Toward Developing a Standardized Core Set of Outcome Measures in Mobile Health Interventions for Tuberculosis Management: Systematic Review

Background Tuberculosis (TB) management can be challenging in low- and middle-income countries (LMICs) not only because of its high burden but also the prolonged treatment period involving multiple drugs. With rapid development in mobile technology, mobile health (mHealth) interventions or using a mobile device for TB management has gained popularity. Despite the potential usefulness of mHealth interventions for TB, few studies have quantitatively synthesized evidence on its effectiveness, presumably because of variability in outcome measures reported in the literature. Objective The aim of this systematic review was to evaluate the outcome measures reported in TB mHealth literature in LMICs. Methods MEDLINE, EMBASE, and the Cochrane Database of Systematic Reviews were searched to identify mHealth intervention studies for TB (published up to May 2018) that reported any type of outcome measures. The extracted information included the study setting, types of mHealth technology used, target population, study design, and categories of outcome measures. Outcomes were classified into 13 categories including treatment outcome, adherence, process measure, perception, technical outcome, and so on. The qualitative synthesis of evidence focused on the categories of outcome measures reported by the type of mHealth interventions. Results A total of 27 studies were included for the qualitative synthesis of evidence. The study designs varied widely, ranging from randomized controlled trials to economic evaluations. A total of 12 studies adopted short message service (SMS), whereas 5 studies used SMS in combination with additional technologies or mobile apps. The study populations were also diverse, including patients with TB, patients with TB/HIV, health care workers, and general patients attending a clinic. There was a wide range of variations in the definition of outcome measures across the studies. Among the diverse categories of outcome measures, treatment outcomes have been reported in 14 studies, but only 6 of them measured the outcome according to the standard TB treatment definitions by the World Health Organization. Conclusions This critical evaluation of outcomes reported in mHealth studies for TB management suggests that substantial variability exists in reporting outcome measures. To overcome the challenges in evidence synthesis for mHealth interventions, this study can provide insights into the development of a core set of outcome measures by intervention type and study design.


Introduction
Tuberculosis (TB) is one of the deadly infectious diseases that have claimed millions of lives worldwide. According to the World Health Organization (WHO), globally, there were 10.4 million new TB cases causing approximately 1.2 million deaths in 2016 [1,2]. The mortality rate of TB is disproportionately higher in low-and middle-income countries (LMICs). Over 95% of TB deaths occurred in these countries, and 7 LMICs (India, Indonesia, China, Philippines, Pakistan, Nigeria, and South Africa) accounted for 64% of the total burden [3]. In fact, previous studies have shown that there is empirical evidence of positive associations between poverty indicators and TB incidence both at the macro and individual levels [4]. Considering the vicious cycle of poverty and TB, alleviating the burden of TB is more challenging for the LMICs because it requires adequate resources for "prolonged treatment with multiple drugs [5]." The 6-month course of first-line therapy can be burdensome with the possibility of adverse reactions and the treatment of multidrug-resistant (MDR)-TB requires more toxic and expensive drugs [6]. For example, the cost of bedaquiline, a second-line medication to treat MDR-TB, was US $3000 per treatment in middle-income countries and US $900 in low-income countries [7]. In fact, premature discontinuation of the treatment, which can lead to MDR-TB, is common among TB patients not only for its toxicity but also for socioeconomic costs associated with it [8]. Therefore, the management of TB is notoriously difficult especially in LMICs.
In this context, using mobile devices for TB treatment has been recognized as an innovative approach for LMICs where mobile subscription rates have dramatically increased over the past decade. Mobile health (mHealth) interventions involving mobile devices in the management of TB have the potential for reducing costs of information delivery and improving the quality of communication [9]. mHealth can be useful for TB treatment adherence support such as short message service (SMS) for medication reminders or mobile apps for remote directly observed treatment (DOT) strategy [10,11].
Despite the potential of mHealth interventions for improving TB management, the empirical evidence on its effectiveness is mixed. Some studies have demonstrated the effectiveness and feasibility of the mHealth interventions for TB [12,13], whereas others have shown no significant impact [14,15]. Moreover, no study has attempted to synthesize the results quantitatively to rigorously evaluate the effectiveness of mHealth in TB management. Presumably, one of the reasons for such difficulty in synthesizing and evaluating the findings comes from wide variations in the outcomes reported from the mHealth studies for TB.
To respond to this knowledge gap, this study aimed to systematically review previous mHealth studies for TB management and critically evaluate and categorize the outcome measures for different mobile technologies and study designs. The goal of this study was to provide researchers insights into the development of a core set of outcome measures for mHealth interventions intended to improve TB treatment adherence. In doing so, the study can facilitate the evidence synthesis of mHealth interventions for TB.

Search Strategy and Review Process
Electronic databases (MEDLINE, EMBASE, and Cochrane Database of Systematic Reviews) were searched to identify peer-reviewed studies of mHealth interventions for TB. The systematic search was supplemented by reviewing relevant review papers identified from the initial search. The search strategy for the study population includes key terms describing LMICs such as "resource poor" or "developing country." The search strategy for the mHealth intervention combined multiple keywords such as "mHealth" and "text-messaging." As for the target disease, "tuberculosis," "TB," "multi-drug resistant tuberculosis," and "MDR-TB" were used as search terms. No restrictions were applied to the publication type or publication date, but the language filter was applied to identify studies published in English. The search included articles published up to May 2018. The full search strategy is available in Multimedia Appendix 1.
Furthermore, 3 authors (ShL, YL, and SmL) independently reviewed the retrieved studies throughout the selection process. Each study identified from the databases was screened by 2 reviewers and then a full-text review was conducted for the potentially eligible studies. The disagreement on the selection process was resolved by the other authors who were not involved in the review of the specific study under discussion.

Eligibility for Review
The inclusion criteria for this systematic review were as follows: First, studies conducted in the context of LMICs, as defined by the World Bank's income cutoffs [16]; second, studies involving an intervention using mobile devices (ie, mHealth intervention); third, the target disease of the study should be TB or MDR-TB; fourth, studies designed to evaluate the effectiveness or benefits of mHealth interventions for TB (eg, observational study, mixed-methods study or implementation project, randomized controlled trial [RCT]); fifth, studies reporting more than one type of outcome; finally, only full-text studies published in English were considered eligible. In addition, the authors attempted to identify individual studies from reviews or systematic reviews, which were included in this study.

Data Extraction and Analysis
The qualitative synthesis of evidence focused on the outcome measures reported in each type of mHealth intervention. Information about the study setting, mHealth technologies used, target populations, and types of outcome measures was extracted. To classify the diverse types of detailed outcome measures, the following categories were used: (1) treatment outcome; (2) treatment outcome as defined by WHO; (3) adherence; (4) process measure; (5) perception; (6) technical outcome; (7) health outcome; (8) quality of life; (9) knowledge; (10) cost-effectiveness; (11) cost; (12) psychosocial outcome; and (13) mortality. Some explanations about these measures are provided in the next few paragraphs.
The treatment outcome includes any outcome measure that deals with the result of TB treatment, such as sputum smear conversion or microscopy test result. The treatment outcome following the WHO definition was separately categorized [17]. The WHO definition was developed to make a distinction for treatment outcomes between the drug-susceptible TB and drug-resistant TB, which are mutually exclusive groups. According to the WHO definition, any patient with TB should belong to either group and then 1 of the 7 treatment outcome cohorts: (1) cured; (2) treatment completed; (3) treatment failed; (4) died; (5) lost to follow-up; (6) not evaluated; and (7) treatment success. The WHO definitions for each of these 7 categories differ between the drug-susceptible TB and drug-resistant TB as described in Multimedia Appendix 2.
The adherence outcome includes medication adherence or treatment adherence. The process measure is any outcome measure related to treatment process, including the receipt of diagnostic test, attendance to appointments, or reporting of adverse events. The perception indicates any outcome measure that captures the user's thoughts on mHealth for TB management. The technical outcome relates to the outcome measures that investigate the technical feasibility such as processing times or system installation. Health outcome, quality of life, knowledge (eg, patients' understanding of the disease or the technology), cost-effectiveness (ie, the extent to which an alternative provides value for money), cost (ie, costs associated with an intervention from different perspectives), psychosocial outcome, and mortality outcomes are additional categories which are self-explanatory.

Risk of Bias: Quality Assessment
To evaluate the quality of individual studies included for our review, risk of bias was assessed with the existing tools. As this systematic review includes various types of studies, it is important to have a coherent set of quality assessment tools for different study designs. Therefore, we used the modified version of the Critical Appraisal Skills Program (CASP) that provides the checklists specific to various types of studies ranging from RCTs and qualitative studies to economic evaluation [18]. In case of mixed-methods studies whose CASP tool has not been developed yet, the quality assessment criteria for mixed-methods studies from the previous study were employed [19]. The quality assessment of studies is presented in Multimedia Appendix 3.

Overview of Included Studies
Among the 312 studies identified after removing duplicates, 260 articles were excluded during the screening process based on the titles and abstracts. Therefore, 52 articles were assessed for eligibility through a full-text review. Of those, 27 studies were included for the qualitative synthesis of evidence. A flow diagram for the selection process, based on the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines, is provided in Figure 1 [20]. Table 1 presents the results of qualitative synthesis of the mHealth studies for TB management. Approximately, half (15 out of 27) of the studies were conducted in African countries. The study designs were diverse, including 6 RCTs, 5 mixed-methods studies, 1 cohort study, 3 qualitative studies, 4 observational studies, 6 implementation projects, and 2 economic evaluation studies. The types of mHealth technologies utilized were diverse as well: 12 studies employed SMS, 6 studies used mobile app, 5 studies used SMS plus other technology, 3 studies utilized phone calls, and only 1 study applied mHealth for mobile data collection. With regard to the study population, a majority (20 out of 27) of the studies were targeted for TB patients or TB/HIV patients, but there were a few studies that examined the experience of health care workers or general patients at the clinic for TB test results notification.

Qualitative Synthesis of Evidence
In terms of the outcome measures, there was a wide range of variants in their definitions even within each category of the outcomes. For instance, both Mohammed et al and Bediang et al defined treatment success as the primary outcome measure of SMS intervention for TB medication adherence in their RCT, but they defined treatment success differently [15,22]. Mohammed et al defined it as "the sum of patients clinically reported as cured (ie, a patient whose sputum smear or culture was positive at the beginning of treatment but who was smearor culture-negative in the last month of treatment and on at least one previous occasion) or treatment completed (ie, a patient who completed treatment but who does not have a negative sputum smear or culture result in the last month of treatment and on at least one previous occasion)" [22]. On the contrary, Bediang et al defined treatment success as "having completed 6 months treatment and having negative sputum smears at 5 months" [15]. Table 2 summarizes the types of outcome measures reported in the included studies, using the categories defined in this review. The most frequently reported outcome type was treatment success. Approximately half of the studies reviewed (14 out of 27) included treatment outcome but only 6 studies among them followed the WHO definition for the treatment outcome (Multimedia Appendix 2). The second most frequently reported category was perception on the mHealth intervention (13 out of 27). Other categories reported were diverse and included technical outcome, medication or treatment adherence, process measure, etc. However, there was substantial variability within each category of outcome, as shown in Table 1. For example, acceptability and satisfaction within the perception category were defined differently from one study to another [15,[28][29][30]32,35,36,40,44]. Also, SMS-only intervention studies did not focus on technical outcome [12,15,22,29,30,32,35,36,42,43,45,46] whereas studies involving other mHealth technologies such as an app or mobile data collection did so [21,28,33,37,[39][40][41]. On the contrary, outcomes related to cost or cost-effectiveness were reported only via studies involving SMS [29,45,46].

Phone calls or interactive voice response (3 studies)
Mobile data collection (1 study)

Principal Findings
This systematic review critically evaluated the outcomes reported in mHealth studies for TB management in LMICs. The reason why rigorous evidence synthesis is warranted is that recent literature for TB reports mixed results despite the rapid implementation of mHealth technology for TB management.
The fragmented pieces of evidence on the effectiveness partly resulted from the wide variations in the definitions of outcome measures in TB mHealth interventions. Even though treatment outcome has been reported by many studies, they often did not adopt the standard definition recommended by the WHO [17].
The WHO definition of TB treatment outcome is part of an effort to standardize outcome measures for TB at the global level. To promote the use of standardized sets of outcome measures for TB, WHO provided the standard definitions and classifications of TB in terms of diagnosis or treatment outcomes [17]. The intention for this standardization effort was to coordinate international comparison of TB treatment outcomes through health information systems. However, the findings of our review revealed that mHealth studies for TB have not comprehensively adopted this standardized approach for TB treatment.  [46].
Another issue to consider is related to evidence for cost. As this study suggested, insufficient evidence exists in terms of cost-effectiveness or cost of mHealth interventions for TB management. Those previous studies that have evaluated the cost-effectiveness or cost of mHealth interventions only considered SMS as mHealth channels and did not consider or evaluate other mHealth channels and technologies. However, other mHealth channels and technologies such as mobile apps or global positioning system are now available for TB patients [48]. Therefore, future studies need to assess the cost-effectiveness of such channels and technologies for improving TB management in LMICs.
This study has some limitations. First, the database used for identifying relevant studies is limited to the 3 most frequently cited sources, namely MEDLINE, EMBASE, and Cochrane Database of Systematic Reviews. Grey literature or other sources of information can supplement our findings. To complement this limitation, we attempted to identify additional related studies from relevant systematic reviews searched from our study. Second, the effectiveness of the mHealth interventions for TB was not quantitatively evaluated because of the heterogeneity of the outcomes reported.
Despite these limitations, this study provides an overview of the currently reported outcome measures for mHealth interventions intended to improve TB management in the context of LMICs. The results from this review can be used as a starting point for discussion to adopt standardized definitions within different categories of outcome measures for future mHealth interventions for TB management in LMICs.

Conclusions
This systematic review of mHealth studies for TB suggests that substantial variability exists with regard to the definitions of outcome measures across studies. Our review highlights that a standardized method for measuring the different outcomes is warranted to improve comparability of outcome measures across studies for a more rigorous and reliable evaluation of the effectiveness of mHealth interventions for TB. In doing so, the coordination among researchers and the development of a core set of outcome measures based on standardized methods would be necessary. Our study provides useful information for researchers to better assess the effectiveness of mHealth interventions for TB. In addition, the study provides insights into the possibility of developing a core set of outcome measures by intervention type and study design based on a standardized or coordinated set of methods.