Public health priorities in emergencies

1Why people die in emergencies

In the first weeks of a crisis, excess deaths are rarely due to injuries themselves. They come mainly from preventable and treatable diseases: diarrhoea (including cholera), acute respiratory infections, measles and malaria, made worse by acute malnutrition. Added to these are complications of pregnancy and childbirth, neonatal deaths and the interruption of treatment for chronic diseases (diabetes, hypertension, HIV, tuberculosis).

Risk factors combine: overcrowding, lack of water and sanitation, inadequate shelter, low vaccination coverage, collapse of health services, insecurity. As a result, the health response is not limited to health facilities. It also relies on water, sanitation, nutrition and shelter.

2Measuring mortality: CMR and U5MR

The crude mortality rate (CMR) is the most important indicator of a crisis. It is expressed as deaths per 10,000 people per day:

• CMR = number of deaths ÷ (population × number of days) × 10,000.
• The under-5 mortality rate (U5MR) is calculated in the same way, for children under 5 only.

Example: 30 deaths in 15 days in a camp of 20,000 people give 30 ÷ (20,000 × 15) × 10,000 = 1 death per 10,000 per day. Sources include home death surveillance by community workers, health facility registers, gravediggers and religious leaders, and retrospective mortality surveys (SMART method).

3Sphere emergency thresholds

Sphere compares observed mortality with the region's baseline mortality before the crisis:

• An emergency is declared when the CMR reaches or exceeds double the baseline mortality.
• If the baseline mortality is not known, the threshold of 1 death per 10,000 per day is used for the CMR and 2 deaths per 10,000 per day for the U5MR.
• The aim of the response is to bring the CMR below 1/10,000/day and, ideally, below double the baseline mortality.

Above 2/10,000/day, the situation is out of control; above 5/10,000/day, it is catastrophic. A U5MR rising faster than the CMR often signals a measles or diarrhoea outbreak, or a worsening of malnutrition.

4Initial assessment and priorities

The rapid assessment (often in the form of a multi-cluster initial rapid assessment, MIRA, or a cluster health assessment) answers simple questions within a few days: how many people, what mortality, which diseases, what nutritional status, which services are still functional (HeRAMS), which shortages, which epidemic risks.

The classic priorities of the first weeks: measles vaccination with vitamin A, water and sanitation (at least 15 litres per person per day), screening and treatment of malnutrition, access to primary health care with essential medicines, management of diarrhoea, pneumonia and malaria, the Minimum Initial Service Package for reproductive health, an early warning system. These priorities are decided with the Ministry of Health and the Health Cluster, based on data.

5Conducting the initial health situation analysis step by step (D0 to D14)

In the first two weeks of a crisis, the costliest mistake is not a lack of data: it is the lack of a method for deciding with partial data. For this purpose, the Global Health Cluster offers the Public Health Situation Analysis (PHSA, Public Health Situation Analysis), which structures the available information around four blocks: the context and the affected population, health status and threats, the availability and performance of the health system, and humanitarian access and operational constraints. The method below adapts this logic for an NGO or cluster team arriving in a newly affected area.

Step 1: secondary data review (D0 to D3)

Before any field visit, a health information officer and a public health doctor gather what already exists. The usual sources are the most recent Demographic and Health Survey (DHS) or MICS survey, data from the national health information system (often DHIS2), the Ministry of Health's weekly surveillance reports, SMART nutrition surveys from the past two years, administrative vaccination coverage and coverage surveys, HeRAMS profiles, displacement data (IOM Displacement Tracking Matrix, UNHCR figures) and the cluster's contingency plans. The output is a 4- to 6-page note giving, for each piece of information, its source, date and level of reliability.

Step 2: estimating the population and its structure (D1 to D5)

Every indicator relies on a denominator. A 30% error in the population figure carries straight through to the mortality rate, vaccination coverage and medicine requirements. The methods, from fastest to most accurate:

• Figures from authorities or leaders: quick, often inflated to obtain more aid or underestimated for political reasons; they must always be cross-checked.
• Counting shelters from satellite or drone imagery, multiplied by the average household size obtained from a sample of visited shelters (often 5 to 7 people in rural African settings).
• Quadrat method: the population is counted in 10 to 20 randomly selected squares of known area (for example 25 m by 25 m), and the average density is then extrapolated to the total area of the site.
• T-square method: density estimation based on distance measurements between shelters, suited to low-density and irregular sites.
• Registration or census (UNHCR, IOM, authorities): the most reliable, but rarely available before several weeks have passed.

In the absence of local data, planning proportions are used: children under 5 about 17 to 20% of the population in high-fertility settings, children aged 6 months to 14 years about 40 to 45%, women of reproductive age (15 to 49 years) about 22 to 25%, and pregnant women at any given time about 3 to 4%. These proportions must be replaced by registration data as soon as they are available.

Step 3: rapid primary data collection (D3 to D10)

Primary data collection combines three tools, carried out in parallel by mixed pairs (a clinician and a data collector, women and men so that all groups can be spoken to):

1. Health facility visits using the HeRAMS checklist or a short version: condition of the building, staff present, services available, stock-outs of tracer medicines, cold chain, consultation and death registers for the past four weeks.
2. Key informant interviews: facility managers, community health workers, traditional birth attendants, religious leaders, gravediggers, women's representatives. They are asked about recent deaths, illnesses observed and barriers to accessing care.
3. Direct observation: site density, water sources and quantities available, latrines, open defecation, shelters, visibly malnourished people, new graves.

Rapid screening by mid-upper arm circumference (MUAC) among children aged 6 to 59 months, on an exhaustive or convenience sample, gives a first nutritional indication. It does not replace a SMART survey but is enough to trigger a response if the proportion with MUAC below 125 mm is high.

Step 4: analysis and prioritisation (D10 to D14)

The team meets for half a day with the Ministry of Health and the cluster partners. For each health threat, it scores the likelihood and the expected impact on mortality, identifies service gaps and proposes priority interventions with their area, deadline and lead organisation. The final output is a 10- to 15-page PHSA and a one-page summary for the Humanitarian Coordinator and donors.

A useful initial analysis answers a single question: what will kill the most people in the next 30 days if nothing is done, and what can be done right now to prevent it? Anything that does not contribute to this answer can wait for the second assessment phase.

Typical timeline and division of roles

6Measuring mortality in the field: prospective surveillance and retrospective surveys

The crude mortality rate is only meaningful if we know where the counted deaths come from and which population they occurred in. In practice, two complementary approaches are combined: prospective surveillance, which shows the weekly trend, and the retrospective survey, which provides a representative estimate with a confidence interval.

Prospective mortality surveillance

It is set up in the very first week, without waiting for a survey. Its components:

• Community home-based surveillance: one community health worker or volunteer per 500 to 1,000 people (100 to 200 households) visits every household in their sector once a week and records deaths, age, sex, date and presumed cause from a short list (diarrhoea, fever, cough and difficulty breathing, malnutrition, childbirth, injury, other).
• Facility registers: inpatient deaths, deaths on arrival, deaths in nutrition centres.
• Graveyard surveillance: a paid watchman counts new graves every day, distinguishing children's graves (often by their length, less than about one metre). This method is simple and hard to manipulate; it underestimates deaths buried elsewhere.
• Religious informants and shrouds: in some settings, the distribution of shrouds or burial cloth by the organisation makes it possible to cross-check the count.

Deaths are aggregated every week by sector and age group (under 5, 5 and over). The week's CMR and U5MR are calculated with the most recent denominator and plotted on a curve. An increase over two consecutive weeks, or a U5MR above 2 deaths per 10,000 per day, triggers an investigation of the causes.

Calculating excess mortality

For advocacy and prioritisation, the rate is translated into a number of preventable deaths. Excess mortality over a period equals the difference between the observed CMR and the baseline CMR, multiplied by the population and by the number of days, divided by 10,000. With an observed CMR of 1.8, a baseline CMR of 0.5, a population of 40,000 people and a 30-day period, we get (1.8 minus 0.5) × 40,000 × 30 ÷ 10,000, or about 156 excess deaths in one month. This figure speaks more to decision-makers than the rate itself.

The retrospective mortality survey (SMART methodology)

The SMART methodology (Standardized Monitoring and Assessment of Relief and Transitions) most often combines an anthropometric survey of children aged 6 to 59 months with a retrospective mortality survey covering all household members. Key points:

1. Recall period: around 90 days, bounded by an event everyone remembers (religious festival, start of the fighting, date of arrival). A period that is too long increases recall bias; one that is too short does not contain enough deaths.
2. Household census method: for each household, the members present at the start of the period, arrivals, departures, births and deaths are listed. This makes it possible to calculate person-time at risk rather than a simple average population.
3. Sample size: calculated with the ENA for SMART software, based on the expected CMR, the desired precision (often plus or minus 0.3 to 0.5 deaths per 10,000 per day), the design effect (often 1.5 to 2), the recall period and the average household size. Mortality surveys often require several hundred households.
4. Sampling: cluster sampling (probability proportional to the size of villages or blocks), or simple random sampling when a site is small and well mapped, which removes the design effect.
5. Reporting: the CMR and U5MR with their 95% confidence intervals, the reported causes, the places of death (home, facility, en route), compared with the emergency threshold.

Results are validated by the cluster's nutrition or assessment technical working group before dissemination. A CMR of 1.2 with a confidence interval of 0.7 to 1.9 does not allow us to assert that the threshold of 1 has been exceeded: this must be stated clearly, and prospective surveillance must be relied on to follow the trend.

Baseline mortality and choice of threshold

Sphere recommends comparison with the population's baseline mortality before the crisis. The handbook provides a table of baseline mortality rates and emergency thresholds by major world region; a recent national or subnational estimate is preferable. In many sub-Saharan African countries, the baseline CMR is below 0.5 deaths per 10,000 per day, which puts the emergency threshold (double the baseline) below 1. Mechanically applying the threshold of 1 in these settings amounts to tolerating a doubling of mortality without reacting.

7Worked example: the Kalombo displacement site, from arrival to the 30-day plan

Fictional context: following fighting in Haute-Marévi province, families flee to the outskirts of the town of Kalombo and settle on a vacant plot and in three schools. A medical NGO arrives on D5 with a team of four people. The Ministry of Health runs an overwhelmed health centre 3 kilometres away. The rainy season starts in six weeks.

1. Secondary data (D5 to D7)

• Latest DHS: baseline CMR estimated at 0.45 deaths per 10,000 per day, baseline under-5 mortality at 1.1.
• Administrative coverage of measles vaccine (first dose) in the province of origin: 61%; no follow-up campaign for three years; measles outbreak in the neighbouring province last year.
• Cholera: rainy-season outbreaks in Kalombo's riverside neighbourhoods in three of the last five years.
• Latest SMART survey in the province of origin (before the conflict): GAM of 9.8%.

2. Population (D6 to D8)

The authorities announce 55,000 people. Satellite imagery makes it possible to count 6,120 shelters on the vacant plot; 120 randomly selected shelters are visited and give an average of 6.1 people. The three schools house 2,650 people registered by the site committee. Estimate adopted: 6,120 × 6.1 + 2,650, or about 40,000 people. This figure is kept as the provisional denominator, with a range of 36,000 to 44,000, and the planning proportions are applied: about 7,600 children under 5 (19%), 17,200 children aged 6 months to 14 years (43%), 1,400 pregnant women (3.5%).

3. Mortality (D8 to D21)

Forty community volunteers are recruited among the displaced population (one per 1,000 people) and trained in one day. A watchman counts graves at the two burial sites. Results for the first two complete weeks:

Interpretation: the CMR far exceeds double the baseline mortality (0.9) and crosses 2 in week 2, which characterises a situation out of control. The U5MR is rising faster than the CMR, and the occurrence of 14 deaths with fever and rash strongly points to a measles outbreak. Even using the upper population range (44,000), the CMR would remain above 1.8: the conclusion does not depend on the uncertainty about the denominator.

Estimated excess mortality over the two weeks: (1.86 average minus 0.45) × 40,000 × 14 ÷ 10,000, or about 79 excess deaths in 14 days.

4. Services and determinants (D8 to D12)

• The public health centre carries out 180 consultations a day with two nurses, i.e. 90 consultations per clinician per day; stock-outs of amoxicillin and ORS for 10 days; no functioning cold chain.
• Water: two water trucks supply about 160 m³ a day, i.e. 4 litres per person per day, far below Sphere's 15 litres.
• Latrines: 210 emergency latrines for 40,000 people, i.e. 1 per 190 people; widespread open defecation observed.
• MUAC screening of 1,450 children aged 6 to 59 months: 3.1% with MUAC below 115 mm and 12.4% between 115 and 124 mm, indicating a very worrying nutritional situation.
• No clinical management of rape services; home deliveries without kits; no obstetric referral at night.

5. Prioritisation matrix

The figure of 140 births per month comes from a planning crude birth rate of about 4% per year: 40,000 × 0.04 ÷ 12, or about 133, rounded up to 140 to allow for uncertainty.

6. 30-day plan presented to the cluster (D14)

1. D14 to D21: reactive measles vaccination campaign (target 17,200 children, coverage objective 95%), with vitamin A for 6 to 59 months and simultaneous MUAC screening; isolation of cases in a dedicated tent.
2. D14 to D18: opening of a primary health care post on site with two additional clinicians and 10 ORS points; community case management of malaria, diarrhoea and pneumonia by trained volunteers.
3. D14 to D16: MISP: clean delivery kits distributed to visibly pregnant women, midwife present on site, referral agreement with the district hospital and night-time transport, clinical management of rape kits.
4. D14 to D30: formal alert to the WASH cluster to reach at least 15 litres per person per day and one latrine per 50 people in the immediate term; cholera preparedness.
5. Ongoing: weekly mortality surveillance and EWARN; SMART mortality and nutrition survey scheduled for D45.

The key reasoning: imperfect data were enough to decide. The uncertainty about the population did not change the conclusion, and the breakdown of deaths (rising U5MR, fever with rash) pointed to the intervention that would save the most lives in the short term.

8Reusable tools: key figures, checklist and mortality surveillance form

The tools below are designed to be printed or transferred into a spreadsheet. They bring together the key figures a health manager needs in the first days, without replacing a reading of the full standards.

Key figures for the initial analysis

Initial analysis checklist

• Is the population estimated using at least two methods, with an explicit range?
• Is the baseline mortality used documented (source, year, geographical level)?
• Is prospective mortality surveillance functioning, with at least two sources (community and graveyards or facilities)?
• Are the reported causes of death disaggregated by age (under 5, 5 and over)?
• Is measles vaccination coverage known from a source other than administrative coverage?
• Has nutrition screening been carried out, even partially?
• Have all health facilities in the area been visited or documented in HeRAMS?
• Are water quantities and the number of latrines per person measured rather than estimated by eye?
• Has the availability of MISP services (delivery, referral, clinical management of rape) been verified?
• Have the seasonal epidemic risks for the next three months been identified?
• Have women, older people and persons with disabilities been consulted separately about their barriers to accessing care?
• Have the conclusions been shared with the Ministry of Health and the cluster before external dissemination?

Weekly community mortality surveillance form (template)

Weekly calculation table

9Common mistakes, donor expectations and variations by context

Common mistakes and how to avoid them

What donors, evaluators and auditors look at

• Link between needs and response: the proposal must show how mortality, morbidity and service data justify the activities chosen. DG ECHO's health policy documents stress the priority given to interventions that reduce avoidable mortality and morbidity, and the use of indicators aligned with Sphere.
• Alignment with the cluster and the Humanitarian Response Plan: donors check that areas, targets and indicators match those of the Health Cluster and the HRP, and that the organisation reports to the 4W.
• Survey quality: for nutrition and mortality, BHA (USAID) and other donors expect surveys that comply with the SMART methodology, with a protocol validated by the technical working group and a report that includes data quality tests.
• Disaggregation: sex, age and disability for outcome indicators; European and British donors require it and score it.
• Accountability: the Core Humanitarian Standard (CHS) requires communities to be informed and consulted; an evaluator will look for evidence that assessment results were shared back with communities.
• Traceability: auditors check the original data collection forms, the lists of paid volunteers and the consistency between reports and raw data.

Variations by context

In a protracted crisis, the question is no longer only "are we above the threshold?" but "why has mortality remained at this level for three years?". The analysis must then address the structural causes: health financing, human resources, supply chain breakdowns.

10Interview questions and key references

Interview questions and elements of a strong answer

1. "You arrive at a site with 30,000 displaced people and no data at all. What do you do in the first week?" A good answer mentions the secondary data review, population estimation using two methods, setting up community and graveyard mortality surveillance, facility visits, MUAC screening and, above all, the no-regret interventions launched without waiting (measles if coverage is uncertain, water, MISP, basic care), in coordination with the Ministry and the cluster.
2. "How is the CMR calculated and what does a CMR of 1.4 mean?" The formula, the unit (per 10,000 per day), comparison with baseline mortality rather than the absolute threshold, the interpretation (emergency if the baseline is below 0.7), and mention of the confidence interval if the figure comes from a survey.
3. "The U5MR is rising faster than the CMR. What are your hypotheses?" Measles, diarrhoeal diseases, malaria, worsening malnutrition; verification through reported causes, EWARN surveillance and nutrition screening; immediate actions.
4. "What are the limitations of a retrospective mortality survey?" Recall bias, survival bias, recall period, design effect, time required; complementarity with prospective surveillance.
5. "How would you present your priorities to a donor?" Mortality and excess mortality data, likelihood-impact matrix, gaps quantified against Sphere, alignment with the cluster and the HRP, monitoring indicators, exit plan.
6. "How do you work with the Ministry of Health when its data differ from yours?" Methodological transparency, comparison of sources, joint validation within the cluster, respect for the role of the State, no unilateral dissemination.

Key references

• Sphere Association, The Sphere Handbook: Humanitarian Charter and Minimum Standards in Humanitarian Response, 2018 edition, Health chapter and Water Supply, Sanitation and Hygiene Promotion chapter.
• Global Health Cluster, Public Health Situation Analysis (PHSA), guidance and template.
• Global Health Cluster, Health Cluster Guide, 2020 edition.
• SMART Initiative, SMART Manual and ENA for SMART software.
• Checchi F. and Roberts L., Interpreting and using mortality data in humanitarian emergencies: a primer for non-epidemiologists, Humanitarian Practice Network (ODI), Network Paper 52.
• IASC, Multi-Sector Initial Rapid Assessment (MIRA), revised guidance.
• WHO, Health Resources and Services Availability Monitoring System (HeRAMS), operational guide.
• WHO and UNICEF, recommendations on prevalence thresholds for malnutrition (2018).
• CHS Alliance, Groupe URD and Sphere Association, Core Humanitarian Standard on Quality and Accountability (CHS), revised version 2024.
• Médecins Sans Frontières, Refugee Health: An Approach to Emergency Situations (1997), reference work on the ten health priorities of the first weeks.