Are we missing ‘previously treated’ smear-positive pulmonary tuberculosis under programme settings in India? A cross-sectional study

Background: In 2007, a field observation from India reported 11% misclassification among ‘new’ patients registered under the revised national tuberculosis (TB) control programme. Ten years down the line, it is important to know what proportion of newly registered patients has a past history of TB treatment for at least one month (henceforth called ‘misclassification’). Methods: A study was conducted among new smear-positive pulmonary TB patients registered between March 2016 and February 2017 in 18 randomly selected districts to determine the effectiveness of an active case-finding strategy in marginalised and vulnerable populations. We included all patients detected through active case-finding. An equal number of randomly selected patients registered through passive case-finding from marginalised and vulnerable populations in the same districts were included. Before enrolment, we enquired about any history of previous TB treatment through interviews. Results: Of 629 patients, we interviewed 521, of whom, 11% (n=56) had past history of TB treatment (public or private) for at least a month: 13% (34/268) among the active case-finding group and 9% (22/253) among the passive case-finding group (p=0.18). No factors were found to be significantly associated with misclassification. Conclusion: Around one in every ten patients registered as ‘new’ had previous history of TB treatment. Corrective measures need to be implemented, followed by monitoring of any change in the proportion of ‘previously treated’ patients among all registered patients treated under the programme at national level.


Amendments from Version 1
Based on the reviewer comments, the following key edits were made to the manuscript: 1. Definition of 'Misclassification' was introduced early in the manuscript (both in Abstract as well as in the main text) 2. In the Discussion section, under implications for the TB programme, we added point regarding how misclassification over the last 10 years might have contributed to the amplification of resistance in those who may have primary or acquired drug resistance (from prior treatment) and MDR-TB.

Introduction
India has the highest tuberculosis (TB) burden in the world. The annual estimated TB incidence and deaths is 2.7 million and 0.4 million, respectively 1  Axshya SAMVAD study: This study was conducted among new smear-positive pulmonary TB patients to determine the effectiveness of Axshya SAMVAD on diagnosis and treatment initiation delays, costs due to TB diagnosis and treatment outcomes 6,7 . We included all new smear-positive pulmonary TB patients from marginalised and vulnerable populations that were detected through ACF and registered under the programme in the 18 randomly sampled Axshya districts (simple random sampling) during March 2016 to February 2017. Every month in the same districts, we randomly sampled an equal number of new smear-positive pulmonary TB patients registered through PCF from marginalised and vulnerable populations (simple random sampling) 6,7 . Random numbers for simple random sampling were generated using Microsoft Excel.

Data collection
Under Axshya SAMVAD study, we collected data for each study participant through record review (age, sex, ACF/PCF status, residence (urban/rural), distance of residence from microscopy centre, sputum smear grade, weight, diabetes status and HIV status) and patient interviews at their residence. Patient interviews were set up during the review of the participant's record. Before starting the patient interviews, we enquired about their past history of TB treatment for at least one month either from the public or private sector. Those with a past history of treatment were excluded from the Axshya SAMVAD study and referred to the programme for appropriate management. These constitute 'misclassification' for the purpose of present analysis.

Data analysis
We double entered and validated the data using EpiData Entry software 9 (version 3.1, EpiData Association, Odense Denmark). We analysed the data using STATA (version 12.1, copyright 1985-2011 StataCorp LP USA) 10 . We used frequency and proportions (0.95 confidence intervals (CI)) to summarise (infer) the extent of misclassification. Adjusted analysis was done using log binomial regression to determine the factors associated with misclassification. Variables collected during record review (age, sex, ACF/PCF status, residence (urban/rural), distance of residence from microscopy centre and sputum smear grade) were included in the adjusted analysis. Baseline weight was missing in two-fifths of patients; baseline diabetes status was missing for more than three-fifths and HIV status was missing for two-fifths. Hence, we excluded them from the adjusted analysis. The association was summarized (inferred) using adjusted prevalence ratios (95% CIs).

Ethics
The Axshya SAMVAD study was approved by the Ethics Advisory Group of The Union, Paris, France (EAG number 15/15, dated 28 September 2015). We conducted the study after receiving approvals from the State Tuberculosis Officers in the respective states (18 randomly sampled Axshya districts belonged to seven states). We obtained written informed consent for participation from all the study participants. Figure 1 depicts the misclassification of 'previously treated' smear-positive pulmonary TB patients as 'new'. A total of 629 newly registered smear-positive pulmonary TB patients were enrolled for the Axshya SAMVAD study. We couldn't contact 108 (17%) for interview as patients were not available at their residence during the visit (a maximum of two visits were made). among the PCF group (p=0.18). No factors were found to be significantly associated with misclassification (Table 1). Patients belonging to rural areas had higher prevalence of misclassification when compared to urban areas (12% vs 2%), but this difference was not statistically significant probably due to small sample size.

Key findings
About one in ten 'new' TB patients had a past history of TB treatment. This misclassification meant that these patients received the wrong treatment regimen as per the national guidelines at the time. This is similar to the RNTCP report of 2018 3 and previous documentation in 2007 5 . The misclassification among new smear-positive TB patients was two times higher than the 4.5% reported from Malawi in 2000 11 .
One possible reason for this might be a lack of attention on the part of the medical officer to enquire for previous history of TB before starting treatment. Ambiguity in classification when there was a large gap between previous and current treatment, absence of treatment records and patients' reluctance to disclose previous treatment details due to possible stigma (fear of being seen as a 'problem patient') could be the other reasons 5 .

Limitations
This study has some limitations. First, this programmatically relevant finding was incidental and part of a larger study (Axshya SAMVAD study) and hence, we did not systematically record the details of past TB treatment (when, duration of treatment, whether under programme or in private sector) and the reasons for misclassification. Secondly, as patients with misclassification were excluded from the Axshya SAMVAD study, we do not know what happened to them, including their treatment outcomes. Thirdly, we did not include smear-negative pulmonary TB and extrapulmonary TB patients as they were not part of the Axshya SAMVAD study. In Malawi (2000) 11 , they had a higher risk of misclassification when compared to smearpositive pulmonary TB patients. Finally, non-response was a limitation. However, in a best-case scenario (assuming all 108 non-responders did not have previous history of TB treatment), the proportion of misclassification would have been 8.9% (56/629) which is still programmatically significant.

Implications for the TB programme
Limitations notwithstanding, our study has programme implications. Of the new smear-positive pulmonary TB patients registered in India in 2016, 21% had an unfavourable outcome 3 . Some of these unfavourable outcomes can be explained by wrong management -patients getting an inferior treatment regimen (previously treated patients being treated with a regimen meant for new cases) and missing an opportunity for drug susceptibility testing (as previously treated patients were eligible for DST at the time). Inferior regimen might have also contributed to amplification of resistance in those who may have primary or acquired drug resistance (from prior treatment) and MDR-TB. This has been happening for over 10 years so one can see why India now faces the serious problem of drug resistant TB.

Recommendations
Our findings were based on patients from marginalised and vulnerable populations and this limits our generalisability to TB patients registered from the general population. The programme should consider replicating similar studies among patients from the general population with a possible sub-group to look for rural-urban differences.
Since 2017, the revised laboratory register at the level of designated microscopy centres under the RNTCP (one per 50 000 to 100 000 population) also captures this information of previous treatment 8 . RNTCP staff needs to be re-sensitized to "ask" for previous history of TB treatment. Hence, complete filling of the revised laboratory register at microscopy centres should be closely monitored by the programme and future operational research should focus on this.
Systematic qualitative enquiry is recommended to understand the 'why' (why does it happen) and 'how' (how can it be addressed) of misclassification. In the national case-based TB notification software (NIKSHAY), record linkage and deduplication using key attributes may be considered to identify repeat notification of the same person separated by a time period.

Conclusions
This study demonstrated that 'previously treated' patients were being missed and were being registered as 'new' patients under the RNTCP in India. Corrective measures need to be implemented, followed by monitoring any change in the proportion of 'previously treated' patients among all registered patients treated under the programme at national level.

Consent
Written informed consent for publication of the patients' details was obtained from the patients.

Grant information
The author(s) declared that no grants were involved in supporting this work.

If applicable, is the statistical analysis and its interpretation appropriate? Partly
Are all the source data underlying the results available to ensure full reproducibility? Partly

Are the conclusions drawn adequately supported by the results? Partly
No competing interests were disclosed.

Competing Interests:
I confirm that I have read this submission and believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard, however I have significant reservations, as outlined above.

If applicable, is the statistical analysis and its interpretation appropriate? No
Are all the source data underlying the results available to ensure full reproducibility? Yes

Are the conclusions drawn adequately supported by the results? Partly
No competing interests were disclosed.

I confirm that I have read this submission and believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard, however I have significant reservations, as outlined above.
previously treated for TB (at least for one month).

REVIEWER COMMENT
The authors should revise the entire manuscript, particularly Abstract and Introduction, to clarify about what "misclassification" refers to. Misclassification is an epidemiological term, but it can refer to different variables depending on the context. And during the read, it became clear only after going to the results section. Example: (abstract) "reported 11% misclassification among 'new' patients". Maybe the authors could rephrase as: "reported that 11% of patients with tuberculosis were misclassified as new patients regarding previous treatment history" or "reported that 11% of patients with tuberculosis were misclassified as new patients despite their previous TB treatment" or "reported that 11% of patients with tuberculosis were misclassified as new patients despite previously treated" AUTHOR RESPONSE Thank you for the comment. In the revised manuscript, both at beginning of abstract and main text introduction, we have clarified as to what we mean by 'misclassification' (reproduced below) Abstract "In 2007, a field observation from India reported 11% misclassification among 'new' patients registered under the revised national tuberculosis (TB) control programme. Ten years down the line, it is important to know what proportion of newly registered patients has a past history of TB treatment for at least one month (henceforth called as 'misclassification')." Introduction (main text) "This provided us with a unique opportunity to document the proportion of newly registered smear-positive pulmonary TB patients that had previous history of TB treatment and were therefore misclassified (henceforth called as 'misclassification')."

REVIEWER COMMENT
Regarding the inclusion criteria, did it only include adults? AUTHOR RESPONSE Thank you for pointing this out. Yes it included only adults (≥ 15 y). We have clarified this under study population and titles of tables and figures.  4 (0.9, 48.2)). Likely with a bigger sample size and/or higher number of events, you would have better precision and would achieve "significancy" or "no include 1". AUTHOR RESPONSE Thank you for the comment. We categorized aged, because we wanted to explore whether misclassification was significantly higher among certain meaningful sub-groups. Regarding opening up the category 15-44 y, we decided not to do this as there are only 25 events of interest (=misclassification) in this sub-group. If we further divide this sub-group, the event of interest could get reduced and this would have further widened the 95% CI for the adjusted prevalence ratios. As per reviewer comments, we have added a line under results narrative in the revised manuscript to discuss at adjusted prevalence ratio for the variable 'residence' (reproduced below).

REVIEWER COMMENT Regarding
"Patients belonging to rural areas had higher prevalence of misclassification when compared to "Patients belonging to rural areas had higher prevalence of misclassification when compared to urban areas (12% vs 2%), but this difference was not statistically significant probably due to small sample size." REVIEWER COMMENT How is it possible to have such low prevalence of positive HIV status in a vulnerable population? How was the HIV testing coverage? AUTHOR RESPONSE The HIV percentage among all TB patients in India is around 3%. In our study, HIV status was missing in records for two-fifth patients and among those whose test results were recorded (n=288), one was positive.

REVIEWER COMMENT
The discussion section must be improved: Although recommended and ideal, DST will not be available for all. Even if available, the previous history treatment is fundamental. The authors should propose solutions for the problem, such as linkage deduplication, linked records form the national system, etc. Please, better discuss the potential selection bias and representative of the included population regarding India population. AUTHOR RESPONSE Thank you very much. We have reviewed and revised the discussion section as per the comments. We agree that DST may not be available for all. Hence, in the "Implications for the TB programme" sub-section under discussion section we have included this this. We are reproducing it below: "Second, the universal DST is not a reality in every part of the country and in such instances, prioritizing previously treated patients for DST is a better strategy, given the higher prevalence of drug-resistant TB among them." Regarding DST availability, we are reproducing the relevant discussion lines below "Hence, in the present scenario, the impact of misclassification on individual patient management is minimal. This was not the case at the time of conduct of this study. Despite these developments, we think asking for previous treatment history is still relevant for two reasons. First, the information on the proportion of previously treated patients is epidemiologically an important piece of information and is regularly reported to the WHO for monitoring the global TB epidemic. Second, the universal DST is not a reality in every part of the country and in such instances, prioritizing previously treated patients for DST is a better strategy, given the higher prevalence of drug-resistant TB among them" (second paragraph under 'implications for TB programme' sub-section of discussion) Regarding potential selection bias and representativeness, we are reproducing the relevant discussion lines below: "Our findings were based on patients from marginalised and vulnerable populations and this limits our generalisability to TB patients registered from the general population. The programme should consider replicating similar studies among patients from the general population with a possible sub-group to look for rural-urban differences." (First paragraph under 'recommendations' sub-section of discussion) Regarding proposing solutions, we have revised it as per reviewer suggestions under recommendations subsection of discussion (reproducing below) "Since 2017, the revised laboratory register at the level of designated microscopy centres under the RNTCP (one per 50 000 to 100 000 population) also captures this information of previous treatment . RNTCP staff needs to be re-sensitized to "ask" for previous history of TB treatment. Hence, complete filling of the revised laboratory register at microscopy centres should be closely monitored by the programme and future operational research should focus on this. monitored by the programme and future operational research should focus on this. Systematic qualitative enquiry is recommended to understand the 'why' (why does it happen) and 'how' (how can it be addressed) of misclassification. In the national case-based TB notification software ( ), record linkage and deduplication using key attributes may be considered to NIKSHAY identify repeat notification of the same person separated by a time period." REVIEWER COMMENT Minor: Maybe better to use the notation of 95% CI rather than 0.95 AUTHOR RESPONSE We have revised throughout the manuscript as per reviewer suggestion REVIEWER COMMENT Figure 1. I think the first box at left has a wrong "the" : as AN error AUTHOR RESPONSE We are not clear as to what the reviewer means here. We did not find this error in the first left box of figure 1. However, we re-checked figure 1 and found it no grammatical errors. The manuscript addresses one of the crucial problems with in India's Revised National TB Control Program. Interestingly, the study is based on our prior study (in which I was a lead author) in 2007, which was conducted in Mumbai and rural areas of Pune district. Given the research context and the data, I feel this manuscript should go as a brief communication or notes from the field rather than a full original research article. It does not make any novel contribution, but just confirms the earlier research finding. Even to make it a brief communication, I feel some points need to be provocatively addressed. My comments are as below.

REVIEWER COMMENT
Authors have provided details of their larger study, of which the current study is only a part. I feel that level of details is unnecessary here. On the other hand, unfortunately despite having a large team (as seen from the long list of authors), they did not do in-depth inquiry into the reasons for an erroneous categorization of cases, which is actually the main aim of the manuscript. They just mentioned the same reasons for erroneous categorization by providing a reference to our 2007 1.

4.
mentioned the same reasons for erroneous categorization by providing a reference to our 2007 article without making any new contribution. Identifying the most prominent reasons would have been helpful to identify as a focus area for the policy makers to make some action plan for operational implementation of the program. In my opinion, there is not much substance to publish it as an original research article. Authors nowhere discussed the major implication of their observation that even after 10 years, there remains a big disconnect between the operational research and the program implementation, which is really unfortunate. This finding has another major implication that because of erroneous categorization, previously treated cases are being treated with first-line regimen, which results in amplification of resistance in those cases who may have primary or acquired drug resistance (from prior treatment) and/MDR-TB. This has been happening for over 10 years so one can see why India now faces the serious problem of drug resistant TB. I am not convinced with the factor analysis in Table 1

REVIEWER COMMENT
The manuscript addresses one of the crucial problems with in India's Revised National TB Control Program. Interestingly, the study is based on our prior study (in which I was a lead author) in 2007, which was conducted in Mumbai and rural areas of Pune district. Given the research context and the data, I feel this manuscript should go as a brief communication or notes from the field rather than a full original research article. It does not make any novel contribution, but just confirms the earlier research finding. Even to make it a brief communication, I feel some points need to be provocatively addressed. My comments are as below. AUTHOR RESPONSE Thank you for the comment. F1000Research does not have the article types as suggested: brief communication or notes from the field. 'Research notes' is the closest option, but here too there is no specific word count mentioned. Considering our manuscript is around 1800 words only, we are fine if F1000Research agrees to consider this as a research note. We agree the findings are not novel, but are timely considering 10 years down the line, the issue remains as it is and this is important to highlight. We have addressed your comments, details below.

REVIEWER COMMENT
Authors have provided details of their larger study, of which the current study is only a part. I feel that level of details is unnecessary here. On the other hand, unfortunately despite having a large team (as seen from the long list of authors), they did not do in-depth inquiry into the reasons for an erroneous categorization of cases, which is actually the main aim of the manuscript. They just mentioned the same reasons for erroneous categorization by providing a reference to our 2007 article without making any new contribution. Identifying the most prominent reasons would have been helpful to identify as a focus area for the policy makers to make some action plan for operational implementation of the program. In my opinion, there is not much substance to publish it as an original research article. AUTHOR RESPONSE The actual study was done across 18 randomly spread districts of the country (Axshya SAMVAD study -Axshya SAMVAD is an ACF strategy for detecting TB and is implemented by The Union South East Asia). As mentioned, the findings were accidental (not intended) and we thought it is important to report this and look for differences in rates of misclassification across various patient sub-groups. The Axshya SAMVAD study was conducted by project Axshya staff in operational setting without any additional funding for the study itself. The research team in the field (project Axshya staff) were not trained and did not have the required capacity to conduct qualitative research. Hence, we could not go in-depth into the reasons. But, one key point is that our study was conducted in randomly selected 18 districts of the country and hence, the findings are representative. We agree the findings are not novel, but are timely considering 10 years down the line, the issue remains as it is and this is important to highlight. Considering our manuscript is around 1800 words only, we are fine if F1000Research agrees to consider this as a research note.
Authors state in discussion that as per the WHO recommendation, universal DST will be done for all TB cases. This is an ideal situation. The scale up of GeneXpert even remains questionable. The WHO global report 2018 showed that only 40% of TB cases in India were subjected to GeneXert in 2017. There are problems of shortage of cartridges, falcon tubes, power outages etc. Given the glacial speed of translation of findings from operational research in the actual RNTCP implementation, India's claim of TB elimination by 2025 looks really questionable. AUTHOR RESPONSE Thank you for the comment. We agree with the reviewer. We agree that DST for all TB patients is an ideal situation. Therefore, we have also stated that in a non-ideal situation (DST not possible for all TB patients), DST may be focussed on those with previous history of TB treatment. Therefore, correct classification as new or previously treated is still relevant. We have mentioned this in the last three lines of the section 'implications for the TB programme' (reproduced below): "Second, the universal DST is not a reality in every part of the country and in such instances, prioritizing previously treated patients for DST is a better strategy, given the higher prevalence of drug-resistant TB among them." There are no competing interests to declare Competing Interests: The benefits of publishing with F1000Research: Your article is published within days, with no editorial bias You can publish traditional articles, null/negative results, case reports, data notes and more The peer review process is transparent and collaborative Your article is indexed in PubMed after passing peer review Dedicated customer support at every stage For pre-submission enquiries, contact research@f1000.com