Show Me the Data: Rethinking Forecasting for Better Resupply Decisions

A woman brings her baby for vaccination at a health facility in Nampula Province, Mozambique. By Nelson Sardinha

When it comes to providing life-saving vaccines, accurate forecasting makes all the difference. In a recent webinar hosted by TechNet, JSI’s Wendy Prosser and Laila Akhlaghi presented findings from a deep dive into routine data conducted in Mozambique that looked at alternative approaches to vaccine forecasting. This research showed that using existing data in new and simple ways could transform resupply decisions, reduce stock-outs, and improve vaccination coverage.

Most countries develop annual vaccine forecasts based on population estimates derived from census data that may be old. Subnational resupply calculations are determined by geographically divided target populations, but, as populations shift and grow unevenly across regions, this can add another layer of inaccuracy.

The analysis, supported by the Gates Foundation-funded Immunization Collaborative Supply Planning Strengthening (ICSPS) Project, sought to understand true wastage rates, find alternatives to traditional population estimates for forecasts, and generate additional insights to improve forecasting and supply planning.

Understanding Wastage Rates

Standard wastage rates are routinely used for forecasting and re-supply decisions regardless of immunization program or facility size. The World Health Organization and Gavi apply these standard rates to forecasts. However, wastage rates are strongly influenced by facility size and service volume. This study illustrated that larger facilities with higher patient volumes typically had lower wastage rates, while smaller facilities experienced higher rates—particularly for vaccines like BCG and measles that have six-hour usage windows after opening.

While it is often thought that data on wastage rates are not available, if systems track what comes into a facility, what remains, and what doses are administered, wastage can be easily determined. In the webinar, Prosser and Akhlaghi emphasized that some open vial wastage is inevitable because multi-dose vials have to be discarded after a certain period of time. Better calculation of wastage rates will lead to improved understanding of true wastage and how to avoid it.

There’s always going to be wastage with vaccines and to make sure all children can get vaccinated, we need to move away from that being a punitive thing,” said Prosser.

Innovative Approaches to Forecasting

Stock-outs lead to missed vaccination opportunities. More accurate forecasting would reduce these misses while minimizing inter-facility stock transfers that currently compensate for stock-outs.

JSI supported Mozambique with a national annual forecast in 2024 and 2025. Through this study, JSI investigated alternatives to population-based forecasting by analyzing data from 36 health facilities across two districts in Mozambique. The analysis revealed that when using the national statistics office population estimates, facilities consistently under-forecasted vaccine needs by 4–67%.

In contrast, when using BCG doses administered at the facility level as an alternative to population estimates, facilities achieved significantly more accurate forecasts, with a tendency to over-forecast by just 0–33%. In the province in Mozambique studied, this alternative method could be a better option because the high rate of facility births leads to higher BCG coverage needs. Whether this alternative method will work in other countries or even geographic areas within Mozambique depends on BCG coverage and facility-based birth rates.

Supply chain decision-making shouldn’t rely on one forecasting method,” said Akhlaghi. “If you’re looking at multiple methods, then you have better perspectives and can make better-informed decisions.”

Practical Applications for Immunization Programs

The study findings offer several practical considerations for immunization programs worldwide:

1. Calculate actual consumption using existing data points. Many systems already collect beginning balance, receipts, and end balance, which is sufficient information to calculate consumption and wastage.
2. Use multiple forecasting methods and compare their accuracy over time. The most appropriate method may depend on country context, but consumption-based approaches deserve stronger consideration.
3. Analyze facility-specific wastage rates rather than applying standard rates. This enables more accurate resupply quantities and shows which facilities need targeted supervision and training to improve service delivery and inventory management.
4. Use buffer stock for its intended purpose to address unpredictable variability in supply and demand, rather than compensating for routine demand that should be properly forecasted.

By embracing multiple forecasting methodologies, programs can significantly improve accuracy, reduce stock-outs, and ultimately reach more children with life-saving vaccines.

For further study details and discussion, watch the webinar recording and access the presentation slides.