Using clinical guidelines and card sheets for guiding the design of data-driven clinical pathways

Abstract

Clinical pathways(CP) aim to link evidence to practice for standardizing and managing the quality of healthcare services. However, putting evidence into practice is challenging in health systems with limited resources where limited digitalization represents one of the main hurdles. This study aims at investigating the need for designing automated and data driven clinical pathways for low resource settings and more specifically for the case of female reproductive health care. We conducted a case study on the Ethiopian primary health system in general and Jimma Health Center in particular. After securing ethical clearance, (i) the existing paper-based clinical guidelines(CGs), annual reports, point of care charts and card sheets were examined, (ii) a digitized CP dataset was derived from a previously created electronic template, and (iii) a python based interactive CP tool was developed for automating pre-processing, interactive visualization and analysis of the data. We found that the health center patient card sheet only contains very limited information. CGs have demonstrated a potential advantage in identifying and making referral decisions on cases that relate to several concurrent health problems. The existing paper-based point of care instruments have the disadvantage of not being interactive and proved difficult to use for extracting relevant clinical information, summarizing the patient history, constructing a patient flow diagram, diagnosing all potential underlying diseases and in the end for delivering optimal clinical pathways. The study demonstrates that health care services, as they are implemented now, have severe shortcomings and prevailing paper-based methods are inefficient for delivering useful evidence to the frontline health workers.  Utilizing existing care information for delivering adaptive evidence-based health services in low resource settings will require  a suitable algorithm that works with  limited input (i.e. clinical signs and symptoms) and updates the generated clinical pathway incrementally each time additional information becomes available.