Management of drug-resistant tuberculosis in Indonesia: a four-year cascade of care analysis

Summary

Background

In Indonesia, drug resistance testing for TB largely relies on Xpert MTB/RIF, and it is unknown what proportion of drug-resistant (DR) TB is adequately diagnosed and treated.

Methods

We conducted a cascade of care analysis on a cohort of presumptive rifampicin-resistant (RR) TB patients registered in 2015–2018 in a tertiary hospital in Indonesia. Estimated incidences of (presumptive) DR-TB cases were assumption-based using global reports. Data on diagnosis and consecutive cascades steps, including their timing were collected from national electronic registers, and medical records. We described a secondary cascade for patients receiving treatment not supported by phenotypic drug susceptibility testing (pDST). Factors associated with delay and loss between diagnosis and treatment were identified using logistic regression.

Findings

Less than a third of estimated incident TB cases at risk of DR-TB were identified as presumptive DR-TB case and tested, and 9.8% (982/10,065) of estimated true DR-TB cases was diagnosed. Of those diagnosed, only 45.1% (443/982) had treatment regimens supported by pDST results, but this did not significantly influence treatment outcomes. Only 25.5% (250/982) of diagnosed patients completed all steps of the cascade including successful treatment. Delays between diagnosis and treatment were substantial, and more common among those referred from a primary healthcare facility, and among those who were employed, living outside of Bandung, and reporting engagement with the private sector.

Interpretation

The DR-TB care cascade in this urban setting in Indonesia is characterized by substantial attrition and delays. Strategies to increase access to DR-TB diagnosis accompanied by optimisation of clinical care could substantially improve outcomes and reduce onward transmission.

Funding

Radboud university medical center and University of Otago.

Research in context.

Evidence before this study

Undiagnosed and untreated drug-resistant (DR-)TB forms a threat for ending TB globally. Cascade of care analyses identify the extent of diagnostic and management gaps. We searched PubMed and Google Scholar for publications using synonyms of “cascade of care”, “pathways”, “gaps”, and “delays”, combined with “drug-resistant tuberculosis”. Previous studies reported substantial attrition and delays in the DR-TB cascade. However, comprehensive insight for Indonesia is lacking. Furthermore, no study evaluated outcomes of those who drop off the main cascade.

Added value of this study

We found that nine out of ten estimated incident DR-TB cases remained untested and undiagnosed between 2015 and 2018, and that treatment success rates were low in this urban area in Indonesia. We also identified considerable delays between diagnosis and treatment. Many patients did not have phenotypic drug susceptibility testing (pDST) results, but their treatment outcomes were not significantly different compared to patients with pDST results.

Implications of all the available evidence

Rapid point-of-care DST, active case finding and interventions to link detected cases to subsequent cascade steps could improve patient outcomes and reduce onward transmission.

Introduction

Globally, there were an estimated 450 thousand new cases of multi-drug resistant (MDR) or rifampicin-resistant TB in 2021, of which only 37% were laboratory-confirmed and notified,¹ and of those only 60% were successfully treated.¹ Indonesia is the second biggest contributor (13%) to the global gap between estimated incidence and reported number of TB cases, and among the top 10 countries for the number of incident MDR/RR-TB cases that remain untreated.¹ As MDR-TB in high-burden settings is predominantly caused by direct transmission,² undiagnosed and untreated patients are likely to contribute substantially to ongoing transmission.

In order to close gaps in diagnosis and management of DR-TB, we need better understanding of the care cascade of drug-resistant (DR) TB. The ‘cascade of care’ is a framework that identifies and quantifies gaps and delays between sequential steps of a patient’s pathway. Defining these gaps can facilitate more efficient targeting of resources, to improve care continuity, successful management and reduce ongoing transmission. Previous studies have reported substantial gaps in the cascade of care for DR-TB,3, 4, 5, 6, 7, 8 but the entire care cascade from estimated disease burden until treatment completion remains undefined for Indonesia, which has the second highest TB burden globally and a broad range of health system challenges.^1,9 Furthermore, to our knowledge, no previous study has described the course of patients who drop off the main cascade, such as those who do not have a result from phenotypic drug susceptibility testing. Therefore, this multi-year study, performed in a large tertiary hospital in Indonesia, aimed to describe the full cascade of care for RR-TB, identify factors associated with gaps and delays across the care cascade, and define a secondary care cascade for those who received treatment not supported by pDST results.

Methods

Study design and population

We conducted a cohort study in the DR-TB clinic at Hasan Sadikin General Hospital (RSHS), Bandung, Indonesia. This provincial-level tertiary hospital was the main referral site for DR-TB cases in West Java province (population: 48 million) at the time of data collection. In 2015, RSHS was one of two Programmatic Management of DR-TB (PMDT) sites in West Java, but until 2018 this had grown to six PMDT sites. GeneXpert MTB/RIF has been implemented as a front-line test for presumptive DR-TB in Indonesia since 2012, but mostly centralized until mid-2017, after which it became increasingly available in rural areas and primary care centres. Since 2021, the use of GeneXpert MTB/RIF has been expanded to all presumptive TB patients. Additionally, two laboratories perform culture-based DST in West Java, both of which are located in Bandung. We included all patients who met one or more of the high-risk criteria for identification as a presumptive DR-TB case, according to the Indonesian PMDT guidelines (Panel 1),¹⁰ and who were registered and/or started treatment between January 2015 and December 2018 at RSHS. Patients were excluded if they lived outside of West Java.

Panel 1. High-risk criteria for identification as a presumptive DR-TB case.

According to the Indonesian guidelines for programmatic management of drug-resistant TB at the time of the study, a presumptive drug-resistant TB case is a patient with symptoms of TB and one or more of the following criteria:

Risk category		Definition
1	Treatment failure of first-line second category anti-TB drugs	Patients who are still smear-positive at the end of treatment with first-line second-category anti-TB drugs (i.e. RHZES).
2	No conversion of first line second category anti-TB drugs	Patients who are smear-positive after three months of treatment with first-line second-category anti-TB drugs (i.e. RHZES).
3	TB patients with a history of non-standard TB treatment	Patients who received any type of TB treatment outside of the national program, e.g. non-DOTS or private clinics.
4	Treatment failure of first-line first category anti-TB drugs	Patients who are still smear-positive at the end of treatment with first-line first category anti-TB drugs (i.e. RHZE).
5	No conversion of first line first category anti-TB drugs	Patients who are smear-positive after three months of treatment with first-line first category anti-TB drugs (i.e. RHZE).
6	Relapse	Patients whose most recent TB treatment outcome was ‘cured’ or ‘treatment completed’ and who return with symptoms of TB.
7	Return after loss to follow-up	Patients who interrupted any type of TB treatment for two or more consecutive months and return with symptoms of TB.
8	Contact with DR-TB patient	Suspected TB patient with a history of close contact with an MDR-TB patient.
9	TB-HIV non-responsive to TB treatment	TB-HIV co-infected patients who do not respond clinically or bacteriologically to TB treatment.

TB = tuberculosis; R = rifampicin; H = isoniazid; E = ethambutol; S = Streptomycin; DOTS = Directly Observed Treatment Strategy; MDR-TB = multi-drug resistant TB; HIV = human immunodeficiency virus.

Clinical procedures

Presumptive DR-TB cases were tested with Xpert MTB/RIF, following national PMDT guidelines.¹⁰ All those with confirmed rifampicin resistance underwent baseline examinations (e.g. physical exam, blood tests, chest X-ray, and audiometric tests) and were treated empirically with a standard MDR-TB regimen while awaiting baseline culture and pDST results (Supplementary Material S1).¹⁰ Originally, standard MDR-TB treatment consisted of an injectable long-term regimen (LTR). After 2017, this was changed to injectable short-term regimens (STR), with 4–6 months kanamycin, moxifloxacin, ethionamide, high-dose isoniazid, clofazimine, ethambutol and pyrazinamide, followed by 5 months of moxifloxacin, clofazimine, ethambutol and pyrazinamide.¹¹ However, LTR was still recommended for some patients, including those with resistance to fluoroquinolones or second-line injectables.¹¹

Culture was performed using Solid Lowenstein Jensen (LJ) before 2018, and Mycobacterium Growth Indicator Tube (MGIT) thereafter.¹² Culture-based DST was conducted for first-line drugs (rifampicin, isoniazid, ethambutol, and streptomycin) and selected second-line drugs (kanamycin, amikacin, ofloxacin).¹⁰ If pDST results indicated additional resistance, treatment regimens were adjusted accordingly, following the national PMDT guidelines.¹⁰ Treatment response was monitored by regular smear and culture until treatment completion or any other outcome. Since treatment durations were up to two years, final treatment outcomes were recorded until 2020. Successful treatment outcome was defined based on culture as cured or treatment completed, whereas unsuccessful outcomes included death, treatment failure, or loss to follow-up.^10,13 Failure of the intensive phase was defined as culture positivity at the end of intensive phase.¹⁰

Data sources for care cascades

We defined a pre-diagnosis care cascade for presumptive DR-TB in the hospital’s catchment area, based on both estimated and collected data, as: (1) the estimated number of incident TB cases with a risk factor for DR-TB; (2) the number of presumptive DR-TB cases identified; and (3) the number of presumptive DR-TB cases with a sample collected for GeneXpert MTB/RIF testing. To estimate the number of TB cases with risk factors of DR-TB, we used retreatment TB cases as a proxy. Approximately 16% of pulmonary TB patients are estimated to have a history of TB treatment.¹⁴ These patients should be identified as presumptive DR-TB case by the healthcare system. The approach to this estimation including assumptions is described in Supplementary Material S2. Data regarding identification of presumptive DR-TB cases and testing with GeneXpert MTB/RIF were collected from the national electronic program database system (e-TB manager), laboratory registers, and the medical records.

A similar pre-diagnosis care cascade was defined for true DR-TB, also with both estimated and collected data, as: (1) the estimated number of incident true DR-TB cases; and (2) the number of newly diagnosed RR-TB cases. To estimate the true number of DR-TB cases, we added an estimated number of pulmonary TB cases not previously treated (84%) to retreatment cases from the presumptive cascade, and we calculated the estimated number of MDR-/RR-TB among these retreatment and new pulmonary TB cases (13–16% and 2.4–2.8%, respectively, based on WHO global TB reports 2016–2019). Calculations are provided in Supplementary Material S3. Finally, data regarding the number of newly diagnosed RR-TB cases in the hospital were collected from the national electronic program database system (e-TB manager), laboratory registers, and the medical records.

We then established a clinical care cascade of diagnosed RR-TB cases based on observed data with the following steps: RR-TB diagnosis, baseline clinical assessment, initiation of empirical MDR-TB treatment, support of treatment regimen by pDST results, intensive phase treatment completion, and successful treatment outcome. Data about the number of individuals completing each step and time between the steps were collected from the national electronic program database system (e-TB manager), laboratory registers, and medical records. Due to decentralization of GeneXpert MTB/RIF in 2017, some individuals already had a GeneXpert MTB/RIF diagnosis of RR-TB before admission at RSHS referral hospital. For those, time from sample collection to Xpert results was unknown, but assumed to be within one day. Baseline clinical assessment in our analysis was defined as availability of blood tests and physical examination in the medical records.

Finally, patients whose treatment was not supported by pDST results, but who continued to be treated in RSHS were represented in a secondary care cascade that includes the steps of intensive phase treatment completion and successful treatment outcome.

Statistical analysis

Primary cascade data were analysed using R. Cascade indicators and corresponding gaps were summarized using proportions (%). Time delays between cascade steps were summarized as median time (IQR). Furthermore, to dichotomize patients in “timely” vs “delayed” steps, we used indicator times from national guidelines^12,15 and expert consensus, i.e. one day for sample collection after referral, and subsequently two days for Xpert results, four days for baseline clinical assessment, three days for initiation of standard regimen, and 84 days for pDST results. Next, potential factors associated with delay in the steps from diagnosis to treatment were pre-selected using literature.^16,17 Their association with delay and loss, as a proxy for extreme delay, to the next step were evaluated using univariable logistic regression. We acknowledge other factors, such as income and education, may be important factors associated with delay, but data regarding those factors was limited in our setting. After checking for multicollinearity, all variables were added in a multivariable logistic regression, as all were deemed clinically and scientifically relevant. Results are presented as crude and adjusted odds ratios (OR) with 95% confidence intervals (CI), and a p-value < 0.05 was considered statistically significant. Treatment outcomes for the secondary cascade were represented in a Sankey diagram using the web-based SankeyMATIC tool.¹⁸ Chi-square test was used to compare treatment outcomes for those treated with and without support of pDST testing.

Ethical considerations

Ethical permission for this study was provided by the Health Research Ethics Committee at the Faculty of Medicine, Universitas Padjadjaran (035/UN6.KEP/EC/2022). Data from registers and medical records were de-identified and anonymized.

Role of the funding source

The funders of this project had no role in study design, data collection, analysis, interpretation, or writing of the report. The corresponding author had full access to all data and the final responsibility to submit for publication.

Results

Study population

Over a period of four years, 9663 patients presented with presumptive RR-TB at RSHS referral hospital, and 982 were newly diagnosed as RR-TB using Xpert MTB/RIF (Supplementary Material S4). Of these, about half were male, with a median age of 38 years (Table 1). A majority (84.2%) was previously treated for TB, and most Xpert MTB/RIF tests were conducted to evaluate the possibility of relapse (38.1%) or treatment failure with first-line drugs (36.3%). Baseline smear positivity was 73.5%, and for the 791 (80.5%) who had a culture done, 64.6% (n = 511) had a pDST result. Of these, the majority (400/511, 78.3%) had RR-/MDR-TB, and some had further second line drug resistance (72/511, 14.1%) or were found not resistant to rifampicin in culture (39/511, 7.6%).

Table 1.

Characteristics of newly diagnosed RR-TB cases at RSHS (n = 982).

	n (%) or median (IQR)
Age	38 (28–48)
Gender, male	510 (51.9)
Residence (n = 976), outside Bandung	567 (58.1)
Socioeconomic information
Employment Status (n = 736), employed	386 (52.4)
Monthly income (n = 267)
No income	140 (52.4)
<1 million Indonesian Rupiah (IDR, $67)	28 (10.5)
1–2 million IDR ($67–134)	63 (23.6)
>2 million IDR ($134)	36 (13.5)
Comorbidities and medical history
BMI (n = 711)
Underweight (<18.5)	432 (60.8)
Normal (18.5–22.9)	212 (29.8)
Overweight (23–24.9)	32 (4.5)
Obese (≥25)	35 (4.9)
HIV status
HIV positive	20 (2.0)
HIV negative	893 (90.9)
Not tested	69 (7.0)
Self-reported diabetes mellitus (n = 911)	111 (12.2)
Random blood glucose (n = 707), elevated (>200 mg/dL)	82 (11.6)
History of TB treatment
None	155 (15.8)
Once	405 (41.2)
Twice	294 (29.9)
More than twice	128 (13.0)
For those with history of TB treatment (n = 827): Outcome of most recent TB treatment
Cured	208 (25.2)
Completed	47 (5.7)
Failed	327 (39.5)
Lost to follow-up	93 (11.2)
Unknown	152 (18.4)
Risk factors for MDR-TBa
Treatment failure of first-line first category anti TB drugs (i.e. RHZE)	202 (20.6)
Treatment failure of first-line second category anti TB drugs (i.e. RHZES)	154 (15.7)
No conversion of first category anti TB drugs	3 (0.3)
No conversion of second category anti TB drugs	7 (0.7)
Return after loss to follow up	111 (11.3)
Relapse	374 (38.1)
Contact with DR-TB patient	9 (0.9)
Unstandardized TB treatment using second-line anti TB drugs	8 (0.8)
TB-HIV non-responsive to TB treatment	2 (0.2)
Routine screening for presumptive TB	112 (11.4)
Care-seeking prior to inclusion
History of visiting private provider for TB treatment (n = 835)	179 (21.4)
Source of referral (n = 962)
Primary health care level	427 (44.4)
Secondary health care level	441 (45.8)
Tertiary health care level	94 (9.8)
Diagnosis and treatment at inclusion
Baseline GeneXpert, rifampicin resistant Mtb detected	982 (100.0)
Baseline smear (n = 790), positive	581 (73.5)
Baseline culture (n = 791)
Positive for Mycobacterium tuberculosis	609 (77.0)
Sterile culture	144 (18.2)
Nontuberculous Mycobacteria (NTM)	20 (2.5)
Contaminated	18 (2.3)
Baseline phenotypic DST (n = 511)
MDR-TB	328 (64.2)
Rifampicin mono-resistance	72 (14.1)
Pre-XDR TB: resistance to fluoroquinolones	49 (9.6)
Pre-XDR TB: resistance to second-line injectables	10 (2.0)
XDR TB	13 (2.5)
Other (rifampicin sensitive)	39 (7.6)
Second-line regimen (n = 807)
Injectable long-term regimen	626 (77.6)
Short term regimen	158 (19.6)
Bedaquiline-based regimen	21 (2.6)
Individualized regimen	2 (0.2)

RSHS = Rumah Sakit Hasan Sadikin (Hasan Sadikin hospital); USD = United States dollar; BMI = body mass index; HIV = human immunodeficiency virus; MDR-TB = multi-drug resistant TB; RHZE(S) = rifampicin, isoniazid, pyrazinamide, ethambutol (streptomycin); TB = tuberculosis; DST = drug-susceptibility testing; XDR-TB = extensively drug-resistant TB.

^aUsed to screen for who is tested with Xpert/MTB-RIF.

Primary care cascade

With regards to presumptive DR-TB, we estimated that there were almost 35 thousand TB cases with risk factors for DR-TB (Fig. 1A, Supplementary Material S2), but only 9663 (28%) were identified as person with presumed DR-TB and seen at RSHS. With regards to true DR-TB, of the estimated ten thousand true DR-TB cases in the hospital’s catchment area (Fig. 1B, Supplementary Material S3), only 9.8% (n = 982) were confirmed as new RR-TB patients through GeneXpert MTB/RIF at RSHS.

Fig. 1.

Pre-diagnosis cascade of care describing the presumptive DR-TB cases and true DR-TB burden in the catchment area of RSHS. Bars represent absolute frequencies of individuals completing subsequent steps of the care cascade preceding diagnosis of DR-TB, including suspected (A) and true (B) DR-TB cases. (A) Presents the estimated number of people with risk factors for DR-TB cases in the catchment area of the hospital (i.e. TB cases with a history of TB treatment), the number of presumptive DR-TB cases that were registered at RSHS, and the number of presumptive DR-TB cases tested with GeneXpert MTB/RIF at RSHS. (B) Presents the estimated true number of DR-TB cases, separated for those with and without a history of TB treatment, in the catchment area of the hospital, and the number of newly diagnosed RR-TB cases in the hospital. RSHS = Rumah Sakit Hasan Sadikin (Hasan Sadikin hospital); DR-TB = drug-resistant TB; TB = tuberculosis.

Gaps were also found after Xpert-diagnosis of RR, as only 56.0% (443/791) of patients with completed baseline clinical assessment who initiated standard MDR-TB treatment had a regimen supported by pDST results (Fig. 2, Supplementary Material S5). pDST was conducted in only 56.8% (449/791) of all patients who initiated standard MDR-TB treatment. As expected, for almost all in this group (96.4%), pDST results became available after treatment initiation. For those without pDST, culture was not performed (n = 102), negative (n = 122), positive for Mtb but not followed by pDST (n = 84), positive for nontuberculous mycobacteria (n = 18), or contaminated (n = 16). Among those with pDST results, 98.7% (443/449) had treatment regimens concordant with pDST results. These were mostly culture-confirmed RR-/MDR-TB patients receiving MDR-TB treatment (353, 81.5%), and pre-XDR or XDR-TB patients whose regimen was adjusted after pDST (20, 4.6%).

Fig. 2.

Post-diagnosis cascade of care of newly diagnosed RR-TB cases in RSHS. Bars and lines represent absolute and relative frequencies, respectively, of individuals completing subsequent steps of the care cascade after diagnosis. Of those diagnosed, 90.9% (95% CI 89.0–92.6) had baseline clinical assessment, 80.5% (95% CI 78.0–82.9) initiated treatment, 45.1% (95% CI 42.0–48.2) had treatment regimens supported by pDST results, 31.0% (95% CI 28.1–33.9) completed intensive phase, and 25.5% (95% CI 22.8–28.3) completed treatment successfully. (a) Median time between sample collection and Xpert MTB/Rif results was 0 days (IQR 0–1), with dichotomous delay defined as >2 days. (b) Median time between Xpert MTB/Rif result and completion of baseline clinical assessment was 11 days (IQR 5–20), with dichotomous delay defined as >4 days. (c) Median time between completion of baseline clinical assessment and initiation of standard MDR-TB regimen was 8 days (IQR 6–11), with dichotomous delay defined as time >3 days. (d) Median time between initiation of standard MDR-TB regimen and results of concordant DST results was 66 days (IQR 22–91), with dichotomous delay defined as >84 days. RR-TB = rifampicin-resistant TB; TB = tuberculosis; RSHS = Rumah Sakit Hasan Sadikin (Hasan Sadikin hospital); pDST = phenotypic drug-susceptibility testing.

Furthermore, treatment success rates were low, and only 25.5% of new GeneXpert MTB/RIF diagnosed RR-TB patients completed the entire cascade successfully (Fig. 2, Supplementary Material S5). Of 443 patients who completed baseline clinical assessment and who were treated with a regimen supported by pDST results, 56 (12.6%) died in the intensive phase, and 16 (3.6%) died in the continuation phase (Fig. 3). Just over half (250/443, 56.4%) finished their treatment successfully.

Fig. 3.

Treatment outcomes in primary and secondary care cascade. The bottom green branch represents the primary care cascade, with patients whose treatment regimens are supported by pDST results (n = 443), completing the intensive phase (n = 304) and having a successful treatment outcome (n = 250). The middle and top branch represent the secondary care cascade, including both patients without pDST results (n = 342), or with treatment regimens that are discordant with pDST results (n = 6). MDR-TB = multi-drug resistant TB; TB = tuberculosis; (p)DST = (phenotypic) drug-susceptibility testing.

Secondary cascade of care

Phenotypic DST was only available for 56.8% (449/791) of treated patients, but this did not seem to affect treatment outcome. Of 791 patients who completed baseline clinical assessment and initiated treatment for RR-TB, 342 (43.2%) did not have pDST results, and 6 had treatment regimens that were discordant with pDST results. Hence, these patients were lost in the primary cascade, but represented in a secondary cascade (Fig. 3). Of those initiating treatment without pDST, 56.7% (194/342) finished their entire treatment successfully and 24.0% (82/342) died. There was no significant difference in the number of deaths or failures in patients without pDST results compared to those with treatment regimens supported by pDST (27.5% vs 24.8%, p = 0.413). Furthermore, there was no significant difference in the number of successful treatment outcomes for both groups (56.7% vs 56.4%, p = 0.942). It should be noted that, 14.3% (64/449) had rifampicin monoresistance and 64.4% (289/449) had MDR-TB, which are treated appropriately with the empiric regimen.

Delays from diagnosis to treatment initiation

There were substantial delays in the steps from diagnosis to treatment initiation (Fig. 2, Supplementary Material S5). Of those receiving baseline clinical assessment after diagnosis, only 21.9% (195/891) did so within the targeted four days. Furthermore, only 20.5% (162/791) of those initiating MDR-TB treatment, did so within the targeted three days. In multivariable analysis, the chance of delay or loss, as a proxy for extreme delay, in baseline clinical assessment was smaller for patients referred from secondary or tertiary level healthcare facilities (aOR 0.55, 95% CI 0.34–0.89) (Table 2). Furthermore, the chance of delay or loss in treatment initiation was higher for patients who lived outside of Bandung (aOR 1.79, 95% CI 1.15–2.77), who were employed (aOR 1.88, 95% CI 1.20–2.93), or who have previously sought care at private providers (aOR 2.43, 95% CI 1.25–4.69).

Table 2.

Factors associated with delay and loss in baseline clinical assessment and treatment initiation.

Characteristic	Delayed baseline clinical assessmenta				Delayed treatment initiationb
	Crude OR [95% CI]	p value	Adjusted OR [95% CI]	p value	Crude OR [95% CI]	p value	Adjusted OR [95% CI]	p value
Age (years)	1.01 [1.00–1.02]	0.03	1.00 [0.99–1.02]	0.61	1.01 [1.00–1.02]	0.09	1.01 [0.99–1.03]	0.25
Gender		0.98		0.90		0.05		0.14
Female	Ref		Ref		Ref		Ref
Male	1.00 [0.73–1.36]		1.03 [0.65–1.63]		1.41 [1.00–1.99]		1.40 [0.90–2.17]
HIV		0.66		0.61		0.48		0.30
Negative	Ref		Ref		Ref		Ref
Positive	0.80 [0.29–2.22]		0.70 [0.18–2.77]		0.66 [0.21–2.09]		0.49 [0.13–1.87]
Residence		0.73		0.47		<0.001		<0.01
Inside Bandung	Ref		Ref		Ref		Ref
Outside Bandung	1.06 [0.77–1.46]		1.18 [0.75–1.87]		2.00 [1.42–2.82]		1.79 [1.15–2.77]
Employment		0.45		0.36		<0.001		<0.01
Unemployed	Ref		Ref		Ref		Ref
Employed	0.87 [0.60–1.25]		0.81 [0.51–1.28]		2.18 [1.45–3.28]		1.88 [1.20–2.93]
History of visiting a private provider for TB treatment		0.57		0.84		<0.01		<0.01
No	Ref		Ref		Ref		Ref
Yes	0.88 [0.57–1.36]		1.06 [0.60–1.86]		2.21 [1.32–3.69]		2.43 [1.25–4.69]
History of TB treatment		<0.001		0.45		<0.001		0.55
None	Ref		Ref		Ref		Ref
Once or more	2.75 [1.89–4.01]		1.95 [0.34–11.21]		0.28 [0.14–0.57]		0.51 [0.06–4.60]
Source of referral		0.05		0.02		<0.001		0.05
Primary level	Ref		Ref		Ref		Ref
Secondary or tertiary level	0.72 [0.52–1.00]		0.55 [0.34–0.89]		0.54 [0.38–0.77]		0.64 [0.40–1.01]

Univariable and multivariable logistic regression to identify factors associated with delay and loss.

OR = odds ratio; CI = confidence interval; HIV = human immunodeficiency virus; TB = tuberculosis; Ref = reference.

^aDelay, defined as >4 days, or loss between diagnosis and baseline clinical assessment (785/980).

^bDelay, defined as >3 days, or loss between baseline clinical assessment and treatment initiation (729/891).

Discussion

In this urban setting in Indonesia, we found significant attrition and delays along the DR-TB care cascade. First, only about a third of estimated TB cases at risk of DR-TB were identified and reported, and only a tenth of estimated true DR-TB cases were diagnosed. Secondly, only about half of the treatment regimens were supported by pDST. Thirdly, almost half of all patients initiating treatment had an unsuccessful outcome. Finally, delays between diagnosis and treatment were substantial, especially for those living further away, those with employment, and those with a history of private sector engagement. These results provide key information to guide expansion and optimisation of care for DR-TB patients in Indonesia.

Previous studies have reported widely varying estimates of the proportion of DR-TB detected, ranging from 1.8%–100%.^4,5,7,8 This may be due to different methodology, study setting, programmatic conditions, and mode of diagnosis. Commonly reported barriers to diagnosis include patients’ willingness and/or ability to access diagnostic tests, self-cured or subclinical TB, lack of human and laboratory resources, laboratory operational challenges, and healthcare providers’ not testing for TB or drug resistance.^19,20 Undetected DR-TB hampers global ending of TB, as it leads to onward transmission of DR-TB. At the same time, if more persons with presumed DR-TB present for testing and treatment, health systems could become overburdened, which might cause an exacerbation of issues further on in the cascade. In a nationwide scale-up of GeneXpert MTB/RIF in South-Africa, approximately all estimated MDR-/RR-TB cases were diagnosed, but still only 20% were successfully treated.⁵

Phenotypic DST results were only available for a minority of patients. A meta-analysis has shown that treatment outcomes are better when regimens are guided by DST results,²¹ but in this cohort the absence of pDST results was not associated with death or treatment failure, noting that over three quarters of Xpert RR was due to rifampicin monoresistance or MDR-TB, for which the empiric regimen is appropriate. Other factors contributing to a lack of a difference in outcome include: the fact that pDST results often take months to be available, leading to delayed treatment modifications and physicians relying on their clinical expertise to make initial decisions; to issues with reproducibility and diagnostic accuracy of pDST for certain TB drugs²²; to the fact that pDST was only used for a limited number of second-line drugs during the time of this study; or a selection bias, with a higher likelihood of pDST testing in the most severely ill patients.

Only 57% of patients who initiated treatment also successfully completed it, in line with the global treatment success for DR-TB of 59%.²³ In literature, success among treated DR-TB patients is highly variable, ranging from 44% to 77%.3, 4, 5, 6, 7 Adverse drug reactions, lack of financial and/or social support, baseline comorbidities with increased pill-burden, early discontinuation once symptoms resolve, large distance to healthcare facilities and other factors may hinder treatment success of MDR treatment.^24,25 However, a recent German study suggest that long-term successful treatment outcome may be underestimated using current WHO definitions, as many patients currently classified as treatment failure have long-term relapse-free cure.²⁶ Early losses in the cascade were more common among those with a history of engagement with private practitioners who show lower adherence to national and international guidelines.²⁵ Finally, those referred from higher level healthcare facilities had better outcomes, where one-stop services are more common.

Since the time of this study, DR-TB case management has made progress through scale-up of GeneXpert MTB/RIF and new and shorter treatment regimens. However, the COVID-19 pandemic has jeopardized advances in DR-TB services and ending TB.¹ Implementation of GeneXpert MTB/RIF has improved RR-TB case detection.⁵ It will be important to monitor improvements in long-term patient outcomes,²⁷ given the likely health system limitations in linking diagnosed patients to subsequent steps in the care cascade.²⁸ If implemented sustainably and in the context of a national TB programme, active case finding strategies are cost-effective interventions that aid DR-TB case detection, especially for new cases.²⁹ Furthermore, an improved rapid, comprehensive, point of care second-line DST could enhance equitable access to resistance profiling and could aid regimen selection.³⁰ Implementation of such a tool should be accompanied by efforts to strengthen treatment capacity, to improve linkage of diagnosed patients to effective treatment for instance through a community officer or caregiver, and to improve financial, social and clinical support during treatment.

To our knowledge, this study is the first to evaluate the entire care cascade from the estimated number of DR-TB patients to treatment outcome in Indonesia. We also developed a secondary care cascade that evaluates the impact of pDST on care continuity and patient outcomes. A limitation of this study is the uncertainty regarding the estimate of DR-TB cases in the catchment area, since national MDR-TB prevalence surveys with more precise estimates were not available. We made assumptions that the incidence of TB in West Java is similar to overall national estimates, and that incident DR-TB cases were divided equally among PMDT sites. However, our estimates may be on the conservative side. For example, if incident DR-TB cases were divided proportional to cases treated in each PMDT in West Java, the estimated incidence of true DR-TB in our hospital’s catchment area would go up to 17,000 cases. Furthermore, with regards to estimated presumptive DR-TB, patients without a history of TB treatment could also be at risk of DR-TB, but aren’t included in the estimation. On the other hand, patients in the catchment area may have been transferred to other areas and hospitals. As such, estimations can vary based on the assumptions made, so these results ought to be interpreted with caution. Finally, this study was performed in a tertiary hospital with better resources compared to district hospitals. Therefore, gaps and delays identified in this study may be more severe in other PMDT sites and areas of Indonesia.

In this urban setting in Indonesia, significant losses and delays of DR-TB patients were found in case finding, pDST testing and treatment outcomes, but treatment initiation was relatively high. Strategies to improve access to DR-TB diagnostics, such as active case finding and improved peripheral and rapid DST, accompanied by efforts to link these patients to subsequent steps of the care cascade, could improve case finding, initiation of appropriate treatment, patient outcomes, and reduce onward transmission.

Contributors

BWL, AL, PS, BA, LC, BA, AYS, RvC, and PCH designed the study. AL, GN and BWL contributed to data collection and verification. AYS, BA, and PS supported data collection through patient management. BWL, GN and AL did the analysis and had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. GN, AL and BWL drafted the manuscript. All authors critically revised the manuscript for important intellectual content and all authors gave final approval for the version to be published.

Data sharing statement

Deidentified participant data including the study protocol and data dictionary will be made available to others upon written requests to the corresponding author immediately after publication and subject to a written proposal with detailed description of study objectives, data analysis plan and a signed data sharing agreement.

Declaration of interests

We declare no competing interests.