Delay to initiate treatment is associated with unsuccessful treatment outcome among new pulmonary tuberculosis patients in Tigray, Northern Ethiopia: a follow up study

Background : Previous studies in Ethiopia indicated that tuberculosis (TB) patient’s elapse long time before treatment initiation. However, there is very limited evidence on the association of delay to initiate treatment with treatment outcome. Objective : To investigate the association of time to treatment initiation delay with treatment outcomes of new adult TB patients in Tigray region of Ethiopia. Methods : We conducted a follow up study from October 2018 to April 2020 by recruiting 875 newly diagnosed Pulmonary Tuberculosis (PTB) patients from 21 randomly selected health facilities. Study participants were selected using simple random sampling technique during treatment initiation from October 1/2018 to October 30/2019. Delay to initiate treatment and treatment outcome were collected using standardized questionnaire and laboratory investigation. Adherence of TB patients to their treatment was collected using a 10 points linear visual analogue scale (VAS) at the end of treatment. The association of delay to initiate treatment with treatment outcome was modeled using log binomial regression model. Statistical signicance was reported whenever p-value was less than 0.05. Data was analyzed using SPSS software version 21. Result : The median total delay to treatment initiation was 62 days with inter-quartile range of 16-221 days. A unite increase in a day to initiate treatment results in increment of risk of unsuccessful treatment outcome by 2.3. Other factors associated with unsuccessful treatment outcomes were being less adherent to the treatment, HIV co infection, being smear positive at initiation of treatment and after 2 months of treatment initiation. Conclusion : delay in a day to initiate treatment is associated with increased risk of unsuccessful treatment outcome. Any effort targeted towards reducing the negative effects of PTB should target on strategies that reduces the length of delay to initiate treatment and strengthen community engagement to improve treatment adherence of patients that have started treatment.


Introduction
In 2018, an estimated 10 million people developed TB disease and 8.6% of these individuals were HIV positive. In the same year there were about half a million new cases of rifampicin-resistant TB and 78% of them were multidrug-resistant TB (1). In 2019 there were 161,000 new TB cases, 27,000 deaths and 1,600 drug resistant TB cases making Ethiopia one of the 30 high-TB -burden countries (2).
To halt the burden attributable to TB, WHO developed the End TB Strategy which includes ending the TB epidemic by eliminating catastrophic expenses due to TB, reducing deaths by 95% and by attaining 90% reduction in the incidence TB by 2035 (1,3). Early detection of TB cases and initiating treatment for the detected cases are main pillars for the End TB strategy (4). Nonetheless, patients elapse very long time to initiate care for their illness and to start treatment in different parts of the world (5-7) and in Ethiopia (8- 10). For example median number of days to initiate treatment is 60 days in Tigray (11,12), 97 days in Bale (13) and 64 days in Addis Ababa (10). Delayed treatment intuition results in more serious clinical illness, increased length of infectiousness and poor treatment outcomes including mortality and drug resistance (14,15). The increased length of infectiousness also increases disease transmission within the community and creates a favorable condition for TB epidemic (9,15). In contrast to these ndings, in Israel successful treatment outcome was not associated with patient delay and health system delay to initiate treatment (16).
Achieving high treatment success enables to reduce TB related mortality, prevalence of drug resistant TB and transmission of TB in the community (2). In the year 2017 successfully treated TB cases globally was 85% and it was 89% in Ethiopia (2,17). However, compared to the treatment success, cure rates of TB patients is still low at 81% in Ethiopia which ranges from 38-92% in different parts of the country (18) In previous studies patient related, socio-demographics, lifestyle and behavioral, clinical, and bacteriologic pro le were signi cantly associated with unsuccessful TB treatment outcomes (19,20). Despite of the long time elapsed to initiate TB treatment and poor treatment outcomes in different parts of Ethiopia, there is very limited evidence that investigated the effect of delay to initiate treatment on the outcome of TB treatment provided to the TB patients. This study investigated association of delayed treatment initiation with treatment outcomes of TB patients. Findings from this study will be an evidence for program people in designing interventions that reduce new TB infection and improve treatment success.

Methods And Materials
Study setting Tigray is one of the nine Regional States of Ethiopia with an estimated 5.3 million population (21). There are 712 health posts, 214 health centers and 38 hospitals (1 specialized hospital, 15 general hospitals, 22 primary hospitals) in the region. The study was conducted in 21 health facilities ( ve hospitals and 16 health centers) in two zones of Tigray, Northern Ethiopia. The two study Zones (Eastern and Mekelle) harbor an estimated 2 million people (21). During the data collection period almost all health facilities were providing directly observed short-course tuberculosis treatment by trained health workers (TB focal persons). The program was introduced in all hospitals, health centers and in most of the health posts (21).
Diagnosis and treatment of all forms of TB across Ethiopia has been based on the national TB control guidelines that specify case de nitions, diagnostic and treatment standards (18). Thus all TB cases enrolled for this study were diagnosed using Genexpert, sputum smear microcopy or clinical signs aided with x-ray (18). The treatment regimens for all new adult cases of TB were treated with combination of Ethambutol(E), Rifampicin(R), Isoniazid(H) and pyrazinamide(Z) for the rst two months (2ERHZ) followed by Rifampicin(R) and Isoniazid (H) combinations for the remaining six months (4RH) (18).
Study design, aim and period A follow-up study with the aim of investigating the effect delayed treatment initiation of PTB on treatment outcome was conducted from October 2018 to April 2020 in 21 health facilities ( ve hospitals and 16 health centers) in Eastern and Mekelle zones of Tigray Regional State of Ethiopia that harbor an estimated 2 million people.

Reference population
All newly diagnosed PTB patients and initiated treatment in all health facilities in Tigray Regional State of Ethiopia.

Study population
All newly diagnosed PTB cases who initiated treatment from October 2018 to April 2020 in the 21 health facilities selected for the current study.
Inclusion criteria All newly diagnosed PTB patients whose age is 18 years or more, with no previous treatment history of TB, were eligible for the study. All Patients who were diagnosed and initiated treatment within the study settings but moved to another place inside the study area were eligible. However; patients diagnosed and started treatment out of the study area but transferred in to the study area were not eligible.

Sample size and sampling technique
The sample size was computed using StatCalc program of Epi-Info at 95% signi cance level, 80% power, considering design effect of 1.5. This is assuming that a unit change in time to initiate PTB treatment results in increment of risk of unsuccessful treatment outcome by 2.02 (risk ratio) and 5.8% of the patients with unsuccessful treatment outcome (22), the required sample size at the end of the follow-up was 489. However, the sample size for assessing predictors of delay computed using 95% signi cance level, 80% power, expected frequency of exposure (visit to non-DOTS center) among controls of 43% and odds ratio (OR) 1.63 (11) with design effect of 1.5 and 4% non-response rate; provided a total of 875 new cases. Hence, the same cases were followed for the outcome; the larger sample of 875 was taken the nal sample size.
The total sample size was proportionally allocated to each health facility based on the average number of TB cases reported by the facility in the previous 2 years. List of eligible patients extracted from daily list of TB patient clinic in each facility was used as our sampling frame to select study participants allocated to that facility (8, 15). Each TB patient's registration numbers of TB patients written on the TB registration book in clinical management was used in the random sampling process and the patients were recruited prospectively during treatment initiation. The number of patients selected from each health facility was obtained by dividing the number of study participants allocated to each of health facility to the total number of data collection time (one year).

Data collection
A structured questionnaire was used to collect the data which was adapted from tools used elsewhere (22-24) and standard TB register book (25). Besides, eligible cases were interviewed for sociodemographic, health care seeking, and lifestyle. The following data were also collected from the study participants: Exposure and outcome data The main exposure variable was delay to initiate treatment (total delay) measured by number of days elapsed between onsets of illness and time to initiation of anti-TB treatment. This delay was computed as a sum of patient and health system delay. The main outcome variable was treatment outcome which was ascertained based on standard de nitions (26,27). For the purpose of this paper this outcome was categorized as binary (favorable versus non-favorable outcome) Treatment adherence data was collected using a 10 points linear visual analogue scale (VAS) (ranging from 0%= no single dose taken; to 100%= no single dose missed) at the end month of treatment, where patients were asked 'How much of your prescribed TB medications have you taken until now in the course of your treatment?' The VAS method was chosen because I) it can be used to assess TB treatment adherence for the whole duration of anti-tuberculosis treatment rather than at a speci c time. II) VAS was used to assess adherence to anti-tuberculosis treatment in another study conducted in the same study area, and was found to be accurate (29). Study participants with adherence level above 90% were considered as adherent based on the World Health Organization (WHO) adherence de nition (30) and national guideline (27).
Clinical data symptoms (respiratory symptoms, constitutional symptoms, or both), and the time duration from onset of symptom up to treatment initiation. Respiratory symptoms included dyspnea, hemoptysis, cough, or expectoration while constitutional symptoms included malaise, loss of appetite, weight loss, fatigue, fever, chills, chest pain or night sweats.

Laboratory investigation
Smear status of all study participants was determined. This was done during treatment initiation, 2nd and 6th month of treatment initiation. One spot sputum specimen (about 5 ml) was collected. Two direct smears were prepared from the specimens for Ziehl-Nielsen (Z-N) staining. The smear preparation and Ziehl-Nielsen (Z-N) staining were performed according to the standard operating procedure (25,26).
Data quality and management A total of 18 nurses with BSc degrees collected the data and 3 Senior BSc degree nurses supervised the eld data collection process. These data collectors and supervisors were trained on the objectives of the study, the importance of the ndings and on how to collect the data from study participants. The training also included description of questionnaire, interviewing techniques. Pretest was done at nearby health facilities not included in the study. The PI cross checked the interview for 10% of the study participants. Efforts were made to minimize recall bias by using local calendars listing main religious and national days to de ne the perceived date of onset of cough (TB symptoms) and time of rst health seeking. The questionnaire was translated into regional language (Tigrigna) spoken by almost all residents in the study area.
Data analysis Data was entered, cleaned and analyzed using SPSS software version 21 (29). The data were described using frequency, proportions, measures of location and dispersion (to calculate the mean and median. As total delay was not normally distributed; the data was transformed before analysis. During analysis delay was considered as continuous variable. Association between the exposure and outcome (unsuccessful vs successful) was determined by log binomial regression model. Accordingly, both simple and multiple logbinomial models were tted to estimate crude and adjusted relative risk (RR) of the outcome variables.
The goodness of t of the model was assessed using Pearson chi-square and Deviance tests. In all the statistical tests, statistical signi cance was judged at p < 0.05.

Operational de nitions
Total delay is the time from the onset of TB symptoms to the rst start of anti-TB treatment.  Defaulter a defaulter is a patient who has been on treatment for at least four weeks (26).

Results
Socio-demographic, clinical characteristics and accessibility to health facilities of study participants A total of 875 newly diagnosed adult pulmonary tuberculosis patients were enrolled from 16 health centers and 5 hospitals. The median age [inter-quartile range (IQR)] of the study participants was 35  years, 58.1% were males, and 54.5% were from urban areas. Regarding to educational status, 390 (47.5%) were lower than secondary school, 314 (39.9%) were secondary school and the remaining 93 (12.6%) college and above. Participants with 28.7% and 26.6% had a BMI of below 18 and below 16, respectively. Similarly, 39.6% had MUAC below 220mm and 23.5% had MUAC measurement of below 200mm. Among the study participants 57% were living within the distance of less than 10KM from the nearest health facility.

Tuberculosis treatment initiation delay
The median total delay of the study participants were 62 days with inter-quartile range of 16-221 days. The mean total delay of the study participants were 125 ± 256(sd) days. TB patients with age of more than 45 years old had the highest median delay to initiate TB treatment compared to the other age groups. Similarly, rural residents, house wives and those TB patients who made more than one visit to health facility had higher median time of delay to initiate TB treatment compared to their counter parts.
Furthermore; TB patients with MUAC measurement of lower than 200mm and who traveled for more than 10 KM had delayed longer to initiate TB treatment (Table 1). , being smear positive after 2 months of treatment initiation (ARR = 5.7,95% CI; 2.7-18.5) and being secondary school (ARR = 2.1, 95% CI;1.14-4.33) are signi cantly associated with higher risk of unsuccessful treatment outcomes.

Discussion
In this cohort study we have investigated treatment initiation delay and its association with TB treatment outcome of PTB patients. Hence, here listed below are the main discussion points from the follow up study summarized under two subthemes as association of treatment initiation delay with treatment outcome and other factors affecting treatment outcome.
Association of treatment initiation delay with TB treatment outcome A unite increase in a day to initiate treatment results in increment of risk of unsuccessful treatment outcome. This nding is in line with studies from Ethiopia (24) and China (30). The reason for increased risk of unsuccessful treatment outcome with increasing the time to initiate TB treatment could be (i), those TB patients might have severe clinical presentation which is a previously reported predictor of unsuccessful treatment outcomes (14). In this study, those who had not delayed initiating treatment and those who have initiated treatment after a delay of 30 to 60 days had 4% and 17% lower risk to develop severe clinical presentation compared to those who have initiated treatment after two and more months (15). (ii), studies had indicated that none formal medication sources (traditional medication, prescribed and self-treatment) are associated with delayed treatment initiation which might lead to poor treatment outcome (8). In the current study, the majority (64.8%) of the cases had visited at least two times until the actual diagnosis of TB is made at which time both self and prescribed medicines had been used (8).
Other factors associated with TB treatment outcome In the current study HIV co-morbidity increase risk of unsuccessful outcome which is similar with other studies in Ethiopia (24,31,32). The increased risk of unsuccessful outcome among the HIV co infected could be due to the complex and overlapping drug interactions, toxicities, and TB-associated immune reconstitution in ammatory syndrome (33). Furthermore, this increased risk might also be associated with lower uptake of TB/HIV collaborative interventions.
Being smear positive at initiation of treatment is associated with higher risk of unsuccessful treatment.
This study result is supported by previous reports in Ethiopia (20) and elsewhere (34). This might be due to the reason that patients with smear-positivity have more advanced disease which leads to lower tuberculosis treatment response. This might also be due to the heavy initial bacillary load which is associated with delay in smear conversion leading to treatment failure and death (35).
Similar to other studies in Ethiopia (36, 37) and elsewhere (35) being smear positive after 2 months of treatment initiation was associated with poor treatment outcome. This could be related with patient attitude and behavior towards the disease (36). Patients may feel better and assume they are cured by the subsidence of symptoms after completing the initiation phase. On the other hand, patients may also loss interest in continuing treatment if symptoms persist and no improvements are seen after the initial phase of treatment. The other reason might be related with partly emergence drug resistance.
But this research had limitations like measurement of the exposure relied on patient self-report which is liable to recall bias. However, we tried to minimize this bias through interviewing patients soon after diagnosis and helping them to recall using local events like national holidays, religious days, and dates of some events. Second, we studied only new PTB and adult cases so that the ndings cannot be generalized to all forms of TB cases among all age groups. On the other hand, relatively large sample, geographic coverage (include all levels of the health care system), reduced selection bias through random enrolment, able to measure treatment adherence, being prospective design, direct estimation of risk and use of standard outcome ascertainment could be mentioned as strength of the study. Therefore, the study is valid and could be applied to new TB cases in similar settings. We would like to express our gratitude to Aklilu Lemma Institute of Pathobiology, Addis Ababa University, for giving us the opportunity to undertake this study. We would like also to thank Mekelle University for giving us sponsorship to conduct the research. We would also like to thank all health professionals working in the health facilities where this study was conducted, the study participants and Tigray regional health bureau.

Conclusion And Recommendation
Authors' contributions KT wrote the proposal, participated in data collection and drafted the manuscript. AM, GB, GM and NB commented the proposal with great revisions, participated in data analysis and revised drafts of the manuscript. All Authors revised and approved the nal manuscript.

Funding
This is not funded externally.

Availability of data and materials
All data is contained in the manuscript and other raw data will be provided to readers upon the request to the corresponding author.
Ethics approval and consent to participate Ethical approval for this study was obtained from the Research and Ethical Committee of Addis Ababa University, Aklillu Lemma Institute of Pathobiology. A formal permission letter was also obtained from the regional health bureau of Tigray to the selected district health o ce. Each health facility selected for the study was contacted with a permission letter from the district health o ce. Each case diagnosed as TB according to the national guideline was consented in a written form before the interview. To assure con dentiality interview with TB case was held in a private room and the information collected was recorded anonymously. All patients were treated according to the national guideline and there was no any interference during the treatment.

Consent for publication
Not applicable.

Competing interests
There is no nancial and non-nancial con ict among the authors.