Factors at first diagnosis of tuberculosis associated with compliance with the Directly Observed Therapy ( DOT ) in the Limpopo Province , South Africa

The aim of this study is to compare and contrast health beliefs, demographic and socio-economic variables, causative beliefs, knowledge, health-seeking behaviour and health provider-patient interaction of compliant and non-compliant tuberculosis patients. The sample included 219 consecutive new sputum-smear and/or culture positive pulmonary tuber­ culosis patients registered between October 1999 and March 2000 in three hospitals in the Limpopo Province of South Africa. The patients were 144 (65.8%) men and 75 (34.2%) women in the age range of 18 to 79 years (M age 35.9 yr., SD= 12.6). The consultation at first diagnosis was observed and tape-recorded. Thereafter an interview was conducted and a questionnaire was face-to-face administered with the patient including knowledge, causative beliefs, health seek­ ing, and Health Belief Model items. Discriminant analysis between compliant and non-compliant groups after six months follow-up showed that the quality of the health practitioner-patient interaction and causative belief were associated with compliance behaviour whereas knowledge, onset of TB, sociodemographic variables, health care seeking, and health beliefs were not associated.


Introduction
South Africa is burdened by one of the worst tuberculosis epidemics in the world, with disease rates more than double those observed in other developing countries and up to 60 times higher than those currently seen in the USA or Western Europe.The Medical Research Council estimated that the country had an estimated 180 507 cases (55% reported) in 1997, or 419 per 100 000 of the total population; of these, 32,8% (73 679 cases) were probably infected with HIV (Fourie, 2000: 4f.).Estimates by the MRC National Tuberculosis Pro gramme indicate that current trends in the epidemic will con tinue unless effective control is achieved, resulting in 3,5 million new cases of tuberculosis over the next decade and at least 90 000 patients dying.On the other hand, significant reductions in transmission of HIV infection together with ef fective tuberculosis control would mean a turn-around in the tuberculosis epidemic by the year 2003 (Fourie, 2000: 5f.).In comparison with the MRC estimates based on epidemiologi cal modelling as mentioned above, actual registration reports to the National Department of Health indicated smear posi tive rates per province (per 100 000 of the total population) of 285 for the Western Cape, 300 for the Eastern Cape and 328 for the Northern Cape.All other provinces had rates below 200.The overall rate for South Africa was 163 for smear posi tive and 310 for all pulmonary tuberculosis cases (Fourie, 2000: 7f.).Tuberculosis was declared a top health priority by the De partment of Health in November 1996 and National Health Minister Zuma committed her Department to implementing a new control programme based on the DOT strategy of the World Health Organisation.The pace and extent of imple mentation of the programme is, however, slow in most prov inces.Since 1996, a system of case registration based on strict criteria for case definition was implemented in South Africa.These registrations, based on standardised criteria, are now beginning to present a clearer picture of disease rates in the country than what was available before.Some progress is be ing made in certain provinces in South Africa (Wilkinson & Davies, 1997: 700) Mpumalanga (despite relatively high HIV rates) and the Western Cape are already showing dramatic improvements in cure rates, because of disciplined implemen tation of the DOT strategy of the WHO in these provinces.
Other provinces are at various stages of implementation of the process such as the Limpopo Province (Fourie, 2000: 10).Factors identified for the improvement of the DOT strategy from research are implementation of the DOT right from the beginning of diagnosis, culturally determined beliefs about and knowledge of tuberculosis, adherence of TB treatment (Rubel & Garro, 1992: 626), transportation time, the sex of the patient, patient information and the quality of communi cation between patients and health workers (Com olet, 55 Curationis August 2002Rakotomalala & Rajaoarioa, 1998: 891), substance abuse, emotional disturbance, homelessness, lack of transportation, behavioural problems, dissatisfaction with clinic scheduling, forgetfulness, mental retardation, lack of family or social sup port, migrant status, illiteracy, unemployment, and low in come (Sumartojo, 1993(Sumartojo, : 1311)).Glatthaar and Berends (1995: 179) have shown how DOT with a community approach reached a 90% adherence rate in an urban (Cape Town) area.Jaramillo (1999: 71) reviews that low levels of self-efficacy, poor information about TB and its means of control and the stigma attached to the disease (with the social discrimination it entails) contribute to some extent to poor adherence to treat ment.Furthermore, he states that material support (food, money, fees for transport, etc.) and personal/family income has been reported as a predictor of the outcome of TB control programmes.Bamhoom and Adriaanse (1992: 291) identi fied that the presence of social support (in particular the pres ence of a supportive person, who establishes routine in tak ing pills and in making control visits) and satisfaction with health-care providers contributed positively to compliance.Westaway and Wolmarans (1994: 447) found among black urban South African TB patients that case-holding (compli ance for both patient and system) involves complex behav iours that depend upon symptom recognition and evaluation, cultural and social influences and enabling factors such as time, money, skills and appropriate/accessible health serv ices.Rideout and Menzies (1994: 3) stated that tuberculosis control programmes must be tailored to take into account the importance of cultural factors in promoting compliance with therapy.In many cultures, the largely unremarked social stigma of tuberculosis contributes to abandonment of treat ment and lengthy delays in seeking professional care (Rubel & Garro, 1992: 626).

Statem ent of the problem
The DOT strategy has been shown to be more cost effective than conventionally delivered treatment for the treatment of new cases of tuberculosis in adults in South Africa (Floyd, Wilkinson & Gilks, 1997: 1407f.;Westaway, Conradie, & Remmers, 1991: 447).However, the implementation of the DOT strategy in the Limpopo Province has been slow.Sev eral studies have identified factors for the improvement of DOT strategy (Glatthaar & Barends, 1995: 179;Gamer, 1998Gamer, : 1326;;Rubel & Garro, 1992: 626;Comolet et al., 1998: 892;Sumartojo, 1993: 131 Iff.).Other studies have suggested ad ditional factors to be considered (Jaram illo, 1999: 71;Barnhoorn & Adriaanse, 1994: 2 9 If.).These studies were urban-based and did not focus on rural black South Africans.Westaway and Wolmarans (1994: 447) studied black South African TB patients but this study was also urban-based.There is paucity of information on DOT strategy in South Africa.This information has not resulted in significant improvement of the pace, extent of implementation and effectiveness of DOT programme in the Limpopo Province.This study aims to in vestigate the predisposing, reinforcing and enabling factors, which can improve the delivery of DOT programme in three predominantly rural black communities in the Limpopo Prov ince of South Africa.

Theorectical models
Directly observed therapy is predicated on the belief that by directly observing the patient consume all required medica tions, a full treatment regimen will be ensured, thereby re ducing the risk of treatment failure (Nazar-Stewart & Nolan, 1992: 58).The health belief model (HBM) has proven to be a useful framework for examining sick role behaviours such as com pliance.It assumes that people comply with regimes under a very specific set of conditions.Patients must possess some minimal level of health knowledge and motivation towards health.They must see themselves as vulnerable to the illness and believe that the illness is of a serious nature.Addition ally, they must be convinced that the treatment can be effica cious, i.e. that it is actually possible to obtain control over the disease and that the cost of such control is not too high in view of the benefits.Another factor included in the model is the predisposition, or motivation, of people to engage in health-related practices (Bamhoom & Adriaanse, 1992: 300).
Effective care for patients also requires understanding of one's ethnic identity and related conception of illness.Kleinman (1980: 80f.) referred to developing an openness to each pa tient's own 'explanatory model' of illness.Because of their primary contact with the patient, health care providers can strongly affect the patient's commitment to a correct regi men, particularly by means of clear communication about the regimen.

Aims and objectives
The aim of this study is to compare and contrast health be liefs, demographic and socio-economic variables, causative beliefs, knowledge, health-seeking behaviour, and health pro vider-patient interaction of compliant and non-compliant tu berculosis patients.If the model distinguishes well between both groups, it would be important to identify what dimen sions have the greater impact on patient compliance and which ones are appropriate to serve as a basis for planning educa tional and health promotive interventions for the improve ment of the DOT strategy in the three main cultural groups and three regions in the Limpopo Province of South Africa.

Methods Design
The design of the study was a prospective case control study divided into (1) assessment of first diagnosed pulmonary tu berculosis patients, and (2) follow-up of the diagnosed tuber culosis patients.Here only the follow-up statistics are used in the analysis.The detailed results of the follow-up study are reported elsewhere (Peltzer, 2001: 191).

Sample
The

Research instrument
(1) A 46-item scoring key for the health carer patient con sultation (Boesch, 1988: 253ff.).Items include forexample: "How does the doctor start the consultation?14, "Symptom question concerning cause11, "Explaining of cause by doctor", "Personal question related to symp tom or other health question", and "Explanation of treatment and medicine by doctor11.(2) A recall interview of the patient on the consultation (Boesch, 1988: 250 Fifteen items on demographic and socio-economic data. swered by yes or no) (.62).(cf.Steen & Mazonde, 1999: 163) (10) Three items on the curability of health care agents (rated from l=agree to 3=disagree  from the files and asked for formal consent to participate in the study.Then the patient was accompanied by one of the researchers to the consultation.The consultation was observed by one of the researchers (P Seoka) and two trained research assistants and tape-recorded.Thereafter an interview was conducted and a questionnaire was face-to-face administered with the patient.This included a recall interview on the health provider-patient interaction.
The questionnaires were translated and back translated by bilingual experts in the major languages used in the study according to scientific standard procedures.The schedule was field tested before the survey and modified where necessary.

Data analysis
The health professional-patient consultation was analysed  (Boesch, 1988: 253ff.).The taperecorded patient-doctor consultation and recall interview of the patient on the consultation were transcribed from the ver nacular language to English according to scientific standard procedures.The messages from the health professional ^m e s sages given) and the messages recalled by the patient were content analysed using the following categories: diagnosis, treatment course, treatment duration, support available, prog nosis, treatment adherence, results of non-compliance (see appendix II) (Boesch, 1988: 250).Causative beliefs of tuber culosis from the Problem Portrait Technique were content and 92 (42%) were single.Almost half (47%) had (some) secondary education and 28 (12.8%) were illiterate.Ninetythree (42.5%) did not have any monthly income and the ma jority were either having unskilled or semi-skilled occupa tions.
Table 1 indicates the sociodemographic characteristics of first diagnosed tuberculosis patients in all three research sites.
Typically the tuberculosis patient is a middle-aged male, mar ried or cohabits, lives in the rural area, comes from a nuclear family, has a semi-skilled occupation, has secondary educa- analysed, coded and also analysed statistically (see Table 4) (MacLachlan, 1997: 84-9).Further, descriptive, Chi-square and discriminant statistical analyses were applied using the SPSS (version 10.0).

Results
The sample included all (N=219) first diagnosed pulmonary tuberculosis patients.Most patients (85.5%) came from a vil lage and a nuclear family type (63.5%).About half of the participants were married or living with a partner (47.5%) tion, lives in a brick house with corrugated iron, uses fire wood as fuel, has no income or is unemployed, belongs to African traditional religion, and is either the son or the head himself of the family.
In order to identify demographic and socio-economic vari ables that discriminated between the compliant and non-compliant groups a discriminant analysis was performed for a

Curationis August 2002
total of 136 patients (81 compliant and 55 non-compliant), which had been successfully followed-up.
From 219 tuberculosis patients assessed at first diagnosis 136 were successfully followed-up after 6 months, 81 (59.6%) had been compliant and 55 (40.4%) had been non-compliant.The compliance rate differed by hospital site from 70.8% to 41.9%.A large group of patients (82) could not be followed-up due to death, had moved out of the province, and could not been traced.
the Problem Portrait Technique.
The compliant group attributed more germs to tuberculosis than the non-compliant group did, whereas the non-compliant group saw smoking tobacco as more responsible for tu berculosis than the compliant group did.The most commonly mentioned causative agents for tuberculosis are believed to be: germs, tobacco smoking, dust, alcohol, dirty environment, and infected by spouse (sex) partner.
Table 7 : Knowledge of patients on when to stop treatm ent (rated from 1 = strongly agree to 5 = stron gly disagree)   None of the variables had significant differences between com pliant and non-compliant groups.Table 3 indicates the onset of tuberculosis symptoms as expe rienced by the patients.
The non-compliant group had a slightly longer onset of tu berculosis symptoms than the compliant group, which was almost significantly different.
Table 4 indicates the causative beliefs of tuberculosis using Prior to first diagnosis a number of participants from both compliant and non-compliant groups had used alternative healing systems.
Table 6 indicates the patients1 knowledge about the transmis sion of tuberculosis.

Curationis August 2002
Knowledge on transmission of tuberculosis can generally be considered high regarding sputum, airborne and contact with someone.Falsely sharing toilet, bath, towels or clothes, ciga rettes, food or drink was seen by the majority as a route of transmission.There was no significant difference between the compliant and non-compliant group.Table 7 indicates the awareness of patients on when to stop the treatment Most patients disagreed with stopping the medication 'when feeling well' and most agreed with after a period of 6 months or more.There were no significant differences between com pliant and non-compliant group.
fied properly, whereas 34.2% (out of 54 cases) of the noncompliant group was classified correctly.
The health practitioner-patient interaction score was signifi cantly higher among compliant than among non-compliant patients.
Messages given at first diagnosis of TB were in both groups about 4.5 per session and messages recalled were between 3 to 3.5 messages.Examples for messages given and recalled are given in Appendix II.

Discussion
The study identified the following factors to be associated with compliance with the DOT strategy, which could serve as

62) .195
Table 8 indicates substance use and family and community history of tuberculosis of the participants.Both past 6 months alcohol and tobacco use seemed with more than 30% high in both compliant and non-compliant tuber culosis patients.Every fourth (about 25%) in both groups had a family member who had tuberculosis.There were no sig nificant differences between compliant and non-compliant patients.
Table 9 indicates the components of the Health Belief Model regarding tuberculosis.
Both compliant and non-compliant patients rated tuberculo sis as a severe disease, saw strong benefits in taking medica tion, felt somewhat susceptible towards tuberculosis, and ap peared to have not much motivation for treatment.There were no significant differences between the compliant and noncompliant groups.Even the analysis of individual items of the 21-item Health Belief Model questionnaire did not show any significant difference on any item between the compliant and non-compliant group.
Table 10 indicates the analysis of the health practitioner-pa tient interaction at first diagnosis.
The overall percentage of cases classified correctly was 61.1%.cases out of 81 (57.9%) of the compliant cases were identi-a basis for planning educational and health promotive inter ventions for the improvement of the DOT strategy in the Limpopo Province: Case detection; Lesser use of herbs, warm fluid/water, and visiting of traditional and faith healer In this sample diagnosis of TB was made about 2.7 months after the onset of symptoms.Among a rural sample in Bot swana a median delay period of 12 weeks was found for the anti-TB treatment in modem medicine (Steen & Mazonde, 1999: 165).
In this sample more than 27% had used herbs, more than 25% had visited a traditional healer and 43% over-the-coun ter drugs prior to TB diagnosis.Steen and Mazonde (1999: 166) also found in Botswana that 52% of the subjects tried one or more alternative treatments during the delay period.
After modem treatment had started in this sample the utili zation of alternative treatment modalities dropped, whereas in Botswana 47% of the subjects visited or planned to visit a traditional or faith healer (Steen & Mazonde, 1999: 169).Brouwer, Boeree, Kager, Varkevisser and Harries (1998: 232) found that 37% of TB patients visited a traditional healer before seeking regular medical care in Malawi.In Ethiopia th era p eu tic preferen ce hinges on the u tiliza tio n o f ethnobotanical remedies (Vechiato, 1997: 185f.).Liefooghe, Baliddawa, Kipruto, Vermeire and De Munynck (1997: 812) reported that the delay in diagnosis of TB, was partly, a result 62 Curationis August 2002 of health-care-seeking practices held by PWT.

D em ographic an d socio-econ om ic fa c to rs w ere n ot id en ti fie d as discrim inating fa c to rs betw een com plian t an d noncom pliant groups
Liam, Lim, Wong and Tang (1999: 300) also found that com pliance with treatment was not affected by age, sex, ethnic group, educational level, occupation, extent of knowledge, tuberculosis symptoms, hospitalisation for tuberculosis.De mographic factors such as age, sex, race, ethnicity, occupa tion, income, and education are often included in successful treatment of tuberculosis, however, such variables are incon sisten t or u n reliab le pred icto rs o f p atien t adherence (Sumartojo, 1993(Sumartojo, : 1311)).Nuwaha (1997: 690) found in TB patients in Uganda that gender and age was not associated with compliance.Werf, Dade and Van der Mark (1990: 249) found in TB patients in rural Ghana that lower educational level, female gender and younger age was associated with compliance.

C ausative beliefs (germ s, sm og/sm oke, an d tobacco sm ok ing)
Bamhoom and Adriaanse (1992: 301) only found witchcraft associated with compliance.Steen and Mazonde (1999: 169) found similar causative beliefs in Botswana such as heavy work, dusty, smoking, from other TB patients, drinking, germs and witchcraft.However, in this sample heredity was also mentioned, likewise in a community sample in the Philip pines (Nichter, 1994: 649).Heredity and hard work were of-lungs, predisposing people to TB (Nichter, 1994: 650).DeVilliers (1991: 70) felt that for Xhosa TB patients witch craft was important.In a study in Honduras 57% of the TB patients studied believed that they would contract the disease by using the eating utensils of someone afflicted by TB (Mata, 1985: 57).Dalai and Singh (1992: 193) found among hospi talised male TB patients in India that similar causative be liefs as in this study, namely inadequate diet, strenuous work routine, unhygienic practices, and addiction.Scientifically founded beliefs are that TB is not transmitted by sharing meals or cutlery, by kisses, hugs, or sexual relationships.

Knowledge
In this study both compliant and non-compliant groups seemed to have basic knowledge about tuberculosis in terms of causes, transmission modes and treatment.However, knowledge did not discriminate between compliant and non-compliant group.Other studies have demonstrated that knowledge about an illness, its origins, its dangers, or its treatment in itself does not necessarily lead to improved compliance (Sbarbaro, 1990: 325f.).Enarson, Grosset, Mwinga, Hershfield, O 'Brien, Cole and Reichman. (1995: 809) feel that the quality of the patient and provider relationship has a strong influence on treatment ad herence.Chaulet (1987: 21) suggests that for improving com pliance in TB patients: a personal interview with the patient  , Lim, Wong & Tang, 1999: 300).Nichter (1994: 650) found among Filipina TB patients that cigarette smoking was associated with TB not only through the perception that smoke was harmful for the lungs, but because it was observed that smoking sometimes reduced appetite.This study found other causative beliefs of lower importance such as hard work and witchcraft.Overwork and exposure to the elements (as well as excessive sexuality) were perceived to weaken the body among Filipinas.Overwork was a risk factor for pulmonya, an illness thought to weaken the lasting at least 20 min, to identify his social, occupational, and family problems, as well as his perception of the disease and its treatment, and the development of personal contact between the patient, the physician, and the nurses through out treatment.In this study the interview with the physician did not normally include the social history of the patient, oc cupational and family problems as well as the perception of the disease as it was rated from the health practitioner-pa tient interaction score.Health care providers do need train ing on the importance of health education to encourage treat ment completion (see also Khan, Walley, Newell & Naghma, 63 Curationis August 2002August 2000: 247): 247).This study found that messages on the instructions by the physician recalled were not associated with compliance.How ever, the patient recalled more than half of the messages given soon after the consultation.Sbarbaro (1990: 325) reviews that roughly half the statements made to a patient will be forgot ten within five minutes.

Patient characteristics
In this sample about 25% indicated that they had a family member and 28% a community member who had had TB.Westaway and Wolmarans (1994: 447-9) found among urban South Africans that 48% of tuberculosis patients had a family that must be taken into account while designing DOT inter ventions, which are acceptable and feasible, and therefore likely to be effective in South Africa.Most important of these factors are: quality of health provider -patient relationship and causative belief.They have shown that powerlessness can be thought of as a broad risk factor for improving the delivery of DOT programme; and empowerment or control over one's destiny, as an important strategy.These findings need to be characterized as preliminary until the concepts can be tested directly in the Limpopo Province of South Africa.If these findings hold, health promotion practitioners should begin -It takes six months be cured when you take treatment correctly.
-You have to come back after two so that we check how treatment works and when you are better we give further treatment which you take for four months and then we check you if you are cured.
-We are going to look for someone in the community who will supervise you while taking treatment.
-The TB officer will accompany home when you are discharged.
-Every time when you take medication you must tick on the green card.
-If you take your treatment regularly you will completely get cured.
Diagnosis -TB is curable if you take medication properly.
-You are going to take treatment for six months.
-You have to come back after two so that we check how treatment works and when you are better we give further treatment which you take for four months and then we check you if you are cured.
-If you don't take treatment you will be admitted here again and the disease will be worse.-You have to take treatment for six months.
-You must be supervised while taking treatment, to be sure that you're taking it.
-The TB officer will drive you home and explain, other things you have to know about TB treatment.
-You have to come back after two so that we check how treatment works and when you are better we give further treatment which you take for four months and then we check you if you are cured.
-Some patients do not finish their treatment, after two months or so they stop thinking they are cured, don't follow suit, because illness come back and you'll infect those around you. Diagnosis sample included 219 consecutive new sputum-smear and/ or culture positive pulmonary tuberculosis patients registered between October 1999 and March 2000 in three hospitals (74 56 Curationis August 2002 in Tsilidzini, 80 in Letaba and 65 in Pietersburg-Mankweng-(a) formal informed consent, (b) age 18 years or older, and Groothoek hospitals).Criteria for inclusion in our study were: (c) no previous treatment history.All patients received super- Appendix II: Exam ples of transcripts and analysis of doctor-patient messages X = N o of messages given Y = N o of messages recalled

Diagnosis
show that you got TB.
Type o f house -Brick house with tiles Assessment of compliance.The criterion for labelling a patient to be 'non-compliant* was the failure to take anti-tuberculosis medication for more than two weeks duration.This was assessed by a) number of pills taken (checked ticks on green card and number of pills pre-

Table 2 :
Group means, standard deviations and significance tests of sociodemographic variables for the com pliant and non-com pliant groups

Table 3 :
Onset of tuberculosis for the com pliant and non-com pliant groups

Table 4 :
Group means, standard deviations and significance tests of tuberculosis causative beliefs from Problem Portrait Technique (P PT) rated from 0 to 10 indicating the importance of the behaviour (10 being most im portant) as well as frequency of agreed responses (in percent) for the compliant and non-com pliant groups

Table 5 :
Help-seeking behaviour for tuberculosis other than biomedical prior to first diagnosis

Table 6 :
Knowledge about transmission of tuberculosis (rated from 1 = strongly agree to 5= stron gly disagree)

Table 8 :
Past six months substance use and fam ily and community history of TB

Table 9 :
Group means, standard deviations and significance tests of health beliefs and other variables for the com pliant and non-com pliant groups (scored from 1 = strongly agree to 5 = stron gly disagree)

Table 1 0
: Analysis of health practitioner-patient interaction