This blog post summarizes Dr. Sophia Kagoye’s presentation about evaluating outpatient and inpatient service utilization data and evaluating health system performance using routine health facility data. Given during the 2025 annual meeting, the presentation offered practical tools for interpreting trends and identifying gaps in curative services for children under five, with a focus on data quality, subnational variation, and health system inputs such as workforce and infrastructure.

Routine data from outpatient departments (OPD) and inpatient admissions (IPD) can be used to assess whether health systems are reaching children with essential services. By examining trends over time, comparing national and subnational patterns, and assessing data completeness, country teams can better understand how well their health systems are functioning—and where to focus improvements.

Below are five key indicators:

Indicators must be reviewed carefully for possible data quality issues, including country’s contextual information. The figure below shows an example of calculating the percentage of under-five children who make up OPD service use three countries:

In these illustrative examples, which are not meant to reflect official statistics:

• Country A showed consistently high OPD utilization (around 3 visits per child annually), with under-fives accounting for 40–45% of all OPD visits. The data should be reviewed further because the usage seems too high to be accurate.
• Country B reported much lower utilization (<0.5 visits per child), and under-fives represented only 16–22% of all OPD visits—suggesting possible underreporting or limited service use.
• Country C demonstrated unstable patterns across years, pointing to major data quality concerns. Only 2019 and 2022 could be deemed plausible.

Similarly, calculating the number of IPD admissions per 100 children under 5 illustrated both the utility and limitations of this data:

In these illustrative examples, which are not meant to reflect official statistics:

• Country D had one of the lowest IPD admission rates in the region (<2 admissions per 100 children under five). Data for 2022 and 2023 appeared particularly problematic, with near-zero admissions reported.
• Country E showed fairly high IPD admission rates, ranging from 5 to 8 per 100 children under five between 2019 and 2023. Rates rose to 7–8 per 100 in 2022–2023, suggesting increasing utilization and possible improvements in access or data completeness.
• In Country F, inpatient admissions reached 10–13 per 100 children under five annually, supported by consistent reporting over time.

These examples underscored the value of tracking patterns over time and questioning outliers based on local knowledge and data completeness.

Integrating Health System Inputs and Outputs

Another useful analysis is to example the relationship between system inputs (such as health workforce, infrastructure and commodities) and outputs (utilization and coverage of preventive and curative services). Benchmarks exist for national health system inputs, such as two health facilities, 25 beds, and 23–44.5 core health professionals per 10,000 population (based on WHO standards and universal health care targets.)

At the subnational level, data from several countries revealed wide disparities:

• Health workforce density ranged from 4 to 17 per 10,000 population across regions.

• These variations were linked to differences in service use. For instance, as illustrated in a scatter plot for under-five OPD visits per child per year against workforce density, regions with higher health worker densities generally had higher OPD and IPD utilization rates for children under five.

Diagnosing Health System Performance

Interpreting health system performance requires both quantitative analysis and contextual understanding. For example:

• A region with unexpectedly low OPD use despite high workforce density might be experiencing bottlenecks in infrastructure or commodity supply.

• Conversely, regions with very low workforce but high service use might reflect data anomalies or overburdened systems operating beyond capacity.

Explore the Shiny app which was used to perform the analyses in the 2025 annual meeting through the Github repository