Bank, 1998. Poverty Assessment Report. Islamabad.
A. Khan et al. / Social Science & Medicine 50 (2000) 247±254254