Improving core outcome sets development in low-and middle-income countries
Background
Standardizing outcomes and their measurement is a crucial aspect of evidence-based medicine. A core outcome set (COS) is a standardized set of outcomes that researchers should measure and report. The use of COS has the potential to reduce research waste, enhance knowledge translation, and ensure that patient-relevant outcomes are always reported. However, most COS work has been led by high-income countries (HICs), leaving a gap in low- and middle-income countries (LMICs). This thesis aims to fill this gap by exploring ways to improve the development and use of COS in LMICs, using the neonatal COS development process in Kenya as a model.
Methodology:
To understand current practice, I undertook a systematic review to describe the extent of inclusion of LMIC stakeholders in the development and use of COS (. I conducted two online surveys to explore views on including LMIC stakeholders in COS development and use. Survey 1 targeted COS developers from HICs, and Survey 2 targeted LMIC stakeholders. In survey 2, I presented three existing COS (Pre-eclampsia, COVID-19 and Palliative care) as case scenarios, and I asked respondents whether they would use the COS (with reasons).
To explore whether an existing COS, agreed predominantly by HIC stakeholders, should be adopted or adapted for the Kenyan context, I undertook key informant interviews with clinicians working in newborn units and policymakers in Kenya to understand what outcomes are important to them. I also undertook focused group discussions with caregivers/mothers who had had their neonates admitted previously to newborn units. This collaborative approach helped me understand the key outcomes from their perspective. I finally conducted a consensus meeting with key stakeholders to generate an adapted COS for use in Kenya.
Key findings
From the systematic review, only one in five (75 of 380, 20%) COS included stakeholders from LMICs, with only four COS projects originating from LMICs. In survey 1, 37 (49%) responses were received from 75 COS developers, 29 of whom had published them between 2015 and 2020. In survey 2, there were 81 respondents from LMICs; 26 had experience using a COS, and nine had been involved in COS development. Across the two surveys, personal research interests were a key driver for initiation/participation in a given COS project; LMIC stakeholders were most frequently involved in determining the ‘what to measure’ stage of COS development as opposed to the other COS development stages like scoping and how to measure. Respondents suggested that the sensitization of stakeholders on the usefulness of COS in LMICs, translation of Delphi and COS materials into local languages, and enhancement of feasibility of outcome measurements would help get more LMIC participants to be part of COS development.
The key informant interviews and Focused Group Discussions (FGD) yielded 16 outcomes (survival, length of stay in hospital, ability to feed or weight gain or growth, cognitive ability, visual impairment or retinopathy of prematurity (ROP), impact on caregivers and wider family, financial costs to the caregiver, pain, adverse events due to medicine, respiratory distress, quality of life, sepsis, future wellbeing, jaundice, necrotizing enterocolitis (NEC), ability to touch/palpate). These outcomes were subjected to a consensus-building workshop and a final set of 12 outcomes (survival, length of stay in hospital, ability to feed or weight gain or growth, cognitive ability, visual impairment or ROP, impact on caregivers and wider family, financial costs to the caregiver, pain, adverse events due to medicines, respiratory distress, quality of life, sepsis/infections) were agreed upon. Seven outcomes were similar to the HIC COS (survival, cognitive ability, visual impairment or ROP, adverse events due to medicines, respiratory distress, quality of life and sepsis/infections). In contrast, four outcomes (NEC, brain injury on imaging, hearing impairment, and general gross motor ability) were not included in this COS.
Conclusion:
Although LMIC stakeholders have been increasingly included in COS development and use over time, more work is required to test the proposed strategies for enhancing COS development and use in LMICs. This could be coupled with other methodological enhancements, such as documenting the adoption or adaptation of existing COS in an LMIC setting, which has the potential to enhance COS utility, as demonstrated in this thesis.