Generalisability of and lessons learned from a mixed-methods study conducted in three low- and middle-income...

Identifying existing care pathways is the first step for understanding how services can be improved to enable early diagnosis and effective follow-up care for non-communicable diseases (NCDs); however, evidence on how care pathways can and should be identified in low- and middle-income countries (LMICs) is lacking.

To describe generalisability and lessons learned from recruitment and data collection for the quantitative component of a mixed methods study designed to determine the care pathway for atrial fibrillation (AF) in Brazil, China and Sri Lanka.

Adults (≥18 years) that spoke the local language and with an AF diagnosis were eligible. We excluded anyone with a hearing or cognitive impairment or ineligible address. Eligible participants were identified using electronic records in Brazil and China; in Sri Lanka, researchers attended the outpatient clinics to identify eligible participants. Data were collected using two quantitative questionnaires administered at least 2-months apart. A minimum sample size of 238 was required for each country.

The required sample size was met in Brazil (n = 267) and China (n = 298), but a large proportion of AF patients could not be contacted (47% and 27%, respectively) or refused to participate (36% and 38%, respectively). In Sri Lanka, recruitment was challenging, resulting in a reduced sample (n = 151). Mean age of participants from Brazil, China and Sri Lanka was 69 (SD = 11.3), 65 (SD = 12.8) and 58 (SD = 11.7), respectively. Females accounted for 49% of the Brazil sample, 62% in China and 70% in Sri Lanka.

Generalisability was an issue in Brazil and China, as was selection bias. Recruitment bias was highlighted in Sri Lanka. Additional or alternative recruitment methods may be required to ensure generalisability and reduce bias in future studies aimed at identifying NCD care pathways in LMICs.