EFFECTIVENESS OF THE COMMUNITY-BASED DOTS STRATEGY ON TUBERCULOSIS TREATMENT SUCCESS RATES IN NAMIBIA

School of Pharmacy, Faculty of Health Sciences, University of Namibia, Windhoek, Namibia; Ministry of Health and Social Services, National Tuberculosis and Leprosy Programme, Windhoek, Namibia; Centre for Health Services and Policy Research, School of Population and Public Health, The University of British Columbia, Canada; Ministry of Health and Social Services, Tuberculosis Clinic, Katutura Intermediate Hospital, Windhoek, Namibia; Anoixis Corporation, Natick, MA, USA. Department of Laboratory Medicine, Division of Clinical Pharmacology, Karolinska Institutet, Karolinska University Hospital Huddinge, Stockholm, Sweden.


INTRODUCTION
Tuberculosis (TB) remains a significant health problem in many lower and middle-income countries. In 2015, there were 10.4 million cases of TB worldwide, leading to an estimated 1.8 million fatalities. 1 The disease is particularly prevalent in Sub-Saharan African countries such as Namibia, where the case notification rate (CNR, i.e. the number of new and relapse TB cases notified in a year) was 489 cases per 100,000 people in 2015. 2 A major strategy to reduce TB incidence has been Directly Observed Treatment Short-course (DOTS), which was implemented in Namibia in 1995.
Directly observed treatment (DOT), i.e standardized anti-TB drug regimens administered to patients under direct observation, remains a critical strategic goal of DOTS implementation in Namibia. 3,4 TB case identification and optimization of treatment outcomes through DOTS are the key global strategies to "end TB" in Namibia by 2035. 1,4 Unsuccessful treatment outcomes however, are important risk factors for the development of drug-resistant TB, a condition that is extremely difficult and expensive to treat. [5][6][7][8][9] In the past decade the community-based DOTS has improved treatment outcomes globally and in Namibia. 9,10 Nevertheless, Namibia, an upper-middle income country in southern Africa with a population of 2.2 million, remains one of the countries with the highest incidence of TB in the world. 1,2,10 Therefore, facility-based DOTS (FB-DOTS) was scaled-up to all public health facilities in Namibia between 1991-1995 as a strategy to control TB and to improve treatment outcomes. 10,11 In Namibia, FB-DOTS refers to when directly observed therapy and related services were only accessible at a health facility before 2005, and CB-DOTS is when DOTS services were extended to villages and households through community based health workers. An assessment of the FB-DOTS strategy in Namibia in 2002 showed that, since its introduction in 1991-1995, TB incidence rates had not declined and treatment success rate (TSR, i.e. the proportion of cases cured or completed TB treatment in a given year) was at its lowest in 2004. 10 12 The access to high quality CB-DOTS was further expanded, i.e to all regions, public-private workplaces and integrated with community-based HIV (Human immunodeficiency virus) care and enhanced i.e improved quality of bacteriological assessments and the standardization of DOTS services such as treatment, DOT support, among others under MTP-II (2010-2016) to empower DOT supporters within each community to deliver quality DOT services. 13,14 The targets for treatment success rate under MTP-I and MTP-II were 85% and 90%, respectively. 12,13 With the implementation of MTP-I in 2004, an electronic TB data base was started to closely monitor treatment outcomes. The objective of this study was to use the annual rates of treatment success, cure, treatment completion, before (1996)(1997)(1998)(1999)(2000)(2001)(2002)(2003)(2004) and after (2005)(2006)(2007)(2008)(2009)(2010)(2011)(2012)(2013)(2014)(2015) the implementation of MTP, to assess the effectiveness of CB-DOTS to improve TB treatment outcomes.

Data collection
Quantitative population level data on annual TB rates of treatment success, cure (i.e. the proportion of cases with pulmonary (PTB), that is TB with lung parenchyma involvement, with bacteriologically confirmed TB at the start of treatment whose sputum was smear-or culture-negative in the last month of treatment), treatment completion (i.e. the proportion of TB cases in a given year that successfully completed TB treatment without bacteriological evidence of success), and case notification for all cases of TB registered during the period 1995 to 2015 were extracted from the annual reports of the National Tuberculosis and Leprosy Programme (NTLP) of the Ministry of Health and Social Services (MOHSS) of Namibia. 11 In Namibia treatment success for extra-pulmonary TB (EPTB, i.e TB disease at sites other than the lung parenchyma) is reported as the proportion of cases with/without aspirate bacteriological or cytology/histology results who are clinically well after completion of 6-8 months of treatment. 11 The National Institute of Pathology, an accredited laboratory performs all the bacteriological testing for TB cases in all DOTS sites in Namibia. Consequently, a case of cure is confirmed by a medical officer base on TB guidelines, which are implemented at all DOT sites with supported training. These annual rates are based on aggregates of quarterly reports collated from district and regional TB case registers. The annual rates were validated against the WHO Analytical Country Summaries for TB, as well as the data reported by the World Bank, United States Agency of International Development (USAID) and Global Fund. 1 Twenty validated annual TSR, CNR from 1995/1996 -2014/2015 for cases with PTB, EPTB, drug susceptible TB (DST) and drug resistant TB (DR-TB) were included in the study. Annual rates reported before 1995 were excluded since there was no systematic reporting on TB outcomes before the establishment of the NTLP programme in 1991. 10,11 During the study period, the case definitions for cure and treatment completion did not change and the DOTS services were free of cost and treatment support as the only incentive during CB-DOTS.

Ethics
Data reported in public documents by the health authorities of Namibia were used as the primary source to assess the effectiveness of an intervention at the population level. Ethical approval for the study was obtained from the human ethics committees of the MOHSS and the University of Namibia.

RESULTS
The annual number of case notifications by TB category and the TSR, CNR, and population covariates from 1996 to 2015 are shown in Figures 1 and 2, respectively.  The mean ( SD) treatment success rate during the pre-intervention period was 58.9  6.9% but varied considerably from year to year (range: 46% to 66%) ( Figure 2). After implementation of CB-DOTS in 2005, a slow but steady increase in the annual TSR was observed: during MTP-I it was on average 76.4  4.8 % and during MTP-II 85.3 %  1.4% (p<0.001). During the post CB-DOTS implementation period, the mean annual TSR was significantly higher than during the pre-intervention period. After the implementation of the CB-DOTS strategy, the CNR, which had been around 800/100,000 just before the intervention, started to gradually decline to 436/100,000 in 2015. A significant inverse correlation (r = -0.65, p=0.001) was found between the CNR and the TSR.
The results of the final, i.e. after correction for autocorrelation, segmented regression model of the TSR, CNR, cure and treatment completion rates for all cases with drug susceptible TB are summarized in Table 1 and Figure 3a. The model estimated TSR at the beginning of the pre-intervention period (0) at 58.0% and the CNR at 596.7/100,000. During the pre-intervention period the annual change in TSR, CNR and cure rate (1) was positive, indicating an increase in trend, which was only statistically significant for cure rate (p= 0.0172). The treatment completion rate during the pre-intervention period showed a slight, non-significant decrease. On the contrary, during the pre-intervention period, the CNR increased significantly by 23.9/100,000 cases/year. After the intervention, the treatment success and treatment completion rates (2) increased abruptly and significantly (p<0.001) by 12.9% and 24.3%, respectively, from the estimated level at the end of the pre-intervention period, e.g. from 60.9% to 68.0% for TSR (Figure 3a). In contrast, the cure rate abruptly dropped after the CB-DOTS intervention by 18.6% (p<0.001). The immediate post-intervention change in the CNR was not statistically significant (Table 1). After the intervention, the trend in the annual TSR, cure and completion rates (3) increased, but this was only statistically significant for TSR and cure rate.   After the intervention there was a significant (p<0.005) immediate increase in level and/or annual rates for treatment outcomes for pulmonary versus extrapulmonary   Table 2 shows the impact of population covariates on TSR, cure and completion rates.
During the post-intervention period, the increased national CB-DOTS and/or ART coverage significantly increased the TSR for all TB cases ( Table 2). The impact of time varying covariates on treatment, cure and treatment completion rates for all TB cases was more significant with increased CB-DOTS and ART coverage ( Table 2). HIV prevalence significantly reduced TSR, cure and completion rates among cases with DST-TB by 4.4%, 3.0% and 2.9%, respectively. The declining CNR had virtually no impact on treatment outcomes, but marginally increased the treatment completion rates among PTB and DST-TB. After the intervention the annual treatment success rate seemed to increase nonlinearly and tended towards a maximum which was estimated at 92.4% (95% CI: 87.7% -97.1% r 2 : 0.961) current interventions (Supplement C, Figure 5). However, the approach to this estimated maximum treatment rate is very slow with a 90% treatment success rate estimated to be reached in 2025.

DISCUSSION
Directly observed therapy, as recommended by the World Health Organization, is used in many countries to deliver TB treatment. 3,4,6 The effectiveness of communitybased versus facility-based (or clinic) DOTS has not been systematically assessed to date. Wright et al. performed a review and meta-analysis of 8 studies, carried out before 2015, comparing treatment outcomes of CB-DOTS versus FB-DOTS. 9 They concluded that CB-DOTS had a higher treatment success rate with a pooled odds ratio of 1.54 (95% confidence interval: 1.01 -2.36; p = 0.046). FB-DOTS was introduced in Namibia in 1991 and was universally accessible at all public health facilities in 1996, and was later expanded in 2005, to CB-DOTS. Before implementation of CB-DOTS, the annual TSR in Namibia was around 60% but showed high variability from year to year (range: 46% to 66%). During the same period, the CNR slowly increased from 652/100,000 to 822/100,000 population, which is among the highest in the world. 2 The first year after the introduction of CB-DOTS, the TSR and completion rate, but not cure rate, showed a It is clear that the community-based DOTS strategy alone will not be able to "end TB". Other factors which cannot be controlled by community-based DOTS must be explaining why the treatment success rates are stagnating around 90%. Similar studies in other countries have concluded that stagnation of treatment success rates below the 95% target may favour drug resistant TB and recommend modifications to the DOTS strategy. [16][17][18][19] In low to middle income countries like Namibia, the effectiveness of DOTS is compromised by false negative smear results, the limited monitoring of bacteriological end points, and the growing burden of drug-resistant TB. [20][21][22][23][24] Consequently, communitybased DOTS should be improved by implementing additional strategies to identify patients at risk of poor treatment outcomes to reach WHO's goal to "end TB" by 2035.
These additional community-based measures should focus on ways to improve treatment monitoring and outcomes in TB patients with co-morbidities such as HIV infection and diabetes, in childhood TB, in malnourished patients and other or mobile patient groups with an increased risk of treatment failure. 17,21,[25][26][27][28][29][30] In addition, the use of treatment completion as a surrogate measure of treatment success should be validated across all TB cases in the context of programmatic challenges. In addition, some communities/patients may require personalized rather than standardized DOTS approaches to optimise treatment outcomes.
In conclusion, the results of this study demonstrate that community-based DOTS is more effective than facility-based DOTS to increase the TB treatment success rate. In Namibia, the community-based DOTS strategy, however, was not, and will not be, able to reach the target of 95% success rate. Additional measures such as bacteriologic monitoring among patients at risk of therapeutic failure, is critical to "end TB" by 2035.
We are currently exploiting the extensive electronic TB database of the NLTP/MOHSS in an attempt to identify significant predictors of poor TB treatment outcome in Namibia.

Segmented regression model for treatment success, cure and completion rate
An interrupted time series analysis was carried out to assess the effectiveness of CB-DOTS strategy on TSR, CNR, cure and treatment completion rates for all cases.
Interrupted time series analysis is a valuable study design for evaluating the effectiveness of population-level health interventions that have been implemented at a clearly defined point in time. 15 In this design, pre-intervention regression level and trend of the outcome measure act as controls for the post-intervention segment. 15  The following segmented regression model was used to determine the level and trend changes in tuberculosis treatment success, cure and completion: Adjustment for serial autocorrelation was carried out by using the Durbin-Watson statistic and an autocorrelation parameter was included in the segmented regression model.

Facility-based and community-based DOTS in Namibia
Namibia achieved a country-wide DOTS coverage at all public health facilities, that is 42 hospitals, 34 health centers and 244 clinics by 1996 12 . Nonetheless, the geographical access to DOT was limited as many patients live too far away from clinics (upto 50 km) to come for daily clinic DOT and led to inadequate tracing of treatment interrupters. Still, there was hardly any provision of community-based DOT. 12 Furthermore, the high pill burden (i. guideline for TB treatment supporters to standardize treatment with supervision and patient support, a system for effective supply and management of TB drugs as well as a monitoring and evaluation system to for effective measurement.

Prediction of the maximum possible treatment success rate under CB-DOTS
To estimate the maximum treatment success rate that could theoretically be expected based on the observed post-intervention treatment success rates, nonlinear regression analysis was carried out using the following model predicting the maximum outcome as a function of time after the intervention: in which A is the TSR level at the intervention estimated by the segmented regression model, TSRmax is the maximum treatment effect rate, T50 is the time at which the outcome is 50% of TSRmax, and T is the time (in years) after the intervention. For all statistical tests, a p-value of 0.05 was considered to be significant.