JMIR Publications

ABSTRACT

Background:

Background:

Hospitals have been identified as very high-risk places for Covid-19 transmission between health workers and non-covid-19 patients. Health care workers are the most at risk population to contract and transmit the infection especially to the already vulnerable non-Covid-19 patients. In Low-income countries, routine testing is not feasible due to the high cost of testing and the high turn around of results therefore presenting the risk of un-controlled transmission within the non Covid-19 treatment wards. This challenge necessitated the development of an affordable intermediary screening tool that would enable early identification of potentially infected health care workers and for early real time DNA-PCR testing prioritization. This would limit the contact time of potentially infected health care workers with the patients but also efficiently utilize the limited testing kits. Materials and methods: Using the WHO, Ministry of Health of Uganda guidelines on the case definition of Covid-19, we developed a screening questionnaire tool for risk assessment of Covid-19. Specific signs and symptoms were weighted based on how prevalent they were among Covid-19 patients and subsequently an algorithm developed for the various case scenarios of Covid-19. Risk sores were computed based on the symptoms and contact history and a daily risk category assigned based on the risk score. The questionnaire, flow charts and algorithms were then integrated into an android mobile application. Following the launch, Health care workers would submit their daily risk scores and high-risk staff would be selected for testing and further intervention including treatment.

Results:

The primary result of this research project was the development of a mobile based daily early warning system for in-hospital transmission of Covid-19. Conclusion: Mobile screening applications can offer an intermediate screening tool for prioritizing which health care workers should undergo routine DNA-PCR for Covid-19 in health care systems where DNA-PCR testing of all health care workers is not feasible. The daily log of risk scores enables trend monitoring among different staff and hence can detect epidemic clusters on specific wards therefore enabling timely intervention.

Objective:

The aim of this research was therefore to create an intermediary daily early warning screening tool that would identify health care workers with the highest likelihood of having contracted Covid-19 and have these prioritized for DNA-PCR testing and timely intervention. In this way, the exposure time of potentially infected health care workers to patients would be significantly reduce thus minimizing the in-hospital transmission of Covid-19 on the general and non-COVID-19 treatment units.

Methods:

IMPLEMENTATION: The Early Warning System for in-Hospital transmission of Covid-19(EWAS) mobile application development comprised of three phases which included; a) risk assessment tool and algorithm development, b) software design and development and finally c) Application use. Risk screening tool and algorithm development: In order to develop are standard risk assessment tool, we identified the World Health Organization (WHO, 2020b) and Ministry of Health of Uganda guidelines(MOH-UGANDA, 2020) for the Clinical Case definition of Covid-19. Based on the above case definitions, we developed an 8-question risk assessment and scoring tool comprising of multiple-choice answers of which the participant could only select a single answer for each question (see additional file 1). This was done to minimize user fatigue from having long examination style assessment tools commonly used for screening. Each selected answer was assigned a score based on whether it comprised of a positive symptom for Covid-19 or not. The 8 questions were grouped into two sections of which section I comprised of the Symptom Score (SC) consisted of 7 questions evaluating for the general health condition as well as the presence of Covid-19 like symptoms. The overall highest score for Section one was 155 with the lowest score being 30. See figure 1 Section two comprised of the Contact Score (CS) consisting of one multiple choice question describing the possible scenarios for a Contact history for the participant in the last week. Each answer was weighted with the highest score being 50 while the lowest being 15. See figure 1. In order to include potential contact with asymptomatic carriers, even when the participant was sure not to have been in contact with a Suspected Covid-19 positive patient, they couldn’t be awarded a zero score. In order to avoid falsely alarming high-risk scores, each of the 8 questions were weighted basing on how prevalent the given symptoms were among the Covid-19 patients in Uganda (Kirenga et al., 2020). Figure 1: Daily assessment tool Section one and Section two each contributed 50% to the overall Risk score which was calculated as a percentage of the participant score to the highest score. Hence in order to calculate the Overall patient risk score, the Patient Symptom Score (SS) and the Contact Score (CS) would be entered into the formula Patient Overall Risk score = [(SS/155) X 50] + [(CS/50) X 50] Second algorithm was to calculate a trends analysis of the daily logged risk score. Based on the Weekly trends of the risk scores, the participant was issued with a weekly risk badge. The Risk badges would generally reflect the risk status of the patient as regards to the Covid-19 symptom and contact score. The intention for the risk badge would be that Participants who had a High-risk Badge would then be identified and prioritized for a weekly DNA-PCR test. Hence if a participant consistently had a high-risk score through the week, their risk badge would identify them as high risk and necessitating urgent DNA-PCR testing. The application also identified clusters of participants with high-risk badges and if these were on one ward, would quickly highlight the affected ward as a potential for a cluster outbreak within the hospital with the intention of timely intervention before putting the lives of the patients and other health care workers at high risk. Once the above risk assessment tools and algorithms were completed, they were reviewed by a clinic al epidemiologist and a Physician directly participating in the treatment of Covid-19 patients at Mulago National Referral Hospital and once cleared, ethical clearance was obtained from the Research and Ethics Review Board of Mulago Hospital. Subsequently, the software developers team undertook the development of the mobile application with multiple pre-tests to ensure good end user experience and relevancy. The Application was piloted for a week among 10 participants and the user feedback was used to make final adjustments prior to the official launch. At the completion and launch of the software development, the Mobile application was introduced to over 100 health care workers involved in direct patient care in the Directorate of Surgery of Mulago hospital. The software application use was supervised and monitored for two months during which period various aspects including user friendliness, fatigue and reliability were evaluated. Health worker experiences and feedback was also obtained to aid further refinement. In order to improve user friendliness, the application risk assessment tool was made to be as simple as possible with minimal drop down and no direct user input.

Results:

The primary intended outcome of this research project was to create an intermediary early warning tool that would help identify symptomatic health care workers to prevent in-hospital transmission of Covid-19 while also identify health care workers who would require to have a DNA-PCR without necessarily having to routinely mass test all health care workers which is costly and unsustainable. With the mobile application in place, health care workers are now able to obtain a daily risk score which is objective and quantifiable. This daily symptom score when used, can help pick the disease process way before the health care worker would have thought that they have Covid-19 and hence limiting the contact time of a potentially infected health care worker with a patient. The health worker can also access a graphical representation of their risk trends for the past week and month (figure 2b). This risk trend is the basis of the Risk Badge (figure 2c) that is assigned to the Health Worker every after 5 days. The risk badge therefore forms the basis for the need for further analysis and intervention including DNA-PCR testing and mitigation measures such as self-isolation. The Software also enables notification and messaging through which an affected health care worker can send out an urgent request for support from the Covid-19 treatment team including request for a dispatch team. The application also comprises of a repository for keeping and tracking the Covid-19 DNA-PCR results which the user uploads onto their database. DISCUSSION: In-hospital transmission of Covid-19 among health care workers and patients is still a recognized challenge in prevention of Covid-19 transmission (Çelebi et al., 2020) especially to the already vulnerable populations. The most ideal way of minimizing transmission would have been routinely performing point of care DNA-PCR testing of all health care workers and patients on the various non-Covid 19 treatment units such as the surgical wards. (Zhu et al., 2020) This has globally been recognized not to be feasible especially in Low an Middle Income (LMIC) countries and therefore in-hospital transmission of Covid-19 remains a big challenge to date (Afzal, 2020). The Early warning system (EWAS) for in-hospital transmission of Covid-19 can therefore come in to reduce the transmission rates on these wards by identifying potentially infected and therefore transmitting health care workers working as a daily screening tool. In highly transmissible diseases like Covid-19, mHealth applications have been described to significantly improve disease screening and symptom monitoring(John Leon Singh, Couch, & Yap, 2020). Daily screening tools have an important role to play as they may identify someone before the person actually becomes suspicious of being infected. In China, mobile Health Applications were shown to significantly reduce disease transmission however these were designed for use in community settings(Wu et al., 2020), while in Sweden a similar mHealth application that tracks logistic use including PPE, and patient care in 5 hospitals has been developed and in use(Dress, 2020). Overall, there is increased utilization of mHealth applications in disease tracking, symptom monitoring as an adjunct to the existing guidelines in the management of Covid-19(Collado-Borrell, Escudero-Vilaplana, Villanueva-Bueno, Herranz-Alonso, & Sanjurjo-Saez, 2020). Very few mobile Health Applications have been developed and actively used in Covid-19 management in Africa. The availability of this mobile Health self-administered screening and risk assessment software in Low income Countries may therefore be relevant in day to day disease transmission prevention as it forms the first in-hospital active screening tool capable of identifying potentially infected health care workers. Without this application in our health care system, essentially leaves in-hospital transmission at the discretion of the affected health worker which may form a potential loop hole for limiting disease transmission. This therefore implies that the mobile application doesn’t replace testing or any other established guidelines in the diagnosis of Covid-19 however it can be used in combination as an intermediary adjunct with these existing guidelines to increase the likelihood of minimizing the in-hospital transmission of Covid-19. By being able to identify the health care workers most likely to be infected, the EWAS application can therefore be able to maximize the highest potential benefits in screening and prioritizing the limited testing resources of DNA-PCR for the health workers in large hospital settings like the Mulago National Referral Hospital. Despite the advantages of the EWAS application, health care provider compliance remains a big challenge especially as the pandemic progresses. The general fear associated with Covid-19 has waned and so is the compliance to the majority of established Standard operating procedures including compliance to the EWAS application. This complacency is partly responsible for the recent spikes in the Covid-19 spread however the increasing spread calls for more aggressive campaigns and adaption of all mitigating measures including the adoption of this the EWAS tools on the non-Covid-19 treatment units. Secondly to successfully utilize these tools would require that all health care workers have access to a smart phone as well as mobile data connection. In Uganda, Smart phone coverage is approximately 42%(Kanaabi, 2020) with over 70% smart phone coverage in urban centers. In Mulago Hospital, based on a preliminary survey, smart phone coverage was approximately 90% among health care workers but despite the good coverage, the ability to fully utilize the smart phones including installation of an application and registration of the user required the establishment of support services. This to a great extent may limit the realization of the full potential of integrating mobile applications including the EWAS in disease surveillance and transmission prevention.

Conclusions:

The Early Warning System for in-hospital transmission of Covid-19(EWAS) significantly reduces the interaction of symptomatic health workers with patients therefore minimizing disease transmission on the wards while also identifying and optimizing routine DNA-PCR testing by identifying the health care providers who actually require to have the test done. This application therefore compliments the existing measures to prevent the transmission of Covid-19 in hospital settings. Clinical Trial: N/A