(68) Measuring Access to Medicines for Children: The Robustness of a Child-Specific Indicator

Publication Number: 399

Background: Sustainable Development Goal (SDG) indicator 3.b.3 is the key indicator to monitor progress in access to medicines as per SDG targets 3.8 and 3.b, but the needs of children are insufficiently covered in the present indicator. A novel methodology was developed to measure access to essential medicines specifically for children, within the existing framework of SDG indicator 3.b.3. This methodology consists of multiple components, including a basket of medicines, the availability and price of these medicines, the National Poverty Line (NPL), the number of units needed for treatment (NUNT) and disease burden (DB).

Objectives: To evaluate the robustness of this child-specific methodology and assess the influence of individual parameters, by using variations of the different components.

Methods: Data on availability and prices of essential medicines for young children (aged 1 month to 5 years) from ten historical World Health Organization/Health Action International (WHO/HAI) datasets were combined into a single database, to increase the volume of data. Two datasets were extracted from this database, both containing 25 composite facilities: dataset 1 (selected at random) and dataset 2 (preference given to medicines that were available). The methodology as proposed was used to run a base case scenario on both datasets. Univariate sensitivity analyses were performed to test the assumptions related to the NUNT (+/-60%; representing differences in body weights of children within age group) and different scenarios for weighting for DB. An additional sensitivity analysis was run on a continuously smaller basket of medicines to find the minimum number of medicines that should be surveyed to get a stable estimate. Mean facility scores were calculated, which express medicines’ accessibility across health facilities.

Results: The mean facility score for dataset 1 within the base case scenario was 35.5% (range 8.0-58.8%). Different NUNT scenarios lead to variations in mean scores of +0.1% and -0.2%, or differences of +4.4% and -2.1% at a more conservative NPL of $5.50. Different approaches for weighing for DB induced a variance of 9.0%. For dataset 2, the base case scenario resulted in a score of 76.3% (range 57.2-90.6%). Variations to the NUNT generated differences of +0.0% and -0.6%. At an NPL of $5.50 the differences were +5.0 and -2.0%. Applying other procedures for DB weighting led to a variance of 11.2%. Less than 5% change in mean facility score was observed for a medicine basket down to 12 medicines. For smaller baskets, scores increased more rapidly with a widening range.

Conclusions: We conclude that the proposed methodology provides robust estimates of access to medicines with regard to the NUNT and with baskets of at least 12 medicines. Although small differences were found compared to the base case scenario, we consider a difference of 5% not relevant. Alternative approaches for DB weighting may cause relevant fluctuations, and should be investigated further.