Index

Countdown Analytical Approach

Data Quality Assessment

Reporting completeness

Scientific Foundation:

Incomplete reporting creates systematic underestimation of true coverage. Assuming non-reporting facilities delivered zero services is statistically implausible; they simply didn’t submit data.
Adjustment factors account for this gap using a defensible assumption: non-reporting facilities operate at lower capacity (typically 25% of reporting facility volume).

The completeness of reporting may affect the analysis, especially if completeness is low or varies between years. Extreme outliers, such as an accidental extra zero in a number, can have a large impact, especially on subnational numbers. Following the assessments, several steps are necessary to obtain a clean data set for analysis. This implies adjusting for incomplete reporting and correcting for extreme outliers.

If we do not consider reporting completeness that means we assume all non-reporting facilities provided zero services, which is not likely to be true. Adjustments depend on how much services (e.g., pregnancy care, vaccinations) were provided at non-reporting facilities compared to those that reported. The adjustment factor $k$ - defined as the ratio of the volume of services provided by non-reporting facilities to the volume of services provided by reporting facilities - is used to adjust the reported numbers for incomplete reporting.

To account for incomplete reporting, the reported number of events can be adjusted using completeness and facility reporting ratio, with the following formula:

$N_{\text{adjusted}} = N_{\text{reported}} + N_{\text{reported}} \times \left(\frac{1}{c} - 1\right) \times k$

Where:

$N_{\text{adjusted}}$ — Total number of events adjusted for incomplete reporting
$N_{\text{reported}}$ — Number of events reported
$c$ — Reporting completeness (e.g., proportion of facilities that reported)
$k$ — Adjustment factor to account for lower service volume in non-reporting facilities

Interpretation:

As a default value, we use $k=0.25$ , which means the non-reporting facilities provided services but only at a volume which was a quarter of the reporting health facilities.
The adjustment assumes that non-reporting facilities still provide services, but at a lower volume compared to reporting facilities. The factor $k$ controls the magnitude of this adjustment.
The factor k can be different for different services. For instance, if private facility reporting is poor but they are in the national system and they provide a considerable number of deliveries, k maybe greater than 0.25 or even as high as 1.0.
If the facility reporting rate is below 75%, it becomes more difficult to impute district data. Therefore, no adjustments are made if reporting is lower than 75%. In that case, further analysis to determine coverage with the facility data is not considered sufficiently reliable.

The following k-values are used depending on the reporting used to adjust the reported numbers for incomplete reporting:

k = 0 : No services in non-reporting facilities (default k-value)
k = 0.25 : Some services, but much lower than reporting facilities
k = 0.50 : Half the rate compared to reporting facilities
k = 0.75 : Nearly as much as reporting facilities
k = 1.0 : Same rate of services as reporting facilities

Adjustment Action summary

Problem	Action	Adjustment
Low reporting rates: identifying low rates that were adjusted	If below 75% (default), data were imputed	Median monthly value for the district year was imputed for the month with low reporting
Incomplete reporting by districts, variable over time, affecting trend assessment	If reporting rates were >=75% and ≤ 100% (default), an assumption was made about the volume of services provided by the non-reporting facilities	Adjustment factor k value was used to adjust for incomplete reporting k default value 0.25 (replace if different value used; state if used for all reporting forms or different k factors between forms)
Extreme outliers can greatly affect coverage trend assessments	If a monthly value was greater or smaller than 5 times the median absolute deviation (MAD) from district monthly median value, an adjustment was made	Extreme monthly outliers are corrected and given the district median value for the same year
Missing values	If there is a missing value, data were imputed	District median monthly value for the year was imputed for the month with missing value

Outlier detection

An extreme outlier is defined as a monthly value that is 5 times the median absolute deviation (MAD) from the monthly median value for a particular year.

Internal Consistency

Internal consistency assesses the degree to which related indicators are aligned within the dataset, based on expected epidemiological and service delivery relationships.
Within the CD2030 analytical framework, this step complements other (DQA) metrics by evaluating whether indicators that are biologically or programmatically linked follow plausible patterns.
Unlike standalone indicator checks, internal consistency focuses on relationships between indicators, ensuring that reported values reflect realistic service delivery pathways across the continuum of care.

Key indicator relationships assessed for internal consistency

1. ANC1 vs Penta1

These indicators are expected to follow a consistent pattern, as both represent entry points into maternal and child health services. After adjusting for expected losses (e.g., stillbirths, neonatal deaths), the ratio should fall within a plausible range.

The relationship between ANC1 (first antenatal care visit) and Penta1 (first dose of pentavalent vaccine) reflects continuity between maternal and child health services.

The expected ratio is derived from biological and demographic assumptions, including:

Pregnancy loss: (~3%)
Stillbirths (~2%)
Twinning (~1.5%)
Neonatal mortality (~3%)

These factors yield a baseline expected ratio of:

Expected\ Ratio = 1.07

This is further adjusted using survey coverage:

Expected\ Ratio = 1.07 \times \left(\frac{\text{Survey ANC1 Coverage}}{\text{Survey Penta1 Coverage}}\right)

Threshold: A tolerance range of ±0.05 around the expected ratio is considered acceptable.

Example:

If ANC1 coverage = 90% and Penta1 = 95%:

1.07 \times \left(\frac{0.90}{0.95}\right) = 1.01

Acceptable range: 0.96 – 1.06

Penta1 vs Penta3

This ratio reflects immunization programme continuity and is primarily driven by drop-out rates rather than biological factors.

The expected ratio is based on survey coverage:

Expected\ Ratio = \frac{\text{Survey Penta1 Coverage}}{\text{Survey Penta3 Coverage}}

Threshold: A tolerance range of ±0.05 is applied.

Example:

If Penta1 = 95% and Penta3 = 85%:

\frac{0.95}{0.85} = 1.12

Acceptable range: 1.07 – 1.17

Institutional Deliveries vs Penta1
The number of facility-based deliveries should broadly align with subsequent child health service uptake, including immunization.

Institutional Deliveries vs Penta1

Interpretation in the DQA Process

Internal consistency should be interpreted alongside:

Reporting completeness
Outlier detection
Data completeness

Strong alignment between related indicators increases confidence in the reliability of the data. Significant deviations from expected ratios should be investigated within the local context before making adjustments.

Not all deviations indicate poor data quality. Programme changes, campaigns, or service disruptions may explain observed differences.

Coverage Analysis

Rationale: Scientific basis for the analysis[#coverage-rationale]

Scientific Foundation:

Coverage of interventions is a critical and direct output of health systems. Regular tracking of coverage at national and sub-national levels has become the mainstay of monitoring progress in national health plans and international initiatives.
Reproductive, maternal, newborn, child and adolescent health indicators with targets are the most common indicators of national health plans and global monitoring.
Both health facility data and household surveys can provide coverage statistics, and an integrated analytical approach is desirable.

Many coverage indicators can be estimated in both surveys and from health facility data. Both are critical pieces of information and need to be considered in conjunction with each other. The facility data can be used to generate annual coverage estimates, and the coverage results should be compared and interpreted alongside the results from recent surveys.

Key Coverage Indicators (Survey vs. Facility Denominators)

Indicator	Survey denominator	Facility data denominator
Antenatal care (ANC1, ANC1 Trimester, ANC4)	Women aged 15-49 years with a live birth in the last 2 years	Estimated total pregnant women in the population
Intermittent preventive therapy (IPT2, IPT3)	Women aged 15-49 years with a live birth in the last 2 years	Estimated total pregnant women in the population
Institutional deliveries & SBA	Live births in the last 2 years	Estimated livebirths as denominator
Low birth weight & C-section	Low birth weight / C-section counts	Estimated livebirths as denominator
Postnatal care within 48 hours (PNC48h)	Postnatal care within 48 hours	Estimated livebirths as denominator
BCG, Penta1, Penta3	Children 12-23 months	N of surviving infants (beyond neonatal period)
Measles1	Children 12-23 months	N of surviving infants (beyond post-NN period)
Measles2	Children 24-35 months	N of surviving infants (beyond postneonatal period)

Antenatal care (ANC)

Most countries have at least one ANC indicator with a target in the national plan. The global ENAP/EPMM coverage targets for 2025 are: globally, at least 90% of pregnant women with 4 or more ANC care visits, and 90% of countries with at least 70% coverage.

ANC1 is often considered an indicator of basic access to health services. It is high in most countries, and in many instances, the numbers of ANC1 visits in the routine health facility data can provide a better denominator for the ANC and delivery indicators than population projections.

Sometimes, an indicator may reach an unlikely high coverage at the national level, say over 125%. This may be because the data quality of the numerator of the coverage indicator is poor, the denominator is wrong, or the intervention is given and recorded more than once during pregnancy (e.g., IFA supplementation).

In that case, the computation of coverage is not useful. It is better to express it differently. For instance, if coverage is 200%, it is better to compute the average number of courses of 90 IFA tablets that a pregnant woman received (in this case 2.0 per pregnant woman in the population).

**** [insert ANC facility vs survey graphs Output] ****

Delivery Care

All countries have at least one delivery care indicator with a target in the national plan. Institutional (live) birth coverage and SBA are closely related, as almost all deliveries with a skilled attendant occur in health facilities. From the analytical perspective, institutional birth coverage is preferred because it is a more objective measure and avoids issues with varying national definitions of what constitutes a skilled birth.

Caesarean section is a life-saving intervention and an important indicator. A general rule of thumb put forward by the World Health Organization (WHO) is that the population-level need for Caesarean section is in the range of 10–15% of all deliveries.

Below 10%: Indicates an unmet need for surgical delivery care.
Over 10–15%: Implies a likely overuse of Caesarean section. It may also imply a combination of unmet need among certain population groups (e.g., the poorest or rural women) alongside overuse in other subgroups.

**** [insert institutional delivery coverage charts Output] ****

Immunization

Immunization coverage indicators are included in virtually every country’s health sector monitoring plan. A general target is at least 90% coverage for essential vaccines given in childhood and adolescence.

Denominator Formulations for Facility Data: The number of vaccinations given to infants is used as the numerator. The denominator is the number of eligible children in the population, approximated as:

\text{Denominator (BCG, Penta)} = \text{Live Births} - \text{Neonatal Deaths}

\text{Measles 1 Denominator} = \text{Live Births} - \text{Neonatal Deaths} - \text{Postneonatal Deaths}

\text{Measles 2 Denominator} = \text{Live Births} - \text{Neonatal Deaths} - (2 \times \text{Postneonatal Deaths})

**** [insert immunization coverage charts Output] ****

Family Planning

Family planning coverage estimates are derived from a collaboration between Countdown to 2030 and Track20, using the Family Planning Estimation Tool (FPET).

\text{Demand Satisfied (Coverage)} = \frac{\text{Use of Modern Contraceptives}}{\text{Unmet Need} + \text{Use of Modern Contraceptives}}

**** [insert figure FPET projection Output] ****

Equity Analysis (Equiplots)

Rationale: Scientific basis for the analysis

Scientific Foundation:

National coverage averages mask critical disparities that determine whether health systems are truly leaving no one behind.
Household surveys provide critical information on inequalities across three major dimensions: area of residence (urban/rural), wealth (richest-to-poorest quintiles), and education of the mother.
Equiplots are used to assess whether the country has made progress in reducing the poor-rich gap or the education gap over time.

For wealth quintiles, it can be assessed what the type of inequality is, as all categories are of the same size. Each pattern requires a different strategy of health programming and targeting:

Top inequality (mass deprivation): If the richest are well ahead of all other wealth quintiles, only the rich escape.
Linear inequality: If the coverage differences are equidistant, this is a linear pattern where the increasing household wealth is linearly associated with higher coverage.
Bottom inequality (marginalization): If the poorest are left behind, this means marginalization of the poorest.

The interpretation should focus on whether inequalities have reduced over time. The following questions should guide interpretation:

What is the level of inequality in the most recent data point?
How have inequalities changed over time?
Is there any inequality pattern that can be observed?
What will be the best approaches to reduce inequalities?

**** [insert equiplots Output] **** **** [insert wealth equiplot Output] **** **** [insert area equiplot Output] **** **** [insert residence-penta3 equiplot Output] ****

Health Systems Performance

Rationale: Scientific basis for the analysis

Scientific Foundation:

The assessment of the burden of disease, intervention coverage, quality, equity, and health system inputs should directly guide policy-making and program targeting.
A comprehensive evaluation requires a baseline comparison of health system inputs (such as financing, health workforce density, and infrastructure availability) against the corresponding outputs of the system (including service utilization and coverage rates) at the sub-national tier (Admin1 level).
The core objective remains isolating specific sub-national bottlenecks where high system inputs fail to translate into expected RMNCAH-N service coverage outputs, or conversely, identifying high-performing regions exhibiting strong service delivery efficiency.

Health Systems Inputs

The assessment focuses on the quality of data for the health system indicators at both national and sub-national levels, targeting global benchmarking and plausibility auditing.

Core Health System Input Indicators

Health Financing: Frequently limited to budgeted government allocations.
Core Health Professionals per 10,000 Population: Includes physicians, non-physician clinicians, nurses, and midwives.
- Historical Threshold (WHO 2006): Minimum baseline of 23 per 10,000.
- Modern Threshold (UHC Era): At least 44.5 per 10,000.
Health Facility Density:

\text{Health Facility Density} = \left( \frac{\text{Total Number of Health Facilities}}{\text{Total Population}} \right) \times 10,000

Hospital Density:

\text{Hospital Density} = \left( \frac{\text{Total Number of Hospitals}}{\text{Total Population}} \right) \times 100,000

Inpatient Bed Density:

\text{Inpatient Bed Density} = \left( \frac{\text{Total Inpatient Beds}}{\text{Total Population}} \right) \times 10,000

**** [insert subnational_health_inputs_scatter_matrix Output] ****

Health Systems Outputs by Inputs

The analysis explores the statistical association between health system inputs and matching service utilization or coverage indicators across sub-national units (Admin1 level):

Inpatient Infrastructure vs. Under-5 Hospital Admissions: We expect regions with lower hospital/bed density to exhibit lower pediatric admission rates. Outliers may indicate infrastructure under-reporting or high clinical throughput.
Health Workforce Density vs. Under-5 Outpatient Utilization: We expect a positive slope on a linear regression line. Outliers may flag staff workload burdens or numerator inflation.
Health Workforce Density vs. Institutional Live Birth Coverage: Assesses if workforce availability translates directly into institutional delivery coverage.

**** [insert 6-ipd_use Output] **** **** [insert 8-health_system_ratio_opd_u5_hwf.png Output] ****

Institutional Mortality

Rationale: Scientific basis for the analysis

Scientific Foundation:

Maternal and perinatal mortality in health facilities are critical indicators of the quality of care (institutional mortality).
If coverage of births in health facilities is high (e.g., over 75%), the facility-based statistics become a useful input into the estimation of population levels of mortality.
The main challenge with mortality data from health facilities is under-reporting of deaths. Maternal deaths do not only occur during birth, but also in pregnancy and post-partum. Reporting of stillbirth deaths requires well-maintained maternity registers and cross-checks to operation theatre registers.

Institutional vs. Population Mortality Definitions

Indicator	Numerator	Denominator
Institutional maternal mortality ratio (iMMR)	Number of maternal deaths in health facilities	$\frac{\text{Live births in health facilities}}{100,000}$
Population maternal mortality (MMR)	Number of maternal deaths in the population	$\frac{\text{Live births in the population}}{100,000}$
Institutional stillbirth rate (iSBR)	Number of stillbirths in health facilities	$\frac{\text{Births in health facilities}}{1,000}$
Neonatal mortality (before discharge)	Number of neonatal deaths before discharge	$\frac{\text{Live births in health facilities}}{1,000}$

Data Interpretation and Quality Metrics

When interpreting iMMR and iSBR, assess extreme high outliers (data entry errors) and implausibly low rates (e.g., iMMR < 25 per 100,000 or iSBR < 6 per 1,000), which strongly indicate major under-reporting rather than superior clinical outcomes.

Ratio stillbirth to maternal deaths: We expect maternal mortality and stillbirth to be positively correlated. We expect the ratio of stillbirth to maternal death to be in the range of 5 to 25 for countries in sub-Saharan Africa.

Ratio < 5: Under-reporting of stillbirths is likely greater than under-reporting of maternal deaths.
Ratio ≥ 25: Under-reporting of maternal deaths is likely to be the main issue.

**** [insert immr_trends_screenshot Output] **** **** [insert isbr_trends_screenshot Output] **** **** [insert ratio of stillbirth chart Output] ****

Consistency with Population Estimates

The population MMR ( $M_p$ ), community MMR ( $M_c$ ), and institutional MMR ( $M_i$ ) must be mathematically consistent.

M_p = (M_i \times P_i) + (M_c \times (1 - P_i))

Where $P_i$ is the proportion of live births occurring in a health institution.

Given a baseline estimate for the population MMR ( $M_p$ ) and an assumed community-to-institutional variance ratio ( $R = \frac{M_c}{M_i}$ ), the expected institutional MMR is calculated as: