Malaria Notification With Mobile Phones in South Africa
Malaria Notification With Mobile Phones in South Africa
In order to establish whether mobile phone reporting could assist in all malaria cases being notified within the required time frame, a mobile phone technical framework was implemented and a nurse was hired to send reports over a smartphone from selected primary health care clinics in a malaria-endemic area in South Africa for eight months. The work was done in close collaboration with the Mpumalanga Malaria Control Programme, which normally receives the paper-based reports from the clinics included in the study. The mobile reporting did not replace the submission of the complete notification forms during this study. As timely reporting is valuable only if action is taken, an evaluation on whether reporting over the mobile phone led to timelier follow-up of the cases was also conducted.
Ethical approval was obtained from the University of the Witwatersrand Human Research Ethics Committee (protocol number M120666) and the Mpumalanga Provincial Government.
A mobile form collection and server solution was implemented, making use of open-source software configured to fit the specific use-case and requirements of the study. ODK Collect was used for mobile data entry, which runs as a Java-based application on an Android smartphone. The software allows for data to be stored on the mobile and sent at a later stage, should networks be temporarily unavailable. Mobile forms were authored using the XLSForm standard and form submissions were maintained using a custom implementation of Formhub running on a dedicated server hosted in South Africa. The reason why a dedicated server was set up for the study, was to not store patient-level data outside of South Africa (in the so called cloud). The Formhub web interface was only accessible over a secure connection using Secure Sockets Layer (SSL) with mobile form submissions secured in the same way. Access to form data and data export functionality was controlled on a per-user level, through appropriate authentication and authorization mechanisms configured in the Formhub implementation.
Patient-level data on each malaria case, as currently reported within the malaria control programme, were submitted to the central server on a form on the smartphone that had the same structure and fields as the paper malaria case notification form.
In this study a designated person reported from the health care facilities, as the regular staff, with their current staff complement and work descriptions, were not able to handle this kind of reporting. To make the reporting as authentic as possible, a nurse was hired for the task. This person was elderly (71 years), with very little experience of computers and no experience of using smartphones. She contacted the clinics each weekday to inquire about any new malaria case, in order to conform to the 24-hour reporting limit. If a case had been found, the nurse would travel to the clinic and then send the required information over the phone to the server. She was asked to keep a manual log of all irregularities during the whole study period.
As the chosen technical solution lacked a function for automatic notification of new data having been submitted, it was decided at the beginning of the study that the nurse should send a text message via SMS to the malaria team coordinator in the district, for each new submission that was made. The malaria team coordinator, in turn, forwarded this message to the local case investigator. This message comprised the name of the clinic, the patient's name and age, and date of malaria diagnosis. This basic information was entered into the malaria surveillance information system in the province upon reception.
The reporting was done from three primary health care facilities in a rural, resource-limited area of Mpumalanga Province with all the challenges regarding communications (fax and conventional phone often do not work), supply of electricity and transportation. The mobile phone reception in this area is good and many more people rely on mobile phones than on conventional land lines.
The reporting was done from 1 October, 2012 to 31 May, 2013 (eight months) to cover the usual malaria season in southern Africa.
The following measures were evaluated using a mixed methods approach: simplicity, flexibility, stability, timeliness, and acceptability. The usability of the framework, ie, how easy it was to use the device and the software was also evaluated.
A structured interview was conducted with all individuals who were involved in the mobile reporting, from the health care workers at the clinics and the nurse doing the reporting to the stakeholders at the district communicable diseases control office (n = 9). The evaluation also included an analysis of the manual logs that the nurse kept and a record on how often the phone had to be charged and how many technical troubles were encountered ('battery and phone trouble study'). The results of the interviews formed the basis for the evaluation of the simplicity, flexibility and stability of the software and mobile device. Based on these interviews, the usability and acceptability of the surveillance system was also evaluated.
The timeliness was evaluated with respect to two events:
The two indicators were investigated for the three clinics as well as for other clinics served by the same malaria case investigator for the study period. To control for other factors that might have affected the reporting, the difference in timeliness between the three study clinics to other clinics served by the same case investigator for the previous year when no study was conducted, was also compared. Additionally, the timeliness for all other clinics in the municipality, to use as an unbiased control, for both time periods was computed.
To enable these comparisons the following attributes for each case were collected:
An exploration into whether the district malaria control programme had received any paper forms from the three clinics without any reports being sent by the mobile device was performed.
Baseline data for all clinics in the area for the study period and the control period were provided by the Mpumalanga Malaria Control Programme in Mbombela.
Methods
In order to establish whether mobile phone reporting could assist in all malaria cases being notified within the required time frame, a mobile phone technical framework was implemented and a nurse was hired to send reports over a smartphone from selected primary health care clinics in a malaria-endemic area in South Africa for eight months. The work was done in close collaboration with the Mpumalanga Malaria Control Programme, which normally receives the paper-based reports from the clinics included in the study. The mobile reporting did not replace the submission of the complete notification forms during this study. As timely reporting is valuable only if action is taken, an evaluation on whether reporting over the mobile phone led to timelier follow-up of the cases was also conducted.
Ethical approval was obtained from the University of the Witwatersrand Human Research Ethics Committee (protocol number M120666) and the Mpumalanga Provincial Government.
Technical Set-up
A mobile form collection and server solution was implemented, making use of open-source software configured to fit the specific use-case and requirements of the study. ODK Collect was used for mobile data entry, which runs as a Java-based application on an Android smartphone. The software allows for data to be stored on the mobile and sent at a later stage, should networks be temporarily unavailable. Mobile forms were authored using the XLSForm standard and form submissions were maintained using a custom implementation of Formhub running on a dedicated server hosted in South Africa. The reason why a dedicated server was set up for the study, was to not store patient-level data outside of South Africa (in the so called cloud). The Formhub web interface was only accessible over a secure connection using Secure Sockets Layer (SSL) with mobile form submissions secured in the same way. Access to form data and data export functionality was controlled on a per-user level, through appropriate authentication and authorization mechanisms configured in the Formhub implementation.
Data Collection
Patient-level data on each malaria case, as currently reported within the malaria control programme, were submitted to the central server on a form on the smartphone that had the same structure and fields as the paper malaria case notification form.
In this study a designated person reported from the health care facilities, as the regular staff, with their current staff complement and work descriptions, were not able to handle this kind of reporting. To make the reporting as authentic as possible, a nurse was hired for the task. This person was elderly (71 years), with very little experience of computers and no experience of using smartphones. She contacted the clinics each weekday to inquire about any new malaria case, in order to conform to the 24-hour reporting limit. If a case had been found, the nurse would travel to the clinic and then send the required information over the phone to the server. She was asked to keep a manual log of all irregularities during the whole study period.
As the chosen technical solution lacked a function for automatic notification of new data having been submitted, it was decided at the beginning of the study that the nurse should send a text message via SMS to the malaria team coordinator in the district, for each new submission that was made. The malaria team coordinator, in turn, forwarded this message to the local case investigator. This message comprised the name of the clinic, the patient's name and age, and date of malaria diagnosis. This basic information was entered into the malaria surveillance information system in the province upon reception.
Study Site
The reporting was done from three primary health care facilities in a rural, resource-limited area of Mpumalanga Province with all the challenges regarding communications (fax and conventional phone often do not work), supply of electricity and transportation. The mobile phone reception in this area is good and many more people rely on mobile phones than on conventional land lines.
Study Period
The reporting was done from 1 October, 2012 to 31 May, 2013 (eight months) to cover the usual malaria season in southern Africa.
Evaluation
The following measures were evaluated using a mixed methods approach: simplicity, flexibility, stability, timeliness, and acceptability. The usability of the framework, ie, how easy it was to use the device and the software was also evaluated.
Qualitative Evaluation
A structured interview was conducted with all individuals who were involved in the mobile reporting, from the health care workers at the clinics and the nurse doing the reporting to the stakeholders at the district communicable diseases control office (n = 9). The evaluation also included an analysis of the manual logs that the nurse kept and a record on how often the phone had to be charged and how many technical troubles were encountered ('battery and phone trouble study'). The results of the interviews formed the basis for the evaluation of the simplicity, flexibility and stability of the software and mobile device. Based on these interviews, the usability and acceptability of the surveillance system was also evaluated.
Quantitative Evaluation
The timeliness was evaluated with respect to two events:
Notification: whether the time from diagnosis to entering the case information into the provincial malaria information system – thus making the information available to relevant public health authorities – was affected with mobile reporting in place. The time it took for the basic information sent via SMS to be entered, as well as the time between diagnosis and the complete information being entered were evaluated.
Follow-up: whether the time it took from diagnosis to case follow-up was affected by the mobile reporting. The number of days it took for the cases to be followed up after diagnosis was measured. The proportion of cases that were followed up within the required time (48 hours), that is, within zero, one or two days from notification was calculated. Only access to dates and not the hours was available, and therefore an assumption was made that if a case was followed up within two days, this complied with the 48 hours' recommendation. In practice this may correspond to some additional hours. The notification date was assumed to be equal to diagnosis date, as a nurse would fill out a form immediately upon diagnosis.
The two indicators were investigated for the three clinics as well as for other clinics served by the same malaria case investigator for the study period. To control for other factors that might have affected the reporting, the difference in timeliness between the three study clinics to other clinics served by the same case investigator for the previous year when no study was conducted, was also compared. Additionally, the timeliness for all other clinics in the municipality, to use as an unbiased control, for both time periods was computed.
To enable these comparisons the following attributes for each case were collected:
Date of diagnosis.
Date of receipt of notification SMS.
Date of entry of paper-based form data into the provincial malaria information system.
Date of case follow-up.
An exploration into whether the district malaria control programme had received any paper forms from the three clinics without any reports being sent by the mobile device was performed.
Baseline data for all clinics in the area for the study period and the control period were provided by the Mpumalanga Malaria Control Programme in Mbombela.