Service utilization analysis¶
Background and purpose¶
Objective of the module¶
Service utilization analysis describes how much of each health service is delivered over time: how many antenatal visits, deliveries, vaccinations or consultations are recorded each month, how those volumes change from one year to the next and from one quarter to the next, and how much the data quality adjustments change the picture.
It answers the first question a program asks of its routine data: are more or fewer people using the service than before, and where? It does not say whether a change is unusual given trend and seasonality. That question is answered by the Disruption detection module, which models each facility's expected volume and flags the months that fall outside it.
Analytical rationale¶
Raw monthly counts are hard to read directly. Volumes rise and fall with the season, a facility that stops reporting pulls the total down, and an outlier can double a district's figure for a month. Service utilization analysis therefore works on the adjusted data produced by the Data quality adjustment module and presents it in three ways that are easy to compare: annual totals with the change from the previous year, quarterly totals with the change from the previous quarter, and the same volumes under each adjustment scenario so that the effect of the adjustments is visible.
Because the analysis is descriptive, its reading depends on the direction of the indicator. More deliveries in facilities is good; more stillbirths is not. Indicator directionality is recorded so that increases and decreases are colored consistently.
Key points¶
| Component | Details |
|---|---|
| Inputs | Adjusted service volumes from the Data quality adjustments module: unadjusted, outlier-adjusted, completeness-adjusted, and both Indicator values from the Indicator values module for calculated indicators |
| Outputs | Monthly and annual service volumes by indicator and area Year-over-year change Quarter-on-quarter change Volume under each adjustment scenario |
| Purpose | Describe and compare service volumes over time and across areas, as the starting point for interpretation and for disruption detection |
Analytical workflow¶
Overview of analytical steps¶
-
Start from adjusted data. The Data quality adjustment module produces one monthly count per facility and indicator under four scenarios: no adjustment, outliers adjusted, completeness adjusted, and both. The service utilization views aggregate these counts to the chosen geographic level.
-
Compare years. Annual totals are computed per indicator and area, and the change from the previous year is expressed as a percentage.
-
Compare quarters. Quarterly totals are compared with the previous quarter in the same way, which captures more recent shifts than the annual view.
-
Flag large changes. Changes beyond ±10 % are highlighted for follow-up. A flag is a prompt to investigate, not a conclusion: the change may be a real program effect, a data quality problem, or an expected event such as a campaign.
-
Show the effect of adjustments. The same volumes are displayed under each adjustment scenario so that users can see how much of a change is due to the data corrections rather than to service delivery.
Key decision points¶
Which count to display
The outlier-adjusted count is the usual choice for reporting. The completeness-adjusted counts estimate what the volume would have been if every facility had reported, which is useful for planning but should be labeled as an estimate. Showing the unadjusted count alongside makes the size of the correction explicit.
Geographic level
National and provincial totals are stable enough to read month by month. District totals are noisier, especially for low-volume indicators, where a small absolute change becomes a large percentage. Cross-check district changes against the provincial trend before acting on them.
Indicator directionality
For most RMNCAH-N service indicators an increase is favorable. For a few, such as stillbirths or maternal deaths, an increase is unfavorable and the color coding is inverted. The direction is set per indicator in the indicator manager.
Analysis outputs and visualization¶
1. Change in service volume
Bar chart of annual service volumes by area and indicator, annotated with the year-on-year percent change.

2. Number of services reported
Monthly volumes over time, by indicator and area, from the adjusted data.

3. Quarter-on-quarter change
Heatmap comparing each quarter with the previous one, with changes beyond ±10 % flagged.

4. Volume change due to data quality adjustments
Grouped bar chart comparing service volumes across the four adjustment scenarios.

Interpretation guide
- Bars and annotations (output 1): annual volumes, with the percent change between consecutive years.
- Heatmap cells (output 3): the percent change for each area and quarter; flagged cells exceed the ±10 % threshold in either direction.
- Four bars per year (output 4): one per adjustment scenario; compare heights to see how much the adjustments change the reported volume.
Detailed reference¶
Change metrics¶
Year-over-year change compares a period with the same period one year earlier:
Quarter-on-quarter change compares a quarter with the previous quarter:
Both are computed per indicator and geographic area. Changes beyond ±10 % are flagged. The threshold is a presentation setting on the chart and can be changed per visualization.
Interpretation¶
- A flagged change asks a question; the answer comes from context. Check for campaigns, stock-outs, strikes, policy changes, and reporting problems in the same months.
- Compare the flagged change under the different adjustment scenarios. A change that disappears once completeness is adjusted was a reporting change, not a service change.
- For low-volume indicators and small areas, read the absolute change next to the percentage.
- Sustained movements over several periods carry more weight than a single flagged quarter.
- The last months of a series are provisional, because late reports arrive after the package is generated.