Background
Tuberculosis (TB) has infected billions and has claimed the lives of millions. Despite the advances in knowledge about the importance of early case finding and treatment, millions of TB cases are being missed by the health system which has led to sustained transmission and increased the pool population at risk of acquiring and developing TB [
1,
2]. It is estimated that one active TB case can infect about 10–15 healthy individuals leaving them with 5–10% life time risk of developing active TB [
3]. Therefore, programme interventions that ensure early identification of index TB cases and support initiation of prompt treatment with the right dose, quality and duration are required to curb the epidemic.
Effectiveness of the National TB Programme (NTP) depends on its capacity to make early case finding and provision of standard treatment. The diagnostic and treatment services should be decentralized and be accessible to community to seek care. Health seeking behavior is the result of a complex interplay between community awareness, access and availability of the services, cost related to seeking care and sociocultural factors including stigma and beliefs. Factors that affect health seeking behavior lead to delay in seeking care and hence increase TB transmission, mortality and remain a challenge to TB programme performance [
4,
5]. Measuring the delay in seeking and getting care is a proxy indicator for good programme performance in reaching and serving the community.
Delay is measured from the time of developing the first symptoms to the time of initiating treatment and is categorized by patient and health facility delay. Patient delay is the time between the onset of the first TB symptom to the time of seeking care while health facility delay is the duration between the first contact with health facility to the time of initiating treatment. Total delay is the sum of TB patient and health facility delays. Treatment delay is measured by the duration from the time of diagnosis to when TB patient started treatment.
Studies reported that delays in seeking care are associated with low awareness about TB, rural residence, high stigma [
6], gender [
7‐
10], number and type of facilities visited [
11‐
13], cost of seeking TB care [
14], poverty, comorbidity with HIV [
15], socioeconomic status [
16] and cultural barriers [
17]. Total patient delay was more than ten weeks in Sub Saharan Africa [
12,
18‐
20], about seven weeks in Asia - highest in Pakistan and Vietnam nearly doubled [
21‐
24] and about six weeks in Latin America [
25]. Though the duration of delay is shorter, it was also reported from low incident states, minority groups and natives due to low index of suspicion about TB [
26,
27]. Unfortunately, after patients sought care and arrived at health facilities, there were also further diagnostic and treatment delays [
28].
Studies have shown that patient delay contributed more to total delay than did health system delays. However, health system delay has been reported to be longer in smear negative, extra pulmonary TB (EPTB) and among those who visited non TB service providing facilities [
29‐
31]. This could be due to the severity of symptoms TB patients present with that resulted in repeated visits to health facilities, led to protracted diagnostic procedures, and further delayed due to unavailability of diagnostic facilities required to confirm TB diagnosis [
32‐
34].
Ethiopia has been implementing TB prevention and control for about a quarter century. It is successfully treating more than 90 % of enrolled drug sensitive cases [
2].
With rapid expansion and decentralization of health services to the community using health extension workers (HEWs), access to TB care has increased and community awareness is expected to also increase.
Studies from Ethiopia reported delays to be associated with educational status, poverty, awareness about TB and accessibility of TB services [
35,
36] with higher proportion of health system delay from northern Ethiopia [
37]. However, most studies were limited to specific population groups and quantitative designs. In this study, we aimed to measure nationwide delay and the associated factors using a mixed method study using qualitative and quantitative data to understand the context and generate an evidence base to advise policy and decision making at the National TB Programme of Ethiopia.
Methods
Study setting and population
Ethiopia is a high TB burden country with a population of more than 100 million. 85% of the population lives in rural areas. Ethiopia is administratively divided into nine regional states and two city administrations. The National TB Programme started DOTS in 1995. Currently 256 hospitals and 3390 health centers provide diagnostic and/or treatment TB services. More than 15,000 health posts provide community based TB services. HEWs are female village residents who completed at least grade ten who received one-year training to provide primary health care to a community of 5000 people. They provide TB services from the health posts under disease prevention and health promotion which includes awareness creation, identifying presumptive TB cases and referral, providing DOT, and retrieving absentees and defaulters [
38].
Study design and sampling
This is a mixed method study conducted from October to November, 2017 in the nine Challenge TB supported regions of Ethiopia that covers 92% of the national population. Challenge TB project is USAID’s global flagship TB control support mechanism designed for 2014 to 2019.
We used a multistage cluster sampling technique to randomly select the zones, districts and kebeles for the study. From six regions and two city administrations, we randomly selected 32 districts and eight sub cities/districts respectively. A health center was randomly selected from each district or sub city with a total of 40 health centers included in the study. From each randomly selected health center, 21 TB patients were consecutively enrolled from each health center. We also conducted qualitative interviews to understand the reasons for delay. The details of the methods are described in a published paper on Knowledge Attitude and Practice about TB in Ethiopia in 2017 [
39].
Operational definition
Patient delay was defined as the time between the first symptoms to date of seeking health care and operationally defined as more than 2 weeks. Diagnostic delay is the interval between first visit to health facilities for TB diagnosis to date of diagnosis for TB. Treatment delay was considered if TB treatment was initiated 1 week or longer after patient was diagnosed with TB. Facility delay is the time between the first visit to treatment initiation. Total delay is the sum of TB patient and facility delay, that is, starting treatment 3 weeks after the onset of symptoms.
Quantitative studies
Single population proportion formula was used to estimate the sample size using EDHS 2011 KAP report [
7], assuming that 50% of the study participants will have delay of at least 2 weeks to seek TB care. We used design effect of 2 and added 10% to compensate for non-response rate. 844 TB patients were enrolled from 40 health centers.
Qualitative studies
We conducted 18 Focus Group Discussions (FGD). Two FGDs, one for women and one for men, were conducted in the selected kebeles in the respective regions. The kebele administrators assisted in the identification of participants of the FGDs from the general population.
The questionnaires for the study were adopted from the WHO guide for KAP [
14]. The questionnaires were initially prepared in English and translated into regional languages and translated back to ensure the quality. FGDs were developed for this study and were conducted in local languages using pretested, semi-structured in-depth interviews (IDI) and open ended topic guides for FGDs. Two data collectors were deployed per site to conduct FGDs and IDIs. 18 FGD sessions and 76 IDIs were conducted and audio recorded. Tablets were used for data collection using a web-based data tool using CSPro software.
Data management and analysis
Quantitative data
Data was extracted from the web-based system and exported to SPSS version 25 for analysis. Mean, median and interquartile range was used to describe the data. Binary logistic and multivariate logistic regression, for independent variables with p < 0.25, were used, which includes marital status, wealth index, gender, travel time to health institution, setting and region to construct multivariate model. Addition of other variables like occupation and education compromised model fitness.
Qualitative data
Was imported to Open Code software and analyzed using thematic content analysis technique. Direct verbatim and results from the coding and categorization were used to develop the narrative, triangulated and synthesized with the quantitative results [
39,
40].
Data quality assurance
We selected and trained experienced data collectors. Pretested questionnaires were used for data collection. Supervisors were assigned and conducted random data and household checks, and reviewed the questionnaires. The CSPro expert regularly checked for completeness and errors.
Ethical considerations
Ethical clearance was obtained from the Ministry of Science and Technology. We obtained support letters from the Federal Ministry of Health to the regions and research areas. Ethics review committee approved informed verbal consent to be obtained from the study participants as most of the study participants are from rural and cannot read and write. We obtained informed consent from the study participants.
Discussion
Early diagnosis and timely initiation of treatment plays a key role in reducing disease transmission, disease severity and risk of death [
20,
21]. We report national level delay in seeking diagnosis and initiation of treatment in Ethiopia. Patients delayed seeking care for 21 days and facility delayed diagnosis for 6 days and treatment for 6 days making total delay of 33 days. This is substantiated by qualitative results from patients and health care workers who identified lack of awareness, cost of seeking care and other socioeconomic factors to influence patient delays.
Compared to delay studies conducted in Ethiopia, we report shorter patient delay of 31 days from northern Ethiopia [
41], 30 days reported from Hadiya zone in southern Ethiopia [
36], 60 days in Addis Ababa [
35] and more than 2 months pretreatment duration from Sidama in southern Ethiopia [
42], but longer than study from southern Ethiopia that reported 4 days [
43]. There is much reduction in the overall delay as a result of increased coverage of health service, better awareness, difference in the study period and changes in socioeconomic conditions. Delays are longer among elderly and patients with EPTB and smear negative TB cases which could be due to the limited diagnostic capacity to detect TB among the elderly, complex diagnostic pathways and delayed health seeking.
Compared to studies from SSA countries, patient delays are shorter in Ethiopia though higher than 2 weeks reported from Cameroon [
44] and similar to 3 weeks reported from Lagos, Nigeria [
18] and shorter than reports from Uganda, rural Nigeria, and other low incidence states which is in the range of 4–8 weeks [
12,
19,
26]. Patients’ residence, age, distance from health facilities, awareness about TB as a curable disease and fear of testing for HIV were the main reasons for delay.
Women had high odds of longer delays compared to men possibly due to higher stigma, lower priority or attention paid to them by the community, socioeconomic problems and domestic responsibilities they shoulder [
8,
24]. This could be complex in settings where there is an inadequately decentralized heath system and gender norms favor men for accessing health services. Hence, gender-sensitive TB services should be identified and designed to address such variations in service delivery.
Patients from urban setting had lower odds of shorter delay compared to rural areas [
43] mainly due to better awareness about TB, access to services, better education status and socioeconomic condition. However, in urban slums, urban poverty could be worse for TB patients. Contrary to other studies, we report longer delay in urban areas. This could be due to diagnostic pathways related to diagnosis of smear negative and EPTB and HIV [
15,
45,
46]. In addition, 85% of the Ethiopian population lives in rural areas and this may have under represented the urban TB patients as we did not estimate the sample size to analyse urban-rural differences.
Due to low awareness about the disease, about the severity and about availability of services, patients ignore TB symptoms and remain at home or visit traditional or private pharmacies. Moreover, visiting health facilities that do not provide diagnostic services, have inadequate technical capacity, and lack diagnostic facilities like chest x-ray, contributes to the delay.
Emerging themes of the qualitative study reported by TB patients, program managers and community showed that lack of awareness, access to informal or illegal drug sellers, inadequate service provision, fear of knowing one is HIV positive, poverty and delay of service provision due to longer waiting time [
8,
15,
34] led to delayed TB care.
Patients from high income countries generally have lower delays in seeking care due to the health infrastructure, improved awareness and health seeking. However, a systematic review indicated similarities [
47]. This indicates the importance of reviewing case finding strategy and making pro-poor strategies in their respective contexts. Minorities and disadvantaged people in developed countries still have more delays [
16].
In resource constrained settings, poverty and inadequate service decentralization complicate patient pathways. Decentralization of services with adequate technical and diagnostic capacity could facilitate early diagnosis. However, stigma, distance and related cost still remain a challenge to the patients to seek diagnosis and initiate treatment [
48].
In the era of the End TB Strategy, reducing patient and household costs of seeking care is one of the key components and yet cost of seeking care is high in Ethiopia [
43,
49]. Thus, patient insurance mechanisms, waiver services and possibility of reimbursing costs related to seeking TB care could be potential interventions to improve health seeking in resource-constrained settings. Moreover, decentralization of services to the community, early identification of cases and linking to care appear feasible and cost effective options. There is a community-based health insurance system which covers cost of seeking care for poor TB patients. Generally, understanding patient pathway of seeking care is important to program TB services, reduce delay and reduce cost [
34,
50].
In the Ethiopian setting, engagement of the HEWs could be a cost effective model (women residing in villages who were trained and employed by the government to provide health care in the community). However, there is limited disclosure to HEWs and their engagement is low compared to the coverage of the HEWs in the country, more than 40,000 HEWs. Generally, mechanisms that reduce the distance between the community, patient and diagnostic units, bringing treatment units in the proximity of the patients at health posts or households and strengthening referral linkages are required to ensure timely diagnosis and prompt initiation of treatment.
The limitation of the study was that it did not consider subnational variations of delay and sample size in the regions was not adequate to measure why delays varied in the regions. Overall delays are likely to be longer since we did analyse the delays by type of tuberculosis. It is likely that smear-negative and extra-pulmonary TB patients would have experienced longer delays due to diagnostic challenges. Patient pathway analysis which could have helped the NTP in designing interventions to improve health seeking is beyond the scope of the study.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.