Mobile application development for improving medication safety in tuberculosis patients: A quasi-experimental study protocol

The COVID-19 pandemic, the growth of smartphones, and the internet have driven the use of technology for monitoring TB patients. Innovation in management of TB patients is needed to improve treatment outcomes. The study was conducted to obtain a predictive model of medication safety and solution model for at-risk patients, and to improve medication safety through mobile applications. The research was conducted in 4 stages, namely qualitative, quantitative (cross-sectional), qualitative, and quantitative (quasi-experimental, post-test group control design). Data were taken at the Public Health Center in Jakarta, Indonesia. Samples were taken by cluster random sampling. For quantitative research, 2nd phase (n = 114) and 4th phase (n = 96) were analyzed using logistic regression. This study analyzed predictors of medication safety to assist in monitoring patients undergoing treatment. At-risk patients were educated using an algorithm programmed in the application.


Introduction
Tuberculosis is one of the world's top 10 causes of death [1]. In 2020, there were 845,000 estimated TB cases with 543,874 notified, 35% of cases unreported, and a treatment success rate of 84%. Cases of Multi-Drug Resistant TB increase annually, with 9,038 recorded in 2018 and 10,097 in 2019 [2]. Treatment failure results in drug resistance, leading to expense, lower cure rates, severe side effects, and longer treatment [3].
The length and complexity of TB therapy affects patient compliance. The most common TB drug therapy problems are adverse reactions (62.3%) and non-adherence (21.5%) which can be solved by therapeutic monitoring [4,5]. Adverse drug reactions cause non-adherence and drug resistance [6].
Under pandemic conditions services must prioritize patient safety, respond to COVID-19 and to keep TB services running [5,7]. Evaluation of a TB program in Indonesia during the pandemic showed medication nonadherence, patients not collecting sputum, cessation of referral laboratory examinations, and issues monitoring patient compliance [8]. a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 The WHO recommends using technology for drug ingestion control. Remote patient monitoring is rapidly increasing while 75.6%, of Indonesia's population uses mobile phones. Up to 47.7%-58.6% urban, 33.8% rural, and 76.2% of Jakarta's population-use the internet [9].
Studies suggest adherence interventions for improving outcomes, such as medication event reminder monitor (MERM) systems, electronic pillboxes, and text messaging [10,11]. Medication reminders appear to improve compliance and staff-patient relationships, but have not been shown to cause significant differences in treatment success [12].
In one study medication monitors (MMs) reduced missed doses [13]. In another study, patients using an electronic medication monitor (EMM) displayed no significant differences from patients without EMM [14]. More research is needed to assess effects of technology and intervention on TB programs [15].
Indonesia's Ministry of Health has developed applications such as SOBAT TB, which focus on independent TB screening, finding health facilities, and connecting the TB patient community. EMPATI-TB monitors and assists treatment of drug-resistant TB, while EMPATI CLI-ENT monitors patients' medication use via video, allowing patients to view medication history and consult with cadres or companions [16,17]. Meanwhile, the Tuberculosis Information System (SITB) is a web-based platform used by officers to report patients in treatment. Each application has advantages, but none uses a predictive model to prevent non-compliance.
The COVID-19 Pandemic has revealed the necessity of made application that remotely detects risks and monitors primary-care tuberculosis treatment.

Research questions
What are predictors of medication safety for primary-care tuberculosis patients?
How can an algorithm monitoring medication safety function within a mobile application? Can mobile applications influence medication safety? Objectives. Assessing effects of mobile applications on tuberculosis medication safety. Specific objectives.
a. Knowing medication safety predictors in primary-care tuberculosis patients b. Crafting a decision-making model for at-risk patients

Materials and methods
The conceptual framework of the research can be seen in Fig 1, which describes research phases 1, 2, and 3. Phase 4 is illustrated in Fig 2. These objectives were addressed using operational research, carried out sequentially in 4 stages (Fig 3) in June-December 2021. Details of each stage can be seen in Table 1. Table 2 describes the flow of quasi-experimental research (4 th stage). The informants involved in qualitative study consisted of various levels. Respondents were selected with various characteristics to enrich the information. Officers were diverse, representing different types of public health centers (sub-district/ward) and profession (general practitioner/nurse). Patients too were diverse, representing compliance 100% or under 100%, experienced the side effects (yes or no), and had families/other than families serve as drug swallowing supervisors. Data validation was carried out by theoretical triangulation. This sampling design is expected to describe the condition in Jakarta.
Cadres were involved in the 3rd phase of this study because they could provide additional information regarding the findings of the 2nd phase. Cadres were more aware of the technical conditions in the field, such as the condition of the patients' families and environmental conditions around them. Besides, they could play a role in reducing the stigma on patients, families, and communities. Cadres could also function as companions while patients were undergoing treatment and carry out contact investigations to track suspected patients until they were controlled and recovered.

Operational definition
The medication safety assessment included monitoring sputum examination, medication adherence, and side effects. It was measured by interview. Intervention group data was    recorded in the application. Failure to conduct a repeat sputum examination, medication compliance at <100%, and mishandled side effects led to risk.

Intervention protocol
Subjects were divided in 2. The intervention group received standard monitoring plus application usage. The control group received standard monitoring. Intervention was carried out for

PLOS ONE
2 months following start of treatment. Data was collected through questionnaires completed in the application (intervention group) or as Google forms (control group), at the study's beginning and end. The effect of applications were measured for 2 months. Early intervention is key, with germs decreasing effectively in the first 2 months of treatment. In this phase, patients often experience drug side effects [20].
Validity and reliability tests were carried out with a 35-person questionnaire. Data analysis was performed by bivariate test. Variables with p value < 0.25 were included in the multivariate test using logistic regression. Variables with p value < 0.05 in multivariate analysis were referred to as predictors. Insignificant variables (p > = 0.05) were excluded. A variable causing a > 10% change in odds ratio of other variables was included and became a confounder.
The application developed is called ERLINA (e-Empowerment Resource for Lowering Ignorance and Negligence Action in therapy) and is free to download through the Android playstore. A decision support system algorithm requires identifying predictors of medication safety alongside specific intervention models (Figs 5 and 6). The application was tested by IT experts and doctors, who assessed the system's functioning and content. The application's functions predict medication safety risks and provide reminders and education (Fig 7). Dashboards allow officers to monitor and provide feedback. Patients input medication, phlegm, and control data when the reminder sounds. Data is collected in the patient compliance chart visible to staff. Patients who miss > 6 doses are included in a non-compliance list in the staff menu. Side effects are reported by the patient to the officer and are automatically addressed by the application and staff.
Potential obstacles include 1. Varied patient absorption of information. Audio-visual media, straightforward language, and officer assistance help ameliorate this.
2. Real-time data monitoring without internet. In the absence of internet, a revision menu is provided, letting patients change data when they have internet access. Patient data is checked with the regularity of taking medication on the control card (TB01 card).
3. Lack of validation. Information content may not follow Public Health Center standards. Officers were involved in developing applications, especially in qualitative research stages 1 and 3. Public Health Center doctors helped develop validation tests to ensure the application could support TB services. 4. Data privacy. All data entered by the patient, excepting passwords, is visible to officers. At the start of the study and during registration, patients consented to data monitoring.

Ethical review
Before data collection, researchers offered information about procedures. Participants consented via electronic questionnaire. In in-depth interviews, verbal consent was recorded.

Results and discussion
Data from the first phase of the research is presented (Table 3) as a basis for developing a conceptual framework. The variables from the data above were selected and developed into operational definitions. The operational definition (Table 4) details helped compose the questionnaire. Grouping variables based on percentiles divides the data equally in each group.
Factors influencing medication safety include patient factors (demographics, social and family situation, comorbidities); staff factors (habits, patient relationships, social and medical conditions); health care facilities (infrastructure, access, information systems, remote services); organizational and service factors (protocol, organizational culture, workload, pharmacy service); external factors (physical environment, social environment, technology, infrastructure, policies); drug factors (effectiveness, side effects, patient adjustment to the regimen); and process (prescription, preparation, checking, storage, administering, information delivery, monitoring) [29,34].
Adequate treatment involves the right combination of drugs at the correct dosage, taken regularly and monitored for a sufficient duration [20]. Patient education consists of which drugs to take, the amount and manner of doing so, the possibility of adverse reactions, the right time to seek treatment, consequences of not taking medication properly, and prevention of TB transmission [35].
Medication safety involves preventing or repairing drug use injury [36]. Medication safety outcome indicators include medication errors, incident type, and impact [37].
Areas monitored include response to therapy, regularity of taking medication, and drug tolerance, including adverse drug reactions [38]. Digital interventions for medication safety are important given pandemic-induced service limitations. Predictions are expected to reduce medication errors, while the Decision Support System (DSS) [39], supports but does not replace decision-makers [40]. Performing predictive maintenance involves observing conditions and making timely diagnoses before failure occurs [41].
Remote monitoring collects patient data using technology [42]. Benefits include rapid detection, continuous monitoring, reduced costs, daily patient information, increased health/ emergency service efficiency, and assistance for limited-mobility patients [43]. c. The medication safety model, implemented in a mobile application, (measured in stages 1,2, and 3) involved a wide range of patients. In limited-equipment conditions, cadres helped ensure safety. e. The application was unintegrated with the tuberculosis recording system (SITB): patients enter medication schedules and checked sputum with manual controls, whereas SITB integration allows automatic data synchronization.
f. The application's impact after 2 months of treatment is unknown. More research is needed to improve ongoing application quality.
g. Application function was developed by continuously updating TB/health information, by tracking patients, especially patients who had moved to other health facilities, and through chat rooms and dialogues containing persuasive messages.

Conclusions
This research improves TB medication safety through mobile application development. It gives contributions to patients, staff, policymakers, and academics.

Measurement result and code
Medication safety The condition of being free from injury or potential injury due to errors in the process of using drugs [21,22]. The assessment includes the presence or absence of incidents in the monitoring process, namely monitoring the progress of treatment through sputum tests at the end of initial treatment, the medication adherence, and the handling of drug side effects [23].  c. Staff-patient interaction The relationship between the officers and patients, including patient trust, sufficient time provided by officers to patients for talk, and good response to them [29] Yes Health care factors (facilities, organizational, and service) a. Remote services The services without direct care that use information and communication technology such as schedule control reminders, online registration, remote monitoring, and consultations The ease of getting services, including appointment and service flow [29] Yes 0 = bad (<25 th percentile) c. Patient safety culture The organizational culture, namely characteristics and norms originating from within the organization [32] related to patient safety culture in this case, e.g. officers ensuring that patients understand the information provided, prioritizing patient safety in service, and motivating them to report errors while taking medication Yes d. Drug services The activities to meet patient needs related to free TB drugs [20,27] Yes e. Quality of service The quality of service is characterized by patient satisfaction [25], including the attitude of the officers, the fulfillment of the patients' therapeutic needs, and the impression of the building and layout Yes 1 = enough (<50 th percentile) 2 = good (<75 th percentile) 3 = very good (�75 th percentile) Family support The assistance (physical or moral) provided to TB patients, including reminding them to seek treatment and giving spiritual encouragement [33] Yes Treatment The perception of the effect of the drugs on recovery, the patients' ability to deal with the side effects, and the difficulty in swallowing the drugs [29] Yes