Non-response to first-line anti-tuberculosis treatment in Sikkim, India: a risk-factor analysis study

Abstract

Setting: Sikkim, India, has the highest proportion of tuberculosis (TB) patients on first-line anti-tuberculosis regimens with the outcome ‘failure’ or ‘shifted to regimen for multidrug-resistant TB (MDR-TB)’.

Objective: To assess the factors associated with non-response to treatment, i.e., ‘failure’ or ‘shifted to MDR-TB regimen’.

Methods: We conducted a retrospective cohort study using Revised National Tuberculosis Control Programme data of all TB patients registered in 2015 for first-line TB treatment. In addition, we interviewed 42 patients who had not responded to treatment to ascertain their current status.

Results: Of 1508 patients enrolled for treatment, about 9% were classified as non-response to treatment. Patient factors associated with non-response were urban setting (adjusted odds ratio [aOR] 2.39, 95%CI 1.22–4.67), ethnicity (being an Indian tribal, aOR 1.73, 95%CI 1.17–2.57, Indian [other] aOR 1.83, 95%CI 1.29–2.60 compared to patients of Nepali origin) and those on retreatment (aOR 2.40, 95%CI 1.99–2.91). Of the patients interviewed, 28 (67%) had received treatment for drug-resistant TB.

Conclusion: In Sikkim, one in 11 patients had not responded to first-line anti-tuberculosis treatment. Host-pathogen genetics and socio-behavioural studies may be required to understand the reasons for the differences in non-response, particularly among ethnic groups.

Tuberculosis (TB) remains a major public health problem worldwide. As of 2016, the global incidence of TB was estimated to be 10.4 million cases.¹ India tops the list of the 22 high TB burden countries, with 2.8 million cases (27% of the global burden) and about half a million deaths annually.^1,2 The response to anti-tuberculosis treatment is crucial, as it not only determines the survival of the patient, it also reflects the effectiveness of the health system.³ Poor response of TB patients to first-line TB treatment indicates drug resistance, inappropriate treatment or poor treatment adherence, and it can affect TB elimination goals. In addition, it is well known that treatment outcomes are affected by factors such as age, sex, human immunodeficiency virus (HIV) status, diabetes, nutritional status and smoking.^4–14

India implemented the Revised National Tuberculosis Control Programme (RNTCP) in 1997. Sputum smear microscopy is the primary tool for TB diagnosis. In case of presumptive drug-resistant TB, sputum samples are simultaneously sent for culture and drug susceptibility testing (CDST). As it takes several days for the test results to become available, patients are initiated on first-line treatment. If resistance is diagnosed, treatment is initiated for multidrug-resistant TB (MDR-TB).

Treatment outcomes are based upon the patients' response to first-line anti-tuberculosis treatment (Table 1). At months 2 or 5 of treatment, a positive sputum smear test signifies that the patient is not responding to first-line treatment and tests for drug resistance are initiated. If the results confirm resistance to isoniazid and/or rifampicin, the patient is moved to MDR-TB treatment. Treatment outcomes of TB patients are reported 1 year after treatment initiation.

TABLE 1.

Definitions of treatment outcomes in accordance with RNTCP guidelines

According to the RNCTP's annual TB report, nearly 1% of new smear-positive patients and 5% of smear-positive retreatment patients were ‘moved to MDR-TB treatment’ in 2016. The state of Sikkim drew the attention of the TB fraternity, as respectively 16% and 23% of new and retreatment smear-positive patients moved to MDR-TB treatment (the total number of TB patients in the state was approximately 1500), which was unusually high compared to the national average. These figures are extraordinary even compared to high MDR-TB burden states such as Maharashtra, where the rates were respectively 2% and 9%,² thus leading to speculation in terms of the quality of programme implementation, the quality of drugs supplied to the region and the type of Mycobacterium tuberculosis strains circulating in the area. The foremost aim was to investigate factors associated with non-responsiveness to treatment in TB patients and their status after 1 year of treatment.

We therefore conducted this study to determine 1) the sociodemographic and clinical factors associated with non-response to first-line treatment, and 2) the status of patients who had not responded to first-line treatment 1 year after their treatment outcome was notified, to investigate whether they were evaluated for drug resistance and if they were responding to MDR-TB treatment.

METHODOLOGY

Study design

This was a retrospective cohort study using routine programme data and a sample patient survey (interviews) based on a semi-structured questionnaire to ascertain their current treatment status.

Study population

All TB patients enrolled with the Sikkim State TB programme in 2015 were included. Patient interviews were conducted for a random sample of patients identified as non-response to treatment.

Study period

The study was conducted from November 2016 to May 2017.

Study settings

The state of Sikkim is located in the hilly North-Eastern part of India, with a population of about 0.6 million spread over 7096 km². The state shares international borders with China, Nepal and Bhutan. Sikkim state is divided into four administrative districts, each with a district hospital and district TB centre (programme management unit for TB at district level). The health care services are predominantly delivered by the public health system, and there are 24 primary health centres and 147 subcentres (primary level health care), two community health centres and four district hospitals (secondary level health care) and one referral hospital.

The ethnicity of the Sikkim population is unique, with the majority being of Nepali ethnic origin. The indigenous tribes of Lepchas, Bhutias and Sherpas make up the remainder. A small proportion of the population is from other parts of India.¹⁵

Specific setting

The RNTCP was implemented in the state of Sikkim in 2002, and programmatic management of drug-resistant TB (PMDT) was initiated in 2012. Unlike the rest of India, Sikkim does not have a thriving private health sector, and the majority of the patients are therefore diagnosed and treated under the public health system only. The state has five RNTCP districts, 31 designated microscopy centres (DMCs), 598 DOT (directly observed treatment) centres, one drug-resistant TB (DR-TB) centre, one intermediate referral laboratory and 16 cartridge-based nucleic-acid amplification test (CBNAAT) machines to detect DR-TB. However, as there are no culture or DST facilities in the state, samples are sent to national referral laboratories in New Delhi and Bangalore.

The operational definition for determining the type of TB cases and treatment outcomes are described in the Tables 1 and 2, respectively. In this study, we defined ‘treatment non-response’ as those new or retreatment patients who had undergone at least 30 days of first-line treatment and whose treatment outcomes were declared to be ‘failure’ or ‘moved to MDR-TB treatment.

TABLE 2.

Case definitions in accordance with Revised National Tuberculosis Control Programme treatment guidelines

Data collection, variables and analysis

Data were collected from TB treatment registers, DMC laboratory registers and referral for treatment registers. Randomly selected patients were interviewed using the semi-structured questionnaire by trained personnel. The key variables for each patient included age, sex, ethnicity, district, type of settlement, type of TB, date of diagnosis, date of treatment initiation, treatment regimen used, number of missed doses, treatment outcomes and date of treatment outcomes.

Data were double-entered and validated using EpiData entry software v 3.1 (EpiData Association, Odense, Denmark). After validation, the data were cleaned; missing data were labelled ‘not recorded’ for each variable. The data were then analysed using Epi-Data analysis software v 2.2.2.178. The data were summarised as numbers and percentages. Logistic regression was used to obtain odds ratios (ORs) and adjusted ORs using STATA v 12 (Stata Corporation, College Station, TX, USA) to identify factors associated with treatment non-response.

To ascertain current treatment status, all patients with treatment non-response were line-listed and 42 patients were randomly selected for interview after providing informed consent. The principal investigator, who was the state programme manager for TB, and her team traced and contacted all 42 patients and interviewed them using a semi-structured questionnaire. Those patients who could not be met in person were contacted via mobile telephone. Patients' current treatment status is presented using a schematic representation (Figure 2).

FIGURE 2.

Flow diagram showing the status of a random sample of patients with non-response to first-line anti-tuberculosis treatment registered in 2015 and followed up after 2 years of treatment outcome declaration in Sikkim state. MDR-TB = multidrug-resistant tuberculosis; XDR-TB = extensively drug-resistant TB.

Ethics

Ethics approval was obtained from Sir Thodup Namgyal Memorial Hospital, Gangtok, India, and the Ethics Advisory Group of the International Union Against Tuberculosis and Lung Disease, Paris, France. Patients were interviewed after providing written informed consent, or verbal consent for those patients interviewed by telephone. No personal identifiers were used in the analysis.

RESULTS

Demographic and clinical profile of patients

In 2015, a total of 1508 TB patients were registered for treatment under the programme. Complete data were available for 1490 patients and included in the analysis. Patients' sociodemographic and clinical characteristics are shown in Table 3; 53% (793) were males and 63% (944) were from rural areas. About 80% (n = 1190) were new TB cases, 41% (n = 618) were smear-positive and 33% (n = 499) had extra-pulmonary TB; 7% (n = 109) were known diabetics and five were HIV-positive. The majority (75%) were of Nepali origin, and 18% were Indian tribal.

TABLE 3.

Sociodemographic and clinical characteristics of TB patients registered in Sikkim state, 2015 (n = 1490)

Treatment outcomes

Of the 1490 patients, complete treatment outcome data were available for 1464. Treatment outcomes, disaggregated by type of TB and case type, are presented in Table 4. Overall, 32% were cured, 52% completed treatment, 1% were lost to follow-up, 3% died, 9% switched to MDR-TB treatment and 2% failed treatment.

TABLE 4.

Treatment outcomes of tuberculosis patients enrolled for treatment under the RNTCP in Sikkim state, disaggregated by type of TB, 2015 (n = 1464) *

Patients who moved to MDR-TB treatment and those with outcome failure were considered patients with treatment non-response (n = 159). Among those who moved to MDR-TB treatment, 19 newly diagnosed patients (12%) received <30 days of first-line treatment. These were presumptive MDR-TB cases and were initiated on first-line anti-tuberculosis treatment while awaiting DST results. As the treatment duration of six patients could not be determined, these patients were excluded (n = 25), and a final 134 patients satisfied our criteria for ‘non-response’ to TB treatment (Figure 1).

FIGURE 1.

Flow chart showing identification of patients who failed to respond to first-line anti-tuberculosis treatment in Sikkim state, 2015. TB = tuberculosis.

Treatment initiation and adherence

Data on treatment initiation were available for 1405 patients. More than 80% were initiated on treatment within 1 week; the proportion was the same among the response and the non-response group. Information on treatment adherence in terms of any missed doses during the intensive phase of treatment was available for only 104 patients in the non-response group. Overall treatment adherence during the intensive phase was 98%. Most of the responders (99%) and only 74% of the non-responders had taken all the doses. As data in the non-response group were limited, we did not conduct further analyses.

Patients with non-response to first-line drugs and risk factors for non-response

Based on the study definition, patients were categorised into two groups: 1255 (cured, n = 474; treatment completed, n = 781) in the treatment response group and (failed, n = 22; ‘moved to MDR-TB treatment’, n = 111) 134 in the non-response group.

The association between various sociodemographic variables and non-response are given in Table 5. Independently associated factors were being from an urban area (aOR 2.39, 95%CI 1.22–4.67; P = 0.010), ethnicity (being an Indian tribal, aOR 1.73, 95%CI 1.17–2.57; P = 0.006; Indian [other] aOR 1.83, 95%CI 1.29–2.60; P = 0.001) and retreatment cases (aOR 2.40, 95%CI 1.99–2.91; P < 0.001).

TABLE 5.

Risk factors for non-response to first-line treatment among TB patients registered in Sikkim state, 2015

Status of patients with treatment non-response 1–2 years after treatment outcome notification

The status of 42 patients who were interviewed, disaggregated by their initial treatment outcome (‘moved to MDR-TB treatment’ or ‘failed) is shown in Figure 2. Over 80% of the patients interviewed had pulmonary TB and were newly diagnosed. Overall, 28 (67%) patients were on MDR-TB treatment or had completed treatment. Of these, 20 were from the subgroup ‘moved to MDR-TB treatment’ and eight from the subgroup ‘treatment failure’. It should be noted that 8/22 patients (36%) from the treatment failure group were moved to MDR-TB treatment after considerable first-line anti-tuberculosis treatment (>5 months).

DISCUSSION

This is one of the first studies conducted in the state of Sikkim to study risk factors for non-response to first-line treatment. Our study reports that 9% of patients had non-response to first-line treatment. The main strength of the study was the inclusion of all patients enrolled for treatment and use of routine programme records. The RNTCP supervision and monitoring systems, including periodic verification of the accuracy and consistency of the recording and reporting systems, are in place and functional in Sikkim. We therefore strongly believe that there are unlikely to be any major errors in the data, and that inferences made from these data reflect ground realities. The main limitation of the study is that most of our study variables on patient factors were restricted to those that are routinely collected and documented in programme records. As several other patient factors or health system factors can affect study outcomes, such as socio-economic status, baseline drug resistance status, availability and functional status of culture and CDST services, etc., we are unable to account for these factors in our analysis and interpretation of the data. Another study limitation was the inability to reach all patients due to difficult terrain, communication issues and lack of adequate transport in the state.

Despite the above limitations, the study findings have the following implications on policies, practices and future research.

First, about 12% of new/retreatment patients who were moved to MDR-TB treatment within the first month were initiated on first-line treatment while waiting for their CDST results. As the state lacks CDST facilities, sending the samples to national laboratories delays the results. The RNTCP in Sikkim must ensure that access to CDST facilities is improved, either by identifying national centres that are closer to Sikkim State or by developing these facilities within the state so that patients are promptly initiated on correct anti-tuberculosis regimens. However, we appreciate the current practice of enrolling all patients who are initiated on first-line treatment in the TB register, which indicates good accountability. This documentation also diminishes the risk of pre-treatment loss to follow-up.¹⁶

Second, treatment non-response could be due to primary drug resistance or development of DR-TB during first-line treatment due to poor treatment adherence. Based on the limited adherence data that we collected (which showed good adherence), we are inclined to suggest primary drug resistance as the most likely reason for non-response. As the RNTCP currently advocates universal DST before the initiation of anti-tuberculosis treatment,¹⁶ the Sikkim RNTCP unit has begun to implement this policy in the state using Xpert^® MTB/RIF (Cepheid, Sunnyvale, CA, USA) technology.

Third, factors associated with non-response included being from an urban area, retreatment TB and ethnicity. While the association between the first two factors is widely known,^4–14 the association between ethnicity (Indian tribal vs. patients of Nepali origin) and non-response is new. Ethnicity is known to cause variations in susceptibility to tuberculous infection and progression to TB disease due to differences in sociobehavioural characteristics and/or genetic make-up.^17–19 However, whether ethnicity can influence non-response to treatment and affect the pathway to non-response is unknown. Future research into this area is required to untangle the mystery of the very high non-response rate in Sikkim. Various host-pathogen genetic aspects are also likely to cause drug resistance, disease susceptibility or progression to active disease.^20–22 The Indian Council of Medical Research (New Delhi, India), the leading government research organisation, and the RNTCP are planning to undertake a genetic study to identify to identify host-pathogen genetic factors associated with TB/MDR-TB in Sikkim.²³ While we welcome this approach, we believe that it is also essential to undertake anthropological studies to assess sociobehavioural differences among the various ethnic groups that could be associated with non-response to treatment.

Finally, the findings from interviews with patients with treatment non-response suggest that the majority were on or had completed treatment for DR-TB. This to a large extent allayed our fears about the status of these patients as to whether or not they had received or were receiving appropriate TB care.

CONCLUSION

In Sikkim, one in 11 patients had not responded to first-line anti-tuberculosis treatment. Ethnic origins (Indian tribal and Indian other) were associated with treatment non-response. Further genetic studies are needed to ascertain the relationship between ethnicity and non-response. CDST facilities should be established within the state for early diagnosis of drug-resistant TB.