Implementation research for digital technologies and tuberculosis

Implementation research for digital technologies and tuberculosis

Case study

Connected TB diagnostics platform in South Africa

(Programmatic management)
CASE STUDIES: 4 Photography by Annie Spratt


Initial loss to follow up (ILFU) is defined as failure to commence treatment within 28 days of microbiologically confirmed TB diagnosis and is a significant problem in South Africa. ILFU is influenced by a variety of factors such as delays in clinics receiving results or failure to receive results from laboratories; the need for frequent patient attendance at clinics to inquire about results and long waiting periods; and missed work due to clinic attendance. This study trialed a new mHealth intervention to help bridge gaps that contribute to ILFU among people with TB in South Africa.

Research objectives

To explore the feasibility, acceptability and potential of an mHealth application to reduce ILFU among newly diagnosed TB patients in Johannesburg, South Africa.


100% of health staff in two targeted health facilities accessed the connected diagnostics platform via a mobile device pre-installed with an application connected to the data warehouse. An electronic notification of TB sputum results to clinic TB staff was sent via the application. This mobile device enabled all targeted TB clinic staff to receive an electronic notification of TB sputum results via the application.


Effectiveness was evaluated as the proportion of diagnostic results available within 48 hours during the implementation period, which was compared with the proportion of diagnostic results available within 48 hours during the pre-implementation period (where results were provided on paper and manually delivered to the TB clinic staff from the laboratory). The proportion of results available within 48 hours during the implementation period was significantly higher (97% using electronic platform, compared to 67% under the pre-implementation paper-based system)


100% adoption was observed during the pilot implementation.

Areas for further investigation: How will adoption differ under normal, 'non-pilot' conditions? Are there any differences in adoption across variables such as facility type, location etc? What strategies or support are needed to promote adoption in other settings?


Interviews with TB patients held after the pilot found that the system was acceptable to most patients, who appreciated the convenience of the electronic communication and the ability to access test results without visiting the clinic. TB clinic staff also found the system acceptable, noting that the electronic communication assisted patients to start TB treatment earlier.

Some of the target staff experienced technical difficulties while using the mobile devices during the initial phase of implementation but this improved with experience. Some patients also reported that they encountered technical challenges that prevented them from retrieving their results either due to technical issues of lack of comprehension of steps to retrieve results. Patients who received results reported that it facilitated their return clinic visit.

Areas for further investigation: What strategies or further support are required to support future roll out and implementation of the platform? Does the acceptability of the platform differ across patient variables, such as age, technological literacy, level of education etc? If so, what strategies are needed to address these barriers?


What is the best strategy or approach for rolling out the platform? What are the ongoing and long-term costs of maintaining the platform? How does the level of use of the platform change across time? Are there differences in key programme indicators over time among sites that are/not continuing to use the platform?

Study design

This study used mixed methods including in-depth interviews and secondary data analysis of routinely collected programme data.

Research outcomes

This study explored various outcomes related to implementation and effectiveness.

  1. Feasibility was defined as the level of concordance between paper and electronic register data.
  2. Acceptability was defined as the perceived acceptability of the mHealth intervention among TB patients and health care providers.

Effectiveness was defined as the proportion of treatment initiation within 48 hours reported during the implementation period, compared to the pre-implementation period.

Study population

The study was conducted among TB health care providers and patients selected from the two primary health clinics (PHC) in inner-city Johannesburg where implementation of the mHealth intervention was occurring.


The mHealth intervention was implemented and evaluated in two PHCs serving high TB-burden communities located in Johannesburg, South Africa.

Intervention/implementation strategy

The mHealth intervention was developed by the study team comprising staff from the Aurum Institute, the School of Public Health at the University of Witwatersrand, Johns Hopkins University, and National Institute of Communicable Disease in collaboration with the District TB Programme, PHC managers and the National Health Laboratory services.

The mHealth intervention was designed to achieve the following goals: i) replace paper-based data collection and reporting (via TB register) with a point-of-care mobile app that enabled direct data entry; ii) automate the electronic delivery of TB laboratory results to PHCs from the central lab; iii) display the status of all registered TB patients through a digital dashboard; iv) deliver TB sputum results electronically to clinic TB staff, and; v) provide electronic, personal identification number (PIN)-protected sputum result notifications directly to patients via cell phone messaging.

The aim of the intervention was to reduce time and effort associated with TB data reporting, enable timely and automatic access to Xpert MTB/Rif TB test results to both TB providers and patients and assist TB providers to track and facilitate patient progress through the TB care continuum.

The mHealth intervention allowed patients to receive results of TB screening directly via mobile phone messaging. The content of these results messages was developed and refined by various staff involved in TB care. Patients who opted to receive their results via text were asked to select a four-digit PIN during their initial clinic visit. When the results were ready, patients received a text asking them to respond with their PIN. If the correct PIN was entered, one of the following messages were provided: 1) "Your result is MTB negative, please visit clinic <name> if symptoms persist"; 2) "Your result is MTB positive, please visit clinic <name> in order to start TB treatment"; or 3) "Your result is unsuccessful specimen, please visit clinic <name> to provide further specimens". Patients who chose not to receive their results by text receive a message asking them to attend the clinic to collect their results.

The mHealth intervention was implemented in two phases. The first phase was a 'pre-implementation' phase from January to March 2015 where training was provided to necessary staff. This phase also included a 'run in' period, where staff were able to begin using the intervention in a trial manner to identify any final issues or training requirements. The implementation period occurred between February and April 2016.

Data collection, management and analysis

Sampling and recruitment

During the implementation period (Feb–April 2016), patients who were aged 18 and above and had been referred for TB investigation including sputum testing at the two PHC implementation sites were sequential recruited. All patient data during this period was captured using the mHealth intervention and used for secondary analysis.

At seven months post-implementation, purposive sampling was conducted to recruit patients at the two implementation sites from the following four categories: patients tested TB negative who i) did, and ii) did not received their results by text message; patients who tested TB positive and iii) did and iv) did not receive their results by text message

Data collection methods

The study undertook various data collection methods for each of the research outcomes:

  1. Feasibility: To observe the level of concordance between data captured in the paper-based register and mHealth intervention, the study team reviewed and compared the following elements in both data sources for each of the eligible patients attending the clinic during the implementation period:

    • Completeness of the electronic records in the mHealth intervention (i.e. have all the components typically captured in the paper-based register been recorded electronically?).
    • Proportion of patients with recorded specimen results.
    • Proportion of patients with positive Xpert MTB/Rif results.
    • Proportion of patients with a positive Xpert MTB/Rif result who also had a subsequent smear result recorded.

    The electronic patient records were also investigated for successful receipt of electronic laboratory results at the clinic using the mHealth application, and success in automatic sending of text notifications and results to patients.

  2. Acceptability: Semi-structured interviews were conducted with eligible TB patients and providers at seven months post-implementation. Patient interviews took approximately 30 minutes and were conducted in the language of the participants' choosing (either English, isiZulu, Sesotho or Setswana). All provider interviews were conducted in English among providers who used the mHealth application at each clinic and the regional TB coordinator. All interviews were recorded and transcribed and translated into English (if necessary).
  3. Effectiveness: Data were extracted from the paper-based register and mHealth intervention to compare the proportion of patients who were initiated on TB treatment within 48 hours (the cut-off point used by the NTP to describe ILFU), and within 28 days (to enable comparison with prior studies from South Africa on ILFU)

Data management and analysis

Analysis was conducted separately based on the individual research outcomes:

  1. Feasibility: A McNemar test was used to compare indicator data extracted from the paper-based registers and the mHealth intervention during the implementation period. A statistically significant difference (as evidenced by a p-value <0.05) between the two sources was taken as an indication of difference in data collection accuracy, allowing conclusions to be drawn regarding feasibility of the mHealth intervention to support the monitoring and evaluation requirements of the district TB programme.
  2. Acceptability: Transcribed interviews were uploaded and analysed using Nvivo. The normalization process theory (NPT) was used as a framework to guide analysis of the interviews. The NPT domains included: i) 'cognitive participation', which was defined as the provider's willingness to capture patient data into the application and search for results, or the patient's willingness to use the application to retrieve and receive results; ii) 'collective action', which was defined as: a) the usability of the mHealth application and how it integrates within existing systems; b) the ease of use of the mHealth application by providers and patients; and iii) 'reflexive monitoring', which was defined as the patient's or providers' experience in using of the device to retrieve results. The study team added an additional domain of 'confidentiality', which was defined as patients' confidence in the system's ability to disclose their results appropriately.

The transcripts were coded by a study team member in line with the four domains.

  1. Effectiveness: The proportion test for two samples was used to compare the proportion of patients initiated within 48 hours and 28 days during the pre-implementation and implementation periods, respectively. A statistically significant difference (p value <0.05) was taken to suggest a true difference in proportions between the two data sources. Differences between time to TB treatment in the pre-implementation and implementation phase was assessed using a Mann-Whitney test

Based on: Maraba N, et al (2018). Using mHealth to improve tuberculosis case identification and treatment initiation in South Africa: Results from a pilot study (

Share this story

See all case studies

Interested in using the toolkit and would like assistance?