SMART Expert Panel Meeting Courtland Robinson Meas

Methods to Measure Mortality: 

Methods to Measure Mortality SMART Meeting July 14, 2004

Quadrat Sampling: 

Quadrat Sampling

Beldangi Camp, Nepal: 

Beldangi Camp, Nepal

Slide4: 

Community Map (IDP camp in Gulu, Uganda)

Measuring Mortality: 

Measuring Mortality What is the context/setting in which we are gathering information? Type of disaster/emergency Phase of emergency Setting (urban/rural, scattered/concentrated, mixed/unmixed) Resources: human, financial Constraints: time, security, geography

Measuring Mortality: 

Measuring Mortality What level of precision do we need? What are the impacts of an imprecise estimate? We could fail to deliver a life-saving intervention to people who need it We could waste resources

Measuring Mortality: 

Measuring Mortality What methods are most appropriate and in what sequence? Preparedness planning Observation Participatory appraisal Surveys Surveillance In-depth sector assessments Validation/cross-checking/triangulation

Methods to Assess Mortality: 

Methods to Assess Mortality Sphere Project’s Minimum Standards in Health Services: Initial Assessment Total disaster-affected population Sex and age breakdown (at least <5 and >5) Average family and household size Crude mortality ( deaths per 10,000 per day) Under-5 mortality (deaths per 10,000 per day) Age and sex-specific incidence rates of major problems and diseases

Methods to Assess Mortality: 

Methods to Assess Mortality Observation Body counts, burial/cremation site activity Participatory Appraisal Community/key informant estimates: ask numbers of deaths in last day/week/month, general age groups, pre-arrival/post-arrival, cause of death Community mapping: locate health clinics, places of worship). Ask about mortality patterns, customs about death (rituals, taboos, disposal of the body, etc.)

Methods to Assess Mortality: 

Methods to Assess Mortality Surveillance Passive: Clinic data Active: Burial sites, body counts, neighborhood interviews Surveys Cross-checking of data sources

Active vs. Passive Surveillance, Bangladesh Camps, 1992 (Diskett, 2003): 

Active vs. Passive Surveillance, Bangladesh Camps, 1992 (Diskett, 2003)

Population Pyramids: Nigeria and USA: 

Population Pyramids: Nigeria and USA

Cambodia (Kampuchea), 1962: 

Cambodia (Kampuchea), 1962 (Holck and Cates, 1982)

Sakaeo and Khao I Dang Camps, 1980: 

Sakaeo and Khao I Dang Camps, 1980

Gaza, West Bank (Fafo, 1993): 

Gaza, West Bank (Fafo, 1993)

Refugee Populations Under 18 (UNHCR, 2004): 

Refugee Populations Under 18 (UNHCR, 2004)

Mortality Rates: 

Mortality Rates Mortality Crude Death Rate (CDR or CMR) = deaths during interval/mid-interval population Infant Mortality Rate (IMR) = deaths of infants under 1 year/ 1,000 live births Under-Five Mortality Rate (U5MR) = deaths of children <5/ 1,000 live births (as used by UNICEF, WHO, etc. expresses the probability of dying by exact age 5 [ 5q0]) Not to be confused with Age-Specific Death Rate of Children <5 (ASDR-U5 or U5DR)= deaths of children <5/ mid-interval pop. of children <5

Methods to Assess Mortality: 

Methods to Assess Mortality Calculating Crude Death Rate CDR=deaths during interval/mid-interval population Question: A local official tells you of reports of 50 deaths among 10,000 IDPs. What is the CDR, calculated as deaths/10,000/day?

Methods to Assess Mortality: 

Methods to Assess Mortality Calculating Crude Death Rate CDR=deaths during interval/mid-interval population Answer: Trick question. You cannot calculate the Crude Death Rate because you are given no time period within which these deaths reportedly occurred.

Methods to Assess Mortality: 

Methods to Assess Mortality Calculating Crude Death Rate CDR=deaths during interval/mid-interval population Question: You have done a survey of IDPs and found that in a sample of 5,000 people, 15 deaths were reported in the last month. What is the CDR, calculated as deaths/10,000/day?

Methods to Assess Mortality: 

Methods to Assess Mortality Answer: 15 deaths/5,000/30 days = 30 deaths/10,000/30 days= 1 death/10,000/day or 15/deaths/5,000/30 days= 0.5/deaths/5,000/day= 1 death/10,000/day

Methods to Assess Mortality: 

Methods to Assess Mortality Units of expression are adjustable: For example, 1 death/10,000/day=36.5 deaths/1,000/year = 3.65% of the population died in one year The Crude Mortality Rate includes all causes of death for all age groups, thus is sensitive to age composition.

Which population has the higher crude death rate?: 

Which population has the higher crude death rate?

Methods to Assess Mortality: 

Methods to Assess Mortality Mortality rates should be standardized from camps/settlements/critical areas to national, baseline rates or other comparative data. Mortality survey instruments should gather data on events that contribute to exposure time: deaths, births, in-migration and out-migration.

Household “Census”: 

Household “Census”

Past Household Census: 

Past Household Census

Past Household Census Methodology: 

Past Household Census Methodology

Current Household Census: 

Current Household Census

Current Household Census Methodology: 

Current Household Census Methodology

Past vs. Current: 

Past vs. Current

Two Options to Factor in Migration: 

Two Options to Factor in Migration

Two Options to Factor in Migration: 

Two Options to Factor in Migration

Partial Birth History: 

Partial Birth History

Methods to Assess Mortality: 

Methods to Assess Mortality Things to think about in designing mortality surveys: Is this for response or “for the record?” What level of precision is needed? What period of reporting or recall is used? What events contribute to h’hold change, exposure time? What kinds of death may be more critical to the life of the population? How to define h’hold, family? How frequently surveys are conducted?

Methods to Assess Mortality: 

Methods to Assess Mortality Reporting/recall periods periods may be shorter (30 days) or longer (1 year) but must be consistent and stable shorter periods: miss earlier mortality trends highly sensitive to “calendar” error may capture relatively small number of deaths (1 death/10,000/day = 15 deaths/5,000/month)

Methods to Assess Mortality: 

Methods to Assess Mortality Mortality surveys are sensitive to error: recall error “calendar” error age heaping/digit preference (numbers ending in 0 or 5) age misreporting (children <1 often identified as 1, adults tend to overstate age) sensitivity/taboos about death

Methods to Assess Mortality: 

Methods to Assess Mortality Mortality surveys are sensitive to error: Mistranslation Interviewer error Data-entry error Analytic error

Slide39: 

Methods to Assess Mortality Mortality surveys Recommended sample size: 900 households (30 x 30 cluster sample) Recommended recall period: 6 months - 1 year Capture key events that contribute to household change, exposure time Get age, sex, date, cause of death If using a questionnaire, pre-test, back-translate Use timeline and/or seasonal calendar to help people recall ages, dates.

Timeline, Uganda: 

Timeline, Uganda

Methods to Assess Mortality: 

Methods to Assess Mortality Mortality surveys Participate with decision-makers, stake-holders to define objectives, necessary precision Establish survey, sampling methodology Identify and document survey, sampling design (note departures from plan) Establish and document training curriculum Establish and document analysis plan Document and share results promptly