Evaluation of Short Message Services (SMS) based data collection system for the surveillance in Nepal.

Abstract

Introduction: Natural disasters increase the risk of morbidity and mortality caused by communicable disease without a well-understood mechanism. The devastating earthquake in Nepal had escalated the threat of disease outbreaks in these areas. Nepal is endemic for many communicable diseases including cholera. Evidence from past experiences has established the importance of infectious disease surveillance to protect survivors during disasters. Real-time monitoring of morbidity can provide early signs of outbreaks, providing an opportunity to prevent outbreaks. Epidemiology and Disease Control Division (EDCD) of Nepal Government started syndromic surveillance right after the earthquake along with pre-existing Early Warning and Reporting System (EWARS). Timeliness and Completeness of reports were the two major problems faced by these surveillance systems. The adaptation of electronic records had shown improvement in timeliness and completeness of reports as well as increased breadth and depth of information collected. The ICD coded health record used for surveillance had been demonstrated as an effective method. Health Management Information System (HMIS)s register number 1.3 (HMIS1.3), records socio-demographic, geographic information along with ICD 10 coded working diagnosis, making it a perfect record for syndromic surveillance. Hence, the study explored the opportunity of using digitalized HMIS1.3 with minimal modification and transmitted via SMS for a surveillance system. Objective: This study aims to develop SMS based syndromic surveillance system and evaluate the feasibility of SMS-based health record collection for surveillance. Methodology: For the purpose, mobile phones with the android platform were selected, as they are cheap and easily available in developing the world. Opendatakit (ODK) collect, widely used open source data collection app developed by the University of Washington was customized to incorporate SMS capabilities. The de-identified individual record form outpatient register (HMIS 1.3) was collected digitally via android app and transferred via SMS to the main server. The system was used for six weeks (from 22nd Feb to 4th April 2016). Center for Disease control and prevention (CDC)s framework for surveillance system evaluation was used. The framework had recommended nine attributes for surveillance but as our system was in early stage of implementation suitable four attributes namely, Timeliness, Data Quality (Completeness, Error Rate), Acceptability (cost of SMS) and Simplicity (Ease of Use) were selected. Metadata related to date and time was used for calculating timeliness. Data elements were categorized as compulsory and optional to calculate completeness. Patient counseling was video recorded and reviewed for error rate calculation. The cost of SMS per individuals record was calculated using a mean number of characters in SMS received. A self-administered questionnaire was used for ease of use. The system was tested in Bahunepati Health Center and Thangsen Health Center, rural health centers from a network of eighteen health centers of Dhulikhel Hospital- Kathmandu University Hospital, Nepal. Bahunepati Health Center lies in Sindupalchowk district which was hit with a maximum number of aftershocks and Thangsen Health Center lies in Nuwakot District which is adjacent district hit by largest magnitude earthquake. Results: During the study period, 459 cases were reported via SMS. The mean time required to fill digital form was 109.3 seconds (SD 182.75) and median time required was 67.25 seconds, difference in mean time required to fill digital form was nominal between two outreach centers. Similarly, mean transmission time was 447.85 seconds (SD 637.71) and median for the parameter was 226 seconds. There is a noticeable difference in mean and median transmission time between outreach centers, for Bahunepati Health Center mean was 569.03 seconds (SD 676.11) and the median was 346 seconds whereas for Thangsen Health Center mean was 26.7 seconds (SD 58.96) and the median was 12 seconds. All the report submitted was 100% complete and error rate was found to be 1.95%. The error was on geographic information and ICD classification. The cost for transmission of data per patient was NRs. 4.26 (equivalent to USD 0.04). The mean user rating of parameters for ease of use was high with the mean point of 6.6 out of 9. Two parameters overall flexibility and performance of the task is straight forward got a rating below 6. Discussion and Conclusion: The surveillance system introduced had made complete data from individual patients available for the analysis within few minutes of entry with minimum error. The study demonstrates the feasibility of using SMS based syndromic surveillance with acceptable cost and addressing the problem faced by current surveillance system in Nepal i.e. completeness and timeliness of data.