Tuberculosis notification in a private tertiary care teaching hospital in South India: a mixed-methods study

Objectives India contributes approximately 25% of the ‘missing’ cases of tuberculosis (TB) globally. Even though ~50% of patients with TB are diagnosed and treated within India’s private sector, few are notified to the public healthcare system. India’s TB notification policy mandates that all patients with TB are notified through Nikshay (TB notification portal). We undertook this study in a private hospital to assess the proportion notified and factors affecting TB notifications. We explored barriers and probable solutions to TB notification qualitatively from health provider’s perspective. Study setting Private, tertiary care, teaching hospital in Bengaluru, South India. Methodology This was a mixed-methods study. Quantitative component comprised a retrospective review of hospital records between 1 January 2015 and 31 December 2017 to determine TB notifications. The qualitative component comprised key informant interviews and focus groups to elicit the barriers and facilitators of TB notification. Results Of 3820 patients diagnosed and treated, 885 (23.2%) were notified. Notifications of sputum smear-positive patients were significantly more likely, while notifications of children were less likely. Qualitative analysis yielded themes reflecting the barriers to TB notification and their solutions. Themes related to barriers were: (1) basic diagnostic procedures and treatment promote notification; (2) misconceptions regarding notification and its process are common among healthcare providers; (3) despite a national notification system other factors have prevented notification of all patients; and (4) establishing hospital systems for notification will go a long way in improving notifications. Conclusions The proportion of patients with TB notified by the hospital was low. A comprehensive approach both by the hospital management and the national TB programme is necessary for improving notification. This includes improving awareness among healthcare providers about the requirement for TB notifications, establishing a single notification portal in hospital, digitally linking hospital records to Nikshay and designating one person to be responsible for notification.

Goals. [3,4] Healthcare delivery in India involves both the public and private sectors. Therefore, it is not surprising that many patients with TB are diagnosed and treated within the private healthcare sector. [5] However, few of these patients receive care from, or are notified i.e., reported, to the Revised National Tuberculosis Control Programme (RNTCP). [4,6] Though, mandatory TB notification introduced in 2012 saw a sharp increase in TB notifications, notifications from the private sector continues to be low. [4,[6][7][8][9][10] Disease notification facilitates TB care through the RNTCP by improving the estimates of disease prevalence that are essential to planning, monitoring, and evaluation of healthcare programmes. Lack of time, low awareness regarding notification, concern towards breaching patient confidentiality, operational complexities in notifying along with lack of trust in the public sector contribute to poor TB notifications from the private sector. [11,12] The information on the extent of notification from private tertiary care teaching facilities is limited. This study was designed to determine the proportion of TB cases notified and the factors that affect notification at such a private tertiary care teaching healthcare facility in Karnataka State, South India. The study also explored qualitatively the gaps in the existing notification systems so as to enable the identification and development of strategies to improve notification.

Study design
A mixed-methods study comprising retrospective review of records to quantitatively assess the proportions of patients with TB notified, and a qualitative component to identify barriers to TB notification was used.

Study setting
The study was conducted at a private tertiary level teaching hospital in Bengaluru, Karnataka  When diagnosed with TB at any of the clinical departments at the hospital, patients can choose to avail anti-tubercular treatment (ATT) either through the DOT center, at no cost, or through the hospital's pharmacy, for a cost. The patient's physician guides the patient's choice of treatment on a case by case basis.

Notification
Irrespective of the source of treatment, all patients with TB that are diagnosed or treated at the hospital are expected to be referred to the DOT center for registration with the RNTCP and subsequent notification via the online notification portal Nikshay (an electronic recording and reporting system).
[13] In the study hospital notification of TB cases is the responsibility of the TBHV.

Study Population
Quantitative component All patients diagnosed with TB and/or treated for TB from 1 st January 2015 to 31 st December 2016 comprised the study population. For this study, a case of TB was defined as (i) a patient with microbiologically confirmed TB using microscopy, bacterial culture, and/or GenXpert MTB/RIF ® OR (ii) a patient with histopathological or radiological findings suggestive of TB, irrespective of microbiological confirmation, OR (iii) a patient who availed ATT from the hospital's pharmacy identified through the pharmacy information system (PIS).

Qualitative component
Health care providers caring for TB patients from various departments including clinicians, staff nurse, researchers, RNTCP LT, and RNTCP TBHV were interviewed in-depth. Participants were chosen purposively to include those involved at various points within the TB case management cascade which is depicted in Figure 1.

Data sources, variables and data collection procedures
Quantitative component Details of patients with TB were extracted from multiple sources. Data was first extracted from the inpatient electronic medical records database using the International Classification of Disease 10 (ICD 10) coding for TB (codes A15 to A19). Subsequently data from the histopathology component of the laboratory information system (LIS) was extracted. For this, search terms such as "tuberculosis", "TB" and for possible typographical errors and "lower and upper case formats" (eg., TB or tb) were used as these diagnoses did not follow the ICD 10 coding. Data was similarly extracted from the Radiology Information System (RIS). These comprised reports from Computerised Tomography (CT) and Magnetic Resonance Imaging (MRI). Chest radiographs were not reported in the RIS as physicians review them in the light of clinical evidence for diagnosis. A laboratory or radiology report that read "acid-fast bacilli (AFB) positive" or "MTB detected" or "strongly suggestive of TB" were considered as a case of TB. When in doubt, two physicians reviewed the reports and arrived at a consensus on the diagnosis. The pharmacy information system (PIS) provided data regarding ATT purchased at the hospital's pharmacy.
Further, details of positive reports from sputum microscopy and culture registers were manually extracted and entered into Microsoft (MS) Excel as they were not available in the LIS.
A "master database" for TB patients diagnosed and/or treated in 2015 and 2016 was created using the unique hospital number (allocated to a patient at registration in the hospital) to A "notification database" for TB patients notified was also created. For this, data from the RNTCP register at the DOT center of the hospital was entered into MS Excel. This was merged with data extracted from Nikshay portal. Patients diagnosed in late 2016 but got notified in the first quarter of 2017 were also incorporated into this database.
In order to identify the proportions of TB cases notified the "master database" was matched with the "notification database", using the VLOOKUP function in MS Excel. The patient's name was used as the primary matching variable. Records with a typographical mismatch in the patient's name were matched using a perfect match for 'sex' within an age range of ±3 years. Flowchart of data sources is depicted in Figure 2.

Qualitative Component
We conducted 11 in-depth interviews (IDI) with various healthcare providers and one focus group discussion (FGD) only with the nursing staff. The first author (AS), a physician trained in qualitative research, conducted the interviews. Two of the interviews were conducted in the local language, Kannada, and the rest in English. All interviews were audio recorded. A rapporteur made field notes during the interviews. After each interview, the key points were summarized and verified with the participants for validation. Data saturation guided the sample size. Each IDI lasted for 15-45 minutes and the FGD lasted for an hour.

Data Analysis
Quantitative component software were used for data analysis. The proportion of TB patients notified was the outcome variable. Associations (unadjusted) between the outcome variable and demographic and clinical characteristics were derived using the Chi square test. All bivariate associations with a 'p value'<0.20 were included in a log-binomial regression model to obtain adjusted prevalence ratios (PR) with 95% confidence intervals. A 'p value'<0.05 was considered statistically significant.

Qualitative component
All interviews and field notes were transcribed and translated into Englishfor analysis using the 'thematic framework approach'. The first and last author (AS, RR) familiarised themselves with a few transcripts and manually coded them. The codes were then compared and categorised based on similarity. This formed the framework for the analysis. [14] The rest of the transcripts were subsequently indexed using the codes generated. Additional codes were added as and when necessary. (Panel 1) The data was then summarised and mapped under various subthemes and themes which were reviewed by the rest of the authors for consensus.

Patient and Public Involvement Statement
Patients were not involved in the design or conduct of the study.  Services at the hospital. Written informed consent was obtained from healthcare providers prior to interviews and included consent to audio-record the interviews. patients with TB were identified through the pharmacy database while nearly 25% were identified through the LIS and laboratory registers.
Of those notified 82 (9%) were also recorded in the Nikshay portal. Factors associated with notification are shown in Table 2. Notification was significantly lower (unadjusted analysis) in children, inpatients and patients identified through the LIS and PIS. Notification was significantly higher for patients whose diagnosis was confirmed microbiologically (sputumsmear microscopy, culture or GenXpert MTB/RIF ® ). The final adjusted regression model comprised age and sputum-microscopy as determinants of notification.    -1·4)  0·819  --Year  ------2015  2071  482 (23·3)  1  ---2016 1749 The four themes that emerged through the qualitative analysis were (i) basic diagnostic modalities and treatment promote notification of TB (ii) misconceptions regarding notification and its process are common amongst healthcare providers (iii) despite training efforts, introduction of the Nikshay portal does not ensure notification of all patients, and (iv) establishing systems for notification will go a long way in improving notifications. (Table 3, Patients whose diagnosis was based on sputum microscopy and those receiving treatment through the RNTCP were more likely to be notified than those requiring complex diagnostics. ii. Misconceptions regarding notification and its process are common amongst healthcare providers: The level of knowledge and awareness regarding notification and its systems was poor.
Healthcare providers did not perceive notification as their responsibility.

Fear of compromising privacy interferes with notification
Fear of stigma from a breach in confidentiality prevents patients from sharing personal identifiers such as phone numbers. This limits entries into the notification portal due to missing information in "mandatory fields".
iv. Establishing systems for notification will go a long way in improving notifications: Notification policy, standard operating procedures, and dedicated personnel supported with innovative technologies such as hotlines and mobile applications were suggested.

Comprehensive institutional notification policy for TB-a necessity
Developing and implementing a comprehensive institutional notification policy to improve notification was suggested. This policy was expected to provide guidance for delegating responsibilities and linking the various components of the hospital information system to enable identification and notification of patients with TB.

DISCUSSION
The Indian private healthcare sector is estimated to cater to approximately 2/3 rd of the inpatients and 3/4 th of the outpatients in the country. [16] The private healthcare sector also accounts for 54% of the healthcare teaching facilities in India. [16] It is therefore not surprising that the private sector treats 2/3rds i.e., 2.2 million annually diagnosed patients with TB. [6] Despite this and India's mandatory TB notification policy, only 19% (0.3 million) of the TB notifications were from the private sector. [8] While poor knowledge, attitudes and organisational systems are considered to limit TB notifications. [12,17] Few reports have explored existing gaps in notification within the private sector. To our knowledge this is the first report on the extent of TB notification and its challenges from a private tertiary care teaching hospital in India.
As in other studies, poor awareness and attitudes along with inadequate systems limited the TB notifications at the hospital to a quarter of those diagnosed. Some private practitioners are of the opinion that notification of TB is unlikely to bring about change in prescription practices and question the need for collecting personal information that does not lead to public health action. [17] Therefore, training and sensitization of healthcare personnel for notification is recommended. Such training should focus on the benefits of notification from the public health and ethical perspective.
[12] It is also essential for the RNTCP to provide annual feedback to healthcare providers of the numbers notified and how this affects policy for TB care. Additionally, obtaining feedback from private practitioners regarding the notification process is expected to boost provider morale and thereby, notifications. [10] Linking hospital records electronically could simplify notification. This does not eliminate manual data entry into the notification portal. Software solutions that feed data to the notification portal automatically could simplify notification and are currently being explored for MDR diagnostic machines. [7] Further, applying ICD-10 codes for diagnoses, commonly used within TB notification systems globally, [10] could standardise diagnoses, enable data capture through software systems and simplify notification.
The guidance for TB notification in India suggests the appointment of a TB nodal officer. [7] The TBHV who currently fulfils this role in our context is probably overburdened with responsibilities in the absence of the 'DOT centre medical officer,' a functionary the RNTCP.
Reports from the private sector also indicate the need for additional human resources in the light of the volume of patients that they carter to. [18] Identifying an additional 'nodal officer' for TB notification from amongst existing institutional personnel could optimise the use of existing resources for notification.
Healthcare providers suggested innovative ICTs such as mobile applications for notification.
However, the short messaging service (SMS), interactive voice calls (IVR) or phone calls to notify TB enabled by the RNTCP for notification, are not as popular as expected. Further, though the Niskhay mobile application that is underway to simplify the notification process holds promise, [19] whose effectiveness remains to be explored.
Healthcare providers elsewhere in India recommend simplifying the existing notification technology to promote uptake. [19] Regular training that includes Nikshay updates, which is much desired, but currently negligible could break existing technological barriers and enhance notification. As the DOT center at the hospital is located within the Chest Medicine department, it is not surprising that sputum positive patients are notified. Only 17% of Chest Physicians notified TB, reflecting the gap between awareness and action. [11] However, in our study, ownership of the DOT center probably made notification a responsibility of the chest physician and enhanced their engagement with the RNTCP. Locating DOT centers within clinical departments with the largest burden of patients with TB patients to improve notifications is worth exploring.
Though the Indian Academy of Paediatrics supports TB treatment through the RNTCP [20] the proportion of children with TB notified was low, reflecting the limited involvement of paediatricians in the RNTCP. The questionable bioavailability of paediatric ATT formulations and alternate day dosing schedules are known barriers to engaging paediatricians with the RNTCP. [21] The introduction of the daily regimen with 'body weight bands' that inform dosing, has the potential to improve provider engagement with the RNTCP and improve TB notifications thereof, irrespective of the patients' age. [22] Also, creating a TB registry within each clinical department could improve department-wise notifications.
Linking patient records using a unique identification number (Aadhaar number) [15] or hospital number, and extending this system to involve the Nikshay portal could minimise duplication, simplify record and help retain patients in care. Studies indicate that patients are weary of sharing personal identifiers i.e., mobile phone numbers and address, for notification. [9,11] This necessitates sensitizing the general public of the need for mandatory disease notification through mass media campaigns and patients through counselling sessions. Further, perceived stigma prevents healthcare providers involved in notification uncomfortable with obtaining personal identifiers from patients. [9,11] Mobile phone numbers and the patients address are mandatory fields in the Nikshay portal, without which notification is incomplete. Therefore, reminding healthcare providers of their obligation to obtain and report personal identifiers of patients with notifiable disease, as per the Indian Medical Council's (MCI) Regulations 2002, [23] might minimise discomfort in the light of responsibility. Simultaneously, mass media, posters and brochures placed in waiting rooms regarding notification could mitigate patients' fears with sharing personal identifiers.
Though punitive action for non-notification exists in India, it is not yet implemented. [23] As in other TB high burden countries, a recent mandate suggests that non-notification could result in heavy fines and even imprisonment. [24] In the light of Government of India's politico-administrative commitment towards TB control, punitive action is an eventuality that is best avoided. Therefore, at institutional level, enabling incentives for notification (tangible or intangible) and disincentives for non-notification ('warnings/ memos', or monetary penalty) could reinforce the importance of notification. Further, the RNTCP provides a cash incentive of 250 INR to a 'private' healthcare provider for every patient with TB notified. [25] Institutional proactiveness to ensure that its healthcare providers receive this incentive could also improve notifications.
Finally, our geneXpert equipment was acquired through the Initiative for Promoting Affordable and Quality TB tests (IPAQT) project, [26,27]

Methodological Issues
The mixed methods design with the quantitative and qualitative components validated and complemented each other. It is possible that our definition of notification overestimated the numbers notified. We were also liberal with our criteria for matching databases. However, we included all patients both diagnosed and treated at the hospital even if they availed a 'onetime' consultation. This probably also inflated the denominator minimising any overestimate.

Conclusions
The low proportions of TB notifications at the hospital call for urgent action to identify strategies that can improve notification. A combined approach from within (managerial) and outside the institution (RNTCP) is necessary. Generating awareness regarding notification and developing a comprehensive notification policy along with establishing a notification portal is essential. Supplementing this with technological innovations such as mobile applications and expanding the scope of the existing hospital information system to capture outpatient data and link patient records is essential. We also call upon tertiary level teaching hospitals both within India and globally to evaluate the TB notifications and its barriers in their setting. Such information is hoped to support the development of evidence based strategies that enhance public private engagement for TB notification and control.

CONTRIBUTORS
AS, GDS and RR conceived the study. AMK and EW alsong with AS, GDS and RR designed the protocol. NA did the literature search. AS and NA did data entry and extracted the electronic data. AS undertook the interviews. AS, AMK, RR, and TMM analysed and interpreted the data. AS drafted with support from AMK, and RR. GDS and EW critically revised the successive drafts. All authors have seen and approved the final version of this manuscript for publication.

FUNDING
The training programme and open access publications costs were funded by the Department for International Development (DFID), UK and La FondationVeuve Emile Metz-Tesch (Luxembourg). The study is supported by grants from the Swedish Research Council and the Welcome trust/DBT India Alliance. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Corresponding author,AS confirms that she had fullaccess to all the data in the study and had final responsibility forthe decision to submit for publication.

DECLARATION OF INTERESTS
AS, GDS and RR are employed at the tertiary level teaching facility that was evaluated in this study.

DATA SHARING STATEMENT
Data will be made available with reasonable request from the corresponding author. Results: Of 3820 patients diagnosed and treated, 885(23.2%) were notified. Notifications of sputum-smear positive patients were significantly more likely, while notifications of children with TB were less likely. Qualitative analysis yielded themes reflecting the barriers to TB notification and their solutions. Themes related to barriers were (i) basic diagnostic procedures and treatment promote notification, (ii) misconceptions regarding notification and its process is common among healthcare providers (iii) despite a national notification system, other factors have prevented notification of all patients and iv) establishing systems for notification(in the hospital) will go a long way in improving notifications.

REFERENCES
Conclusions: Proportion of patients with TB notified by the hospital was low. A comprehensive approach both by the hospital management and the national TB programme is necessary for improving notification. This include, improving awareness among health care providers about the requirement for TB notifications, establishing a single notification portal in-hospital, digitally linking hospital records to Nikshay and designating a notification specific human resource.

Strengths and Limitations:
• A mixed methods design where the qualitative component explains and complements the findings from the quantitative component.   [3,4] Healthcare delivery in India involves both the public and private sectors. The Indian private healthcare sector is estimated to cater to approximately 2/3 rd of the inpatients and 3/4 th of the outpatients in the country. [5] The private healthcare sector also accounts for 54% of the healthcare teaching facilities in India. [5] It is therefore not surprising that approximately 2/3 rd of the 2.2 million patients with TB annually are diagnosed and treated within the private healthcare sector. [6] However, in 2017 only 19% (81% from public sector) of these patients receive care from, or are notified i.e., reported, to the Revised National Tuberculosis Control Programme (RNTCP) [4,7], India's national health program for the prevention and control of TB. Though, mandatory TB notification introduced in 2012 saw a sharp increase in TB notifications, notification from the private sector continues to below. [4,[7][8][9][10][11]. This despite launching Nikshay, the case based web based national TB notification portal, accessible to all healthcare providers, laboratories and diagnostic facilities, both public and private, nationwide.
Improving the estimates of disease prevalence though are essential for planning, monitoring, and evaluation of RNTCP. Yet barriers such as lack of time, poor awareness regarding notification, concern about breaching patient confidentiality, operational complexities in notifying along with lack of trust in the public sector. [12][13][14]  The information on the extent of notification from private tertiary care teaching facilities is limited. This study was designed to determine the proportion of TB cases notified and the factors that affect notification in a private tertiary care teaching healthcare facility in Karnataka State, South India. The study also explored qualitatively the gaps in the existing notification systems so as to enable the identification and development of strategies to improve notification.

Study design
A mixed-methods study comprising retrospective review of records to quantitatively assess the proportions of patients with TB notified, and a qualitative component to identify barriers to TB notification was used.

Study setting
The study was conducted at a private tertiary level teaching hospital in Bengaluru, Karnataka state in South India. The hospital has a 1250 bedded inpatient facility and caters to approximately 2000 out patients from diverse backgrounds, everyday. A network of laboratory, pathology, and radiology services support the clinical departments at the hospital.
TB specific microbiological services available are microscopy, GenXpert MTB/RIF ® , solid culture, and liquid TB culture and drug susceptibility testing (DST) (such as Mycobacterial Growth Indicator Tube). A computerised information system for these services and the pharmacy exist at the hospital. The Indian RNTCP and its relationship with the study hospital [15] Administrative set up: The RNTCP, a vertical national health program, strives to provide care and treatment at no cost to all patients with TB in India. The program adheres to the diagnostic and treatment recommendations of the World Health Organization (WHO) [16] The program delivers its services through a network of designated microscopy center (DMC, population covered: 0.1 million) and peripheral health institutions (PHI) (primary, secondary and tertiary healthcare facilities including all healthcare academia).

The Medical Records department (MRD) compiles and maintains inpatient and outpatient
In addition, Direct observed treatment (DOT) centers at PHIs are responsible for dispensing treatment, observing treatment doses swallowed (DOT), patient follow-up and patient retention in care. Till 2017, the RNTCP followed an alternate day treatment regimen, with DOT thrice a week in the intensive phase (2 months) and weekly once in the continuation phase (4 months). All public PHIs function as DOT enters and has a TB health visitor (TBHV), responsible for DOT and patient retention. DOT centers at academic institutions however, have a medical officer in addition to the TBHV. A PHI may also function as DMC.
Private sector: The RNTCP sets guidelines but does not dictate diagnostic or treatment protocols to the private sector. However, it attempts to deliver public services to the private sector through public private partnerships (PPP) and expects all private healthcare providers to notify TB patients irrespective of a PPP, through Nikshay.
Management of TB at the hospital: By virtue of being a private tertiary care teaching hospital the RNTCP has established a DMC and a DOT center at the hospital through a PPP. The (iii) a patient who availed ATT from the hospital's pharmacy identified through the pharmacy information system (PIS).

Qualitative component
Health care providers caring for patients with TB from various departments including clinicians, staff nurse, researchers, LT, and TBHV were interviewed in-depth. Participants were chosen purposively to include those involved at various points within the TB case management cascade which is depicted in Figure 1.

Quantitative component
Demographic details of patients with TB such as patient's name, date of birth, gender, education, marital status, and residence (urban/rural), and year diagnosed, clinical department visited, source of the record were extracted frommultiple sources. Data was first extracted from the inpatient electronic medical records database using the International Classification of Disease 10 (ICD 10) coding for TB (codes A15 to A19). Subsequently data from the histopathology component of the laboratory information system (LIS) was extracted. For this, search terms such as "tuberculosis", "TB" and for possible typographical errors and "lower and upper case formats" (e.g., TB or tb) were used, as these diagnoses did not follow the ICD 10 coding. Data was similarly extracted from the Radiology Information System (RIS). These comprised reports from Computerised Tomography (CT) and Magnetic Resonance Imaging (MRI). Chest radiographs were not reported in the RIS as physicians review them in the light of clinical evidence for diagnosis. A laboratory or radiology report that read "acid-fast bacilli (AFB) positive" or "MTB detected" or "strongly suggestive of TB" were considered as patients with TB. When in doubt, two physicians reviewed the reports and arrived at a consensus on the diagnosis. The pharmacy information system (PIS) provided patient data for ATT purchased at the hospital's pharmacy.
Further, details of positive reports from sputum microscopy and culture registers were manually extracted and entered into Microsoft (MS) Excel as they were not available in the LIS.
A "master database" for TB patients diagnosed and/or treated in 2015and 2016 was created using the unique hospital number (allocated to a patient at registration in the hospital) to match records and delete duplicate records in the various databases (PIS, LIS, RIS and manual registers).
A "notification database" for TB patients notified was also created. For this, data from the RNTCP register at the DOT center of the hospital was entered into MS Excel. This was merged with data extracted from Nikshay portal. Patients diagnosed in late 2016 but who werenotified in the first quarter of 2017were also incorporated into this database.
In order to identify the proportions of TB cases notified the "master database" was matched with the "notification database", using the VLOOKUP function in MS Excel. The patient's name was used as the primary matching variable. Records with a typographical mismatch in the patient's name were matched using a perfect match for 'sex' within an age range of ±3 years. Flowchart of data sources is depicted in Figure 2.

Patient and Public Involvement Statement
Patients were not involved in the design or conduct of the study.

Quantitative component
A total of 3820 patients were diagnosed with TB and/or treated during the study period. The demographic details of the patients with TB are described in Table 1. The median (interquartile range) age was 40 (27-56) years and 7% of the patients were children <15 years of age.
About a quarter of the patients received inpatient care and of them, nearly half were under the care of department of internal medicine, followed by chest medicine, neurology and paediatrics. About half of the patients with TB were identified through the pharmacy database while nearly 25% were identified through the LIS and laboratory registers.
Of those notified, only 82(9%) were also recorded in the Nikshay portal. Factors associated with notification are shown in Table2. Notification was significantly lower (unadjusted analysis) in children, inpatients and patients identified through the LIS and PIS. Notification was significantly higher for patients whose diagnosis was confirmed microbiologically (sputum-smear microscopy, culture or GenXpert MTB/RIF ® ). The final adjusted regression model showed age and sputum microscopy as determinants of notification.

Misconceptions regarding notification and its process are common amongst healthcare providers:
The level of knowledge and awareness regarding notification and its systems was poor. Healthcare providers did not perceive notification as their responsibility.

a. Those who do not know, do not notify: Awareness could improve notification
Some healthcare providers were unaware that TB was a notifiable disease, others were unsure of the existing system for notifying TB and yet others presumed that notification was common knowledge. Out of 22 health care providers, 14 were aware of the RNTCP requirement of notification.

Despite a national notification system, other factors prevented notification of all patients:
Inadequate training for using the notification portal, Nikshay and mandatory information requirements within the portal were barriers to notification.

. Fear of compromising privacy interferes with notification
Fear of stigma from a breach in confidentiality prevents patients from sharing personal identifiers such as phone numbers. This limits entries into the notification portal due to missing information in "mandatory fields".

Establishing systems for notification (in the hospital) will go a long way in improving notifications:
Notification policy, standard operating procedures, and dedicated personnel supported with innovative technologies such as hotlines and mobile applications were suggested. Linking hospital records electronically could simplify notification. This does not eliminate manual data entry into the notification portal. Software solutions that feed data to the notification portal automatically could simplify notification and are currently being explored for MDR diagnostic machines. [8]Further, applying ICD-10 codes for diagnoses, commonly used within TB notification systems globally, [11]could standardise diagnoses, enable data capture through software systems and simplify notification.

a. Comprehensive institutional notification policy for TB-a necessity
The guidance for TB notification in India suggests the appointment of a TB nodal officer. [8] The TBHV who currently fulfils this role in our context is probably overburdened with responsibilities in the absence of the 'DOT centre medical officer,' a functionary the RNTCP.
Reports from the private sector also indicate the need for additional human resources in the light of the volume of patients that they carter to. [22] Identifying an additional 'nodal officer' for TB notification from amongst existing institutional personnel could optimise the use of existing resources for notification.
Healthcare providers suggested innovative ICTs such as mobile applications for notification.
However, the short messaging service (SMS), interactive voice calls (IVR) or phone calls to notify TB enabled by the RNTCP for notification, are not as popular as expected. Further, though the Niskhay mobile application that is underway to simplify the notification process holds promise, [23]whose effectiveness remains to be explored.   Because none of us know when to notify and how to notify, I may not have notified

Standard operating procedures involved in notification unknown
Lack of knowledge about the process of notification and assuming somebody else has to notify.

Lack of knowledge regarding notification
Notification is someone else's responsibility 2

Gaps in RNTCP notification policy:
There are additions in NIKSHAY, still they haven't given us proper training Gaps in notification Refresher training on Nikshay has not been given to the RNTCP staff involved in notification even when new forms have been updated in the software.

Lack of basic training in Nikshay
Gaps in user training for the notification portal Nikshay 3

Technological involvement:
The moment we have electronic medical records, if anyone is given ATT and is done online…we would get much better way of tracking them  Healthcare providers elsewhere in India recommend simplifying the existing notification technology to promote uptake. [23] Regular training that includes Nikshay updates, which is much desired, but currently negligible could break existing technological barriers and enhance notification.

-Traditional diagnostic procedures promote notification of TB patients; 2-Misconceptions regarding notification and its process is common in healthcare providers; 3-Despit a national notification system, other factors prevented notification of all patients; 4-Establishing systems for notification (in the hospital) will go a long way in improving notifications IRL-Intermediate Reference Laboratory; RNTCP-Revised National Tuberculosis Control Programme; MDR-Multi Drug Resistant; DOT-Directly Observed Treatment short course; ATT-Anti-TB Treatment;
As the DOT center at the hospital is located within the Chest Medicine department, it is not surprising that sputum positive patients are notified. Only 17% of Chest Physicians notified TB, reflecting the gap between awareness and action.
[12] However, in our study, ownership of the DOT center probably made notification a responsibility of the chest physician and enhanced their engagement with the RNTCP. Locating DOT centers within clinical departments with the largest burden of patients with TB patients to improve notifications is worth exploring.
Though the Indian Academy of Paediatrics supports TB treatment through the RNTCP [24] the proportion of children with TB notified was low, reflecting the limited involvement of paediatricians in the RNTCP. The questionable bioavailability of paediatric ATT formulations and alternate day dosing schedules are known barriers to engaging paediatricians with the RNTCP. [25] The introduction of the daily regimen with 'body weight bands' that inform dosing, has the potential to improve provider engagement with the RNTCP and improve TB notifications thereof, irrespective of the patients' age. [15]Also, creating a TB registry within each clinical department could improve department-wise notifications.
Linking patient records using a unique identification number (Aadhaar number) [18] or hospital number, and extending this system to involve the Nikshay portal could minimise duplication, simplify record and help retain patients in care. Studies indicate that patients are  Though punitive action for non-notification exists in India, it is not yet implemented. [26] As in other TB high burden countries, a recent mandate suggests that non-notification could result in heavy fines and even imprisonment. [27] In the light of Government of India's politico-administrative commitment towards TB control, punitive action is an eventuality that is best avoided. Therefore, at institutional level, enabling incentives for notification (tangible or intangible) and disincentives for non-notification ('warnings/ memos', or monetary penalty) could reinforce the importance of notification. Further, the RNTCP provides a cash incentive of 250 INR to a 'private' healthcare provider for every patient with TB notified. [28] Institutional proactiveness to ensure that its healthcare providers receive this incentive could also improve notifications.
Finally, testing for MDR TB cases in the study hospital was done using GenXpert MTB/RIF ® equipment that was acquired through the Initiative for Promoting Affordable and Quality TB

Methodological Issues
The mixed methods design with the quantitative and qualitative components validated and complemented each other. It is possible that our definition of notification overestimated the numbers notified. We were also liberal with our criteria for matching databases. However, we included all patients both diagnosed and treated at the hospital even if they availed a 'onetime' consultation. This probably also inflated the denominator minimising any overestimate.
The retrospective nature of the quantitative component meant that the study procedures did not influence changes in notification, as might have been observed if the study were prospective. Further, the quantitative component, based on a review of records, is limited by the quality of the data in the records, for example, we could not assess the association of the treating clinical department and treatment regimens on notification.
The study included health care providers who encountered patients with TB at different points in the hospital as represented in Fig 1, including hospital staff and RNTCP staff.
Therefore, we believe that this sample is fairly representative of those health care providers who manage patients with TB. This, along with a description of the study context and methodology enables the reader to judge its applicability of the results to their context. The first author's position as a physician in the study setting helped her contextualize the results.
Sharing the results with all authors with diverse backgrounds and skills improved the interpretation of the results further improving generalizability.

Conclusions
The low proportions of TB notifications at the hospital call for urgent action to identify strategies that can improve notification. A combined approach from within (managerial) and outside the institution (RNTCP) is necessary. Generating awareness regarding notification and developing a comprehensive notification policy along with establishing a notification portal is essential. Supplementing this with technological innovations such as mobile applications and expanding the scope of the existing hospital information system to capture outpatient data and link patient records is essential. We also call upon tertiary level teaching hospitals both within India and globally to evaluate the TB notifications and its barriers in their setting. Such information is hoped to support the development of evidence-based strategies that enhance public private engagement for TB notification and control.

CONTRIBUTORS
AS, GDS and RR conceived the study. AMK and EW along with AS, GDS and RR designed the protocol. NA did the literature search. AS and NA did data entry and extracted the electronic data. AS undertook the interviews. AS, AMK, RR and TMM analysed and interpreted the data. AS drafted with support from AMK, and RR. GDS and EW critically revised the successive drafts. All authors have seen and approved the final version of this manuscript for publication.

FUNDING
The training programme and open access publications costs were funded by the Department for International Development (DFID), UK and La FondationVeuve Emile Metz-Tesch (Luxembourg). The study is supported by grants from the Swedish Research Council and the Welcome trust/DBT India Alliance. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Corresponding author,AS confirms that she had fullaccess to all the data in the study and had final responsibility forthe decision to submit for publication.

DECLARATION OF INTERESTS
AS, GDS and RR are employed at the tertiary level teaching facility that was evaluated in this study.

Conclusions:
The proportion of patients with TB notified by the hospital was low. A comprehensive approach both by the hospital management and the national TB programme is necessary for improving notification. This include, improving awareness among health care providers about the requirement for TB notifications, establishing a single notification portal in-hospital, digitally linking hospital records to Nikshay and designating one person to be responsible for notification.

Strengths and Limitations:
 A mixed methods design where the qualitative component explains and complements the findings from the quantitative component.  Goals. [3,4] Healthcare delivery in India involves both the public and private sectors. The Indian private healthcare sector is estimated to cater to approximately 2/3 rd of the inpatients and 3/4 th of the outpatients in the country. [5] The private healthcare sector also accounts for 54% of the healthcare teaching facilities in India. [5] It is therefore not surprising that approximately 2/3 rd of the 2.2 million patients with TB annually are diagnosed and treated within the private healthcare sector. [6] However, in 2017 only 19% of these patients receive care from, or are notified i.e., reported, to the Revised National Tuberculosis Control Programme (RNTCP) [4,7], India's national health program for the prevention and control of TB, as compared to 81% from public sector. Though, mandatory TB notification introduced in 2012 saw a sharp increase in TB notifications, notification from the private sector continues to be low. [4,[7][8][9][10][11]. This is despite launching Nikshay, the case based web based national TB notification portal, accessible to all healthcare providers, laboratories and diagnostic facilities, both public and private, nationwide.
Improving the estimates of disease prevalence though are essential for planning, monitoring, and evaluation of RNTCP. Yet barriers such as lack of time, poor awareness regarding notification, concern about breaching patient confidentiality, operational complexities in notifying along with lack of trust in the public sector prevents complete notification. [12][13][14] 1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57  58  59  The information on the extent of notification from private tertiary care teaching facilities is limited. This study was designed to determine the proportion of TB cases notified and the factors that affect notification in a private tertiary care teaching healthcare facility in Karnataka State, South India. The study also explored qualitatively the gaps in the existing notification systems so as to enable the identification and development of strategies to improve notification.

Study design
A mixed-methods study comprising a retrospective review of records to quantitatively assess the proportions of patients with TB notified, and a qualitative component to identify barriers to TB notification was used.

Study setting
The study was conducted at a private tertiary level teaching hospital in Bengaluru, Karnataka state in South India. The hospital has 1250 beds and caters to approximately 2000 out patients daily from diverse backgrounds. A network of laboratory, pathology, and radiology services support the clinical departments at the hospital. TB specific microbiological services available are microscopy, GenXpert MTB/RIF ® , solid culture, and liquid TB culture and drug susceptibility testing (DST) (such as Mycobacterial Growth Indicator Tube).
There is a computerised information system for these services and the pharmacy exist at the hospital. The Medical Records department (MRD) compiles and maintains inpatient and outpatient hospital records in paper format. Inpatient records are available electronically and outpatient records are available in paper format.  1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57  58  59  The Indian RNTCP and its relationship with the study hospital [15] The RNTCP, a vertical national health program, strives to provide care and treatment at no cost to all patients with TB in India. The program adheres to the diagnostic and treatment recommendations of the World Health Organization (WHO) [16] The program delivers its services through a network of designated microscopy center (DMC, population covered: 0.1 million) and peripheral health institutions (PHI) (primary, secondary and tertiary healthcare facilities including all healthcare academia).
In addition, Direct observed treatment (DOT) centers at PHIs are responsible for dispensing treatment, observing treatment doses swallowed (DOT), patient follow-up and patient retention in care. Till 2017, the RNTCP followed an alternate day treatment regimen, with DOT thrice a week in the intensive phase (2 months) and weekly once in the continuation phase (4 months). All public PHIs function as DOT centers and have a TB health visitor (TBHV), responsible for DOT and patient retention. DOT centers at academic institutions however, have a medical officer in addition to the TBHV. A PHI may also function as DMC.
Even though the RNTCP sets guidelines it does not dictate diagnostic or treatment protocols to the private sector. However, it attempts to deliver public services to the private sector through public private partnerships (PPP) and expects all private healthcare providers to notify TB patients irrespective of a PPP, through Nikshay.

Quantitative component
Demographic details of patients with TB such as patient's name, date of birth, gender, education, marital status, and residence (urban/rural), and year diagnosed, clinical department visited, source of the record were extracted frommultiple sources. Data was first extracted from the inpatient electronic medical records database using the International Classification of Disease 10 (ICD 10) coding for TB (codes A15 to A19). Subsequently data from the histopathology component of the laboratory information system (LIS) was extracted. For this, search terms such as "tuberculosis", "TB" and for possible typographical errors and "lower and upper case formats" (e.g., TB or tb) were used, as these diagnoses did not follow the ICD 10 coding. Data was similarly extracted from the Radiology Information System (RIS). These comprised reports from Computerised Tomography (CT) and Magnetic Resonance Imaging (MRI). Chest radiographs were not reported in the RIS as physicians review them in the light of clinical evidence for diagnosis. A laboratory or radiology report that read "acid-fast bacilli (AFB) positive" or "MTB detected" or "strongly suggestive of TB" were considered as patients with TB. When in doubt, two physicians reviewed the reports and arrived at a consensus on the diagnosis. The pharmacy information system (PIS) provided patient data for ATT purchased at the hospital's pharmacy.  1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57  58  59  A "notification database" for TB patients notified was also created. For this, data from the RNTCP register at the DOT center of the hospital was entered into MS Excel. This was merged with data extracted from Nikshay portal. Patients diagnosed in late 2016 but who werenotified in the first quarter of 2017were also incorporated into this database.
In order to identify the proportions of TB cases notified the "master database" was matched with the "notification database", using the VLOOKUP function in MS Excel. The patient's name was used as the primary matching variable. Records with a typographical mismatch in the patient's name were matched using a perfect match for 'sex' within an age range of ±3 years. Flowchart of data sources is depicted in Figure 2.

Qualitative Component
We conducted 11in-depth interviews (IDI) with various healthcare providers and one focus group discussion (FGD) with 11 nursing staff. At the time of the study, nursing staff looked after activities such as reporting of diseases, and we conducted a FGD with them as they comprised a fairly homogeneous group of female healthcare providers and were therefore included in an FGD. The first author (AS), a physician trained in qualitative research, conducted the interviews. Two of the interviews were conducted in the local language, software were used for data analysis. The proportion of TB patients notified was the outcome variable. Associations (unadjusted) between the outcome variable and demographic and clinical characteristics were derived using the Chi square test. All bivariate associations with a 'p value'<0.20 were included in a log-binomial regression model to obtain adjusted prevalence ratios (PR) with 95% confidence intervals. A 'p value'<0.05 was considered statistically significant.

Qualitative component
All interviews and field notes were transcribed and translated into Englishforanalysis using the 'thematic framework approach'. The first and last author (AS, RR) familiarised themselves with a few transcripts and manually coded them. The codes were then compared and categorised based on similarity. This formed the framework for the analysis. [17] The rest of the transcripts were subsequently indexed using the codes generated. Additional codes were added as and when necessary. (Box 1) The data was then summarised and mapped under various subthemes and themes which were reviewed by the rest of the authors for consensus.

Quantitative component
A total of 3820 patients were diagnosed with TB and/or treated during the study period. The demographic details of the patients with TB are described in Table 1. The median (interquartile range) age was 40 (27-56) years and 7% of the patients were children <15 years of age.
About a quarter of the patients received inpatient care and of them, nearly half were under the care of department of internal medicine, followed by chest medicine, neurology and paediatrics. About half of the patients with TB were identified through the pharmacy database while nearly 25% were identified through the LIS and laboratory registers.

Misconceptions regarding notification and its process are common amongst healthcare providers:
The level of knowledge and awareness regarding notification and its systems was poor. Healthcare providers did not perceive notification as their responsibility.

a. Comprehensive institutional notification policy for TB-a necessity
The guidance for TB notification in India suggests the appointment of a TB nodal officer. [8] The TBHV who currently fulfils this role in our context is probably overburdened with responsibilities in the absence of the 'DOT centre medical officer,' a functionary the RNTCP.
Reports from the private sector also indicate the need for additional human resources in the light of the volume of patients that they carter to. [22] Identifying an additional 'nodal officer' for TB notification from amongst existing institutional personnel could optimise the use of existing resources for notification.
Healthcare providers suggested innovative ICTs such as mobile applications for notification.
As the DOT center at the hospital is located within the Chest Medicine department, it is not surprising that sputum positive patients are notified. Only 17% of Chest Physicians notified TB, reflecting the gap between awareness and action.
[12] However, in our study, ownership of the DOT center probably made notification a responsibility of the chest physician and enhanced their engagement with the RNTCP. Locating DOT centers within clinical departments with the largest burden of patients with TB patients to improve notifications is worth exploring.
Though the Indian Academy of Paediatrics supports TB treatment through the RNTCP [24] the proportion of children with TB notified was low, reflecting the limited involvement of paediatricians in the RNTCP. The questionable bioavailability of paediatric ATT formulations and alternate day dosing schedules are known barriers to engaging paediatricians with the RNTCP. [25] The introduction of the daily regimen with 'body weight bands' that inform dosing, has the potential to improve provider engagement with the RNTCP and improve TB notifications thereof, irrespective of the patients' age. [15]Also, creating a TB registry within each clinical department could improve department-wise notifications.
Linking patient records using a unique identification number (Aadhaar number) [18] or hospital number, and extending this system to involve the Nikshay portal could minimise duplication, simplify record and help retain patients in care. Studies indicate that patients are  1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57  58  59  weary of sharing personal identifiers i.e., mobile phone numbers and address, for notification. [10,12] This necessitates sensitizing the general public of the need for mandatory disease notification through mass media campaigns and patients through counselling sessions. Further, perceived stigma prevents healthcare providers involved in notification uncomfortable with obtaining personal identifiers from patients. [10,12] Mobile phone numbers and the patients address are mandatory fields in the Nikshay portal, without which notification is incomplete. Therefore, reminding healthcare providers of their obligation to obtain and report personal identifiers of patients with notifiable disease, as per the Indian Medical Council's (MCI) Regulations 2002, [26] might minimise discomfort in the light of responsibility. Simultaneously, mass media, posters and brochures placed in waiting rooms regarding notification could mitigate patients' fears with sharing personal identifiers.
Though punitive action for non-notification exists in India, it is not yet implemented. [26] As in other TB high burden countries, a recent mandate suggests that non-notification could result in heavy fines and even imprisonment. [27] In the light of Government of India's politico-administrative commitment towards TB control, punitive action is an eventuality that is best avoided. Therefore, at institutional level, enabling incentives for notification (tangible or intangible) and disincentives for non-notification ('warnings/ memos', or monetary penalty) could reinforce the importance of notification. Further, the RNTCP provides a cash incentive of 250 INR to a 'private' healthcare provider for every patient with TB notified. [28] Institutional proactiveness to ensure that its healthcare providers receive this incentive could also improve notifications.

Methodological Issues
The mixed methods design with the quantitative and qualitative components validated and complemented each other. It is possible that our definition of notification overestimated the numbers notified. We were also liberal with our criteria for matching databases. However, we included all patients both diagnosed and treated at the hospital even if they availed a 'onetime' consultation. This probably also inflated the denominator minimising any overestimate.
The retrospective nature of the quantitative component meant that the study procedures did not influence changes in notification, as might have been observed if the study were prospective. Further, the quantitative component, based on a review of records, is limited by the quality of the data in the records, for example, we could not assess the association of the treating clinical department and treatment regimens on notification.
The study included health care providers who encountered patients with TB at different points in the hospital as represented in Fig 1, including hospital staff and RNTCP staff.
Therefore, we believe that this sample is fairly representative of those health care providers  1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57  58  59  60   F  o  r  p  e  e  r  r  e  v  i  e  w  o  n  l  y 29 who manage patients with TB. This, along with a description of the study context and methodology enables the reader to judge its applicability of the results to their context. The first author's position as a physician in the study setting helped her contextualize the results.
Sharing the results with all authors with diverse backgrounds and skills improved the interpretation of the results further improving generalizability.

Conclusions
The low proportions of TB notifications at the hospital call for urgent action to identify strategies that can improve notification. A combined approach from within (managerial) and outside the institution (RNTCP) is necessary. Generating awareness regarding notification and developing a comprehensive notification policy along with establishing a notification portal is essential. Supplementing this with technological innovations such as mobile applications and expanding the scope of the existing hospital information system to capture outpatient data and link patient records is essential. We also call upon tertiary level teaching hospitals both within India and globally to evaluate the TB notifications and its barriers in their setting. Such information is hoped to support the development of evidence-based strategies that enhance public private engagement for TB notification and control.  1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57  58  59

CONTRIBUTORS
AS, GDS and RR conceived the study. AMK and EW along with AS, GDS and RR designed the protocol. NA did the literature search. AS and NA did data entry and extracted the electronic data. AS undertook the interviews. AS, AMK, RR and TMM analysed and interpreted the data. AS drafted with support from AMK, and RR. GDS and EW critically revised the successive drafts. All authors have seen and approved the final version of this manuscript for publication.

FUNDING
The training programme and open access publications costs were funded by the Department for International Development (DFID), UK and La FondationVeuve Emile Metz-Tesch (Luxembourg). The study is supported by grants from the Swedish Research Council and the Welcome trust/DBT India Alliance. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Corresponding author,AS confirms that she had fullaccess to all the data in the study and had final responsibility forthe decision to submit for publication.

DECLARATION OF INTERESTS
AS, GDS and RR are employed at the tertiary level teaching facility that was evaluated in this study.