Background
Indonesia's National TB Control Programme (NTP) faces many challenges despite its impressive performance in reaching the international targets for case detection (>70%) and treatment success (>85%) in 2006 [
1]. The country still ranks third in the world in terms of tuberculosis (TB) burden and needs to ensure that it progresses beyond current achievements toward the Stop TB Partnership targets of halving prevalence and death by 2015 [
1]. One key challenge for attaining these targets is early diagnosis and treatment of TB patients, which is considered to be the foundation of TB control. Delay in diagnosis and treatment is associated with greater transmission of infection [
2] and poorer outcomes for the patients [
3].
In Indonesia, the mean duration from start of the first TB symptoms to treatment is 63 days [
4], which is on par with the 62 days reported for India [
5] and surpasses the 30 days cut-off point commonly used to define long delay [
6]. The NTP needs to be informed on the mechanisms underlying delays in receiving an appropriate diagnosis and treatment to devise remedial interventions. Studies from other countries have documented a wide range of factors associated with delay, including HIV infection, gender, rural residence, visits to private practitioners, low education, low awareness, self-treatment and stigma [
6]. Little is known however about the importance of these factors in the Indonesian context. With this in mind, we seek to explore patient journeys toward TB treatment in Indonesia.
Methods
Study setting
Jogjakarta province is located in the central southern part of Java, Indonesia. The provincial administration is divided into five districts with a total population of about 3.2 million inhabitants. The province's primary care network consists of approximately 650 private practitioners and 118 community health centres. These first line services are backed up by 9 public hospitals and 24 private hospitals (including maternity and mental hospitals) as well as 5 chest clinics. The number of traditional healers practicing in the area is unknown.
The community health centres form the backbone of the DOTS (Directly Observed Treatment, Short-course) services in the province. The hospitals and chest clinics have also been engaged in the DOTS strategy through the Hospital DOTS Linkage project since the year 2000. Private practitioners (PPs) and traditional healers were not yet formally incorporated into the DOTS strategy at the time of the study. PPs diagnose TB based on clinical symptoms, chest X-ray or smear microscopy, traditional healers diagnose TB on symptoms. Some of the PPs refer their patients to DOTS services for diagnosis and treatment, albeit not systematically [
7]. There are no institutional arrangements for DOTS yet for these care providers and patients have to pay for consultation and treatment.
Study population and sampling
We used a purposive sampling method applying the extreme case sampling technique [
8], selecting the two districts in Jogjakarta province with the lowest (Kulon Progo district) and highest (Jogjakarta municipality) case notification rate (CNR) in 2003, respectively 23/100,000 population and 110/100,000 population. The study population consisted of TB patients, their family and community members. Every individual diagnosed and recorded as TB patient by a public or private health care provider or a traditional healer was eligible for inclusion in the study. We aimed to recruit TB patients through formal and informal health care providers i.e. hospitals, community health centers, chest clinics, PPs, and traditional healers. During a three months data collection period we would be able to recruit 70 patients based on the number of patients detected in quarters before the data collection. These 70 patients were to be recruited using a purposive sampling method applying the maximum variation technique covering various types of care providers (Table
1) and weighting based on the ratio of smear-positive cases reported in Jogjakarta and Kulon Progo (11:3). The TB patient respondents were newly diagnosed or diagnosed a maximum of 3 months before the data collection. Additional inclusion criteria were local residency, clarity of registered home address and consent for interview. Patients meeting inclusion criteria within each care provider category were then recruited randomly. We halted the recruitment process when there were indications of information saturation.
Table 1
Sampling frame for recruitment of TB patients to be interviewed from different health care providers.
Hospital | 10 | 25 | 25 | 18 |
Health centre | 10 | 18 | 14 | 11 |
Chest clinic | 10 | 36 | 36 | 29 |
Private practitioner | 30 | 10 | 9 | 9 |
Traditional healer | 10 | - | - | - |
Total
| 70 | 89 | 84 | 67 |
We aimed to interview 20 family members of interviewed patients who are most knowledgeable of the patients' experience, for DOTS patients this would be those who had been treatment observers. We asked all family members of interviewed patients whether they would be willing to be interviewed. From the list of consenting family members, we recruited family members using purposive sampling applying the maximum variation technique covering the various types of care providers and ratio of patients between Jogjakarta municipality and Kulon Progo district. We additionally considered the potential of the respective family members as respondents (communication ability and richness of information).
Community is defined as a group of people living in the smallest administrative area (hamlet). Community members living in the same hamlet as the included TB patients were eligible for inclusion in FGDs. The hamlets with a TB patient that qualified for participation in the focus group discussions (FGDs) were identified by the district health officer, as the FGDs were organized in conjunction with regular community meetings. Hence the FGD participants were attendants of the meetings who agreed to stay and participate in the FGD after the meeting.
Data collection and analysis
NR, Sh, Su, Pu and YS performed in-depth interviews with the TB patients and with family members of TB patients. The guidelines for the in-depth interviews for patients included questions on age, sex, level of education, economic status [income more or less than standard minimum wage, i.e. IDR 400,000 (equivalent to US$ 47 per month in 2003)], occupation, history of the symptoms, history of their initial and subsequent health care seeking behavior and treatment, details of their management, knowledge and perception of TB, and perceived stigmatization. Interviews of family members included questions on perceptions of the patient's disease history and advice given to the patient.
NR, Sh, Su, Pu and YS also conducted six FGD with community members: three in the Jogjakarta Municipality (urban) and three in Kulon Progo District (rural). The guidelines for the FGD covered knowledge of the community on TB symptoms, cause(s), transmission, consequences, cure, treatment quality and cost, action taken in case a relative or neighbor would suffer from TB, and opinion of existing health care services.
The interviews and FGDs were conducted in the Indonesian language or the local Javanese dialect and recorded on audio-tape with permission from the subjects. The recording was transcribed within 48-hours by a research assistant and checked by at least one core research team member. The transcripts were scrutinized to identify emerging and recurrent themes and a codebook was progressively established and structured. Text units were coded systematically. Coding frequency permitted to identify key issues and trends. All of these analysis procedures were carried out in Indonesian language by at least two different persons within the research team. English translation was only carried out in the process of report writing. The original interviewers had the opportunity to make corrections to the translated quotations.
The study was approved by the ethical review committee of the Faculty of Medicine, Gadjah Mada University, Indonesia.
Discussion
Our study revealed that there are four basic types of TB patient journeys in the study area. The majority of the TB patients (55/67) were diagnosed quite late (delays of >30 days) and many took dubious treatment before TB diagnosis and initiation of appropriate (DOTS) treatment. Most patients and their family and community members were knowledgeable of the fact that TB is a transmissible disease and that it can be cured by taking drugs. There were notable misconceptions regarding the cause of TB. Stigma was present albeit limited. Gender, rural/urban residence and educational level made no apparent difference in care seeking behavior. Income/cost in connection with home-clinic distance, and advice from household members or friends seemed to be more influential.
The patient journey patterns suggest that only 12 patients in the study area were diagnosed by DOTS providers within one month after the first symptoms appeared. Delay for a month or more was found in TB patients in Lusaka [
9] and South India [
10] respectively. Surveys from Ethiopia and South Africa have reported longer patient delays [
11,
12]. Notably, in our study patients with a low income were diagnosed and treated earlier with DOTS than those with a higher income. This was not documented in the other studies. The observed differences reflect differences of contextual factors, most importantly access to general care and DOTS services of the poor, and stereotypes about what people consider high quality care which make them look for other forms of care if they have the means.
A considerable proportion of the interviewed TB patients had been in contact with non-DOTS providers before turning to DOTS services. Lonnroth et al. [
13] reported similar findings in Vietnam where most TB patients preferred to seek care first at private physicians (17.9%) or private pharmacies (29.9%). A study in South India [
10] showed that previous contact with private providers significantly contributed to greater health system delay. Notably, there is also a transfer from DOTS services to non-DOTS services, most likely due to failure to identify TB suspects, poor quality of the services or patients' over-expectation of a speedy recovery, as was observed in Vietnam and Bali [
13,
14].
Over 80% of the TB patients in the study areas consulted several providers, both DOTS and non-DOTS, before they were diagnosed. This is in line with the findings of a study in South India where only 22% of the patients were diagnosed at the health facility where they first sought care, the others shopped around for care at various health facilities before a diagnosis of tuberculosis was made [
10]. Nearly half of all the patients in the study in South India had visited three or more health facilities before a diagnosis of tuberculosis was made. In our setting, shopping around appeared related to interrelated factors such as home-clinic distance, perceived quality, cost, previous experience and advice from others.
WHO recommends in the new Stop TB Strategy to engage non-NTP providers [
15]. However, the evidence on the effectiveness of the PPM DOTS strategy remains controversial with wide variation of outcomes across settings [
16,
17]. Taking into account our study findings, i.e. the large number of TB patients seeking health care at private practitioners, assessment of the potential contribution of a PPM initiative in Indonesia is urgently needed to guide further program policy making.
We did not find evidence that educational level influenced care seeking behavior. Godfrey-Faussett et al. [
9] reported a similar observation in their delay study in Zambia. Rajeswari et al. [
10], however, showed that there is an association between patient delay and educational background in South India. There is a need to take into account the special role of TB knowledge in this association. Portero et al. [
18], reported that knowledge of germs as the cause of TB showed a significant correlation with college level education, while belief of heredity as a cause was significantly associated with non formal education. Others reported that lack of knowledge about TB treatment and large home-clinic distances were factors associated with longer patient's delay in smear negative patients [
10]. Another study by Portero et al. [
18] showed that only 25% of those interviewed knew that TB is due to germs (bacteria or microbes). Watkins and Plant [
14] reported that on the neighboring island of Bali people believe that TB is caused by magic, spirits or heredity so they didn't need to attend the health providers. We also found these beliefs regarding the cause of TB in our study population. It should be noted that almost all the interviewed patients had at diagnosis received health education on TB. Thus it is surprising that so many TB patients did not know the biomedical cause of TB. Contrary to the patients in Bali, however, they had found their way to biomedical care.
Like in Bali [
14], community members related causes mostly to physical factors such as staying up late at night, sleeping on the floor, working too hard and psychological factors including mental stress and worry. These non-infectious causes were also mentioned in another study carried out in Manila [
18]. Despite the varying causes mentioned, participants in our study agreed that TB is an infectious and transmissible disease. Ways of transmission they cited were by talking, coughing, by dirty air, by sputum, or by sharing eating utensils, as was found in the study in The Philippines [
18].
Delay studies have reported conflicting patterns of gender differences. Rajeswari et al. [
10] showed that, contrary to popular belief, in South India men rather than women were more likely to delay seeking care: 32% of the male TB patients waited longer than one month to seek care against 23% of the females. Our findings do not indicate that gender influences care seeking behavior. This however should be further substantiated by a quantitative study with a larger sample.
Income seems to be closely linked to care seeking actions, though these actions not necessarily lead to the most cost-effective care. Those with a higher income, for example, could access a wide range of (private) providers and regarded the quality of DOTS providers at the health centers as poor. In contrast, those with an income below the minimum wage tended to seek low cost public services, particularly at local health centers, though they also might have preferred private treatment. Our findings show that advice of relatives such as spouses, parents or friends played an important role in seeking initial care from particular providers. The most frequently mentioned reason for choosing an initial provider was advice from household or community members. Both family and community members confirmed that they often provided patients with advice. Thus, the opinion of household and community members is quite influential in regard to care seeking behavior.
Our findings indicated that most patients in our study area didn't feel stigmatized because either their spouse, relatives and friends were supportive, or they showed no change in behavior when it became known that a patient had TB. There were however indications that spouses stigmatize patients more than blood relatives. Godfrey-Faussett et al. [
9] reported that in Zambia there was evidence of considerable stigmatization of TB patients by the community as well as by relatives. The limited magnitude of stigma we documented may reflect the general level of education and the intensity of health promotion campaigns related to TB.
We have collected data mainly on patients' related factors rather than providers'. However, our findings revealed that providers seem to contribute rather considerably to total delay by frequently failing to examine TB suspects with smear microscopy. The current body of evidence does point out that the solutions to decrease long delays to TB treatment are to be sought on the providers', not on the patients' side [
19]. Thus, interventions to ensure more proper implementation of NTP guidelines in health facilities seemingly are urgently needed.
This study is limited by qualitative research boundaries. Issues perceived by patients, family members and communities were identified. Although trends emerge, the respective influence of each issue was not quantified. This could be documented through a quantitative survey building on our findings, which points out the key issues to be taken into account. We also could not include traditional healers due to the difficulty of identifying and approaching them as they were largely unregulated and not institutionalized. Additionally, with the few who were identified, we did not manage to establish sufficient rapport for recruiting their patients for interviews within the study data collection period or they did not keep a patient register from which TB patients can be selected. Thus, interpretation of our findings can only be restricted to the context of the formal health services. Although our findings are to a certain extend context bound, they may be relevant to other provinces in Indonesia with similar socio-economic, TB epidemiology and health system characteristics. Some specific findings may even hold in similar settings in other countries.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
NR, YM, Sh, Su, Pu, YS, CV and MVDW made substantial contributions to conception and design. NR, Sh, Su, Pu and YS collected the data. NR, YM, Sh, Su, Pu, YS, CV and MVDW made substantial contribution to analysis and interpretation of data. NR, YM, Sh, Su, Pu, YS, CV and MVDW have been involved in drafting the manuscript. NR, YM and MVDW have contributed to revising the manuscript critically for important intellectual content. NR, YM, Sh, Su, Pu, YS, CV and MVDW have given final approval of the version to be published.