The magnitude and predictors of abandonment of therapy in paediatric acute leukaemia in middle-income countries: a systematic review and meta-analysis.

BACKGROUND: Abandonment of therapy is a significant cause of paediatric cancer treatment failure in low- to middle-income countries (LMIC), but its impact has been underestimated. We performed a meta-analysis to determine the magnitude of abandonment in paediatric leukaemia in LMIC and sought to identify patient-, centre- and country-specific predictors of abandonment. PATIENTS AND METHODS: We searched seven databases to identify paediatric oncology cohorts followed up from diagnosis and treated in LMIC. All languages were included. Two reviewers independently selected articles and extracted data. Authors were contacted for additional information. Subgroup analyses were planned a priori. RESULTS: Of 22,384 publications, 318 in eight languages met criteria for full text review. 157 studies met analysis inclusion criteria. Abandonment rates (ARs), obtained for 83 of the 157 studies (52.9%), ranged from 0% to 74.5%. ARs were frequently unreported and available only directly from authors. Forty studies (10,494 children in 20 countries) were quantitatively analysed. ARs for acute lymphoblastic laeukemia in lower-middle-income countries (lower-MICs) were higher than in upper-middle-income countries (29%, 95% confidence interval (CI) 23-36% versus 2%, 95% CI 1-3%; p<0.0001) but were heterogeneous (I(2)=98%; p<0.0001). This heterogeneity was not explained by centre-specific (free versus paid treatment) or country-specific (government health expenditure, per-capital income) subgroups. CONCLUSIONS: In LMICs, ARs are highest in lower-MICs. However, their broad range suggests that low ARs are possible in resource-constrained settings. Analysis of outliers may suggest interventions for use at other centres. Methodologically appropriate reporting of ARs should be adopted. Future research should evaluate interventions targeting abandonment.