Mortality Survey & Surveillance Toolkit

Stage 1
Listen
Week 1

Stage 2

Build

Weeks 2-3

Stage 3

Test

Week 3

Stage 4

Listen

Week 4

Stage 5

Train

Weeks 4-5

Stage 6

Deliver

Week 5

10+

Deliverables

6

Toolkit Components

6

Stages

The 10 Deliverables

1

Finalised Toolkit Outline

Structure and scope agreed with TAG co-chairs before production begins.

Stage 1: Listen

2

Technical Guidance Document

60-80 page practitioner-facing reference. Survey design, sampling, analysis, surveillance, ethics. Three drafts across stages.

Stages 1-5

3

Concept Note and Protocol Templates

Adaptable planning documents for survey managers to customise for their context.

Stage 2: Build

4

Data Collection Tools

Two pathways: Survey (household listing + mortality questionnaire) and Surveillance (death notification). Verbal autopsy shared across both. Paper and XLSForm formats, multi-language, encryption-at-rest.

Stage 2: Build

5

Analysis Resources

Excel CDR/U5DR calculator, R scripts, R Shiny point-and-click application. Sample size and PPS cluster selection tools.

Stage 2: Build

5b

Validation Findings Report

"Think Aloud" usability testing results. Tool revisions documented before training development.

Stage 3: Test

6

Reporting Templates

Factsheet (1 page), slide deck, short technical report. Structured for decision-makers: Is this an emergency? Is it getting worse?

Stage 5: Train

7

Manager Training Materials

2-day training: sampling, data management, analysis, reporting, ethics. Slides, facilitator guide, exercises.

Stage 5: Train

8

Enumerator Training Materials

3-day training: consent, questionnaire administration, tablet use, distress management, data quality. Field guide included.

Stage 5: Train

9

Community Engagement Guidance

How to introduce mortality data collection, manage expectations, share results, and maintain trust.

Stage 5: Train

10

Final Toolkit Package

Integrated, quality-assured, Creative Commons licensed. Ready for the Initiative's digital hub.

Stage 6: Deliver

Timeline

Indicative 7-week schedule. Grey bars are TAG/stakeholder review periods where we pause production and wait for feedback before proceeding. Actual dates depend on contract start and TAG availability.

Activity	Wk 1	Wk 2	Wk 4	Wk 5	Wk 6	Wk 7
1. Listen - Inception	Toolkit outline, TAG briefings
TAG review of outline		Review
2. Build - Core production		Guidance, XLSForms, analysis tools, protocols
TAG review of draft tools			Review
3. Test - Validation			"Think Aloud" usability testing
4. Listen - Incorporate feedback				Revise tools based on testing
5. Train - Training materials				Manager, enumerator, community engagement materials
TAG review of full package					Review
6. Deliver - Final package						QA, package, handover

Consultant production TAG/stakeholder review (we pause, you review) Usability testing Final delivery

Six Standards We Build On

Resource	What we use it for
SMART survey protocols	Sampling and survey design framework
WHO 2022 Verbal Autopsy	Cause-of-death assessment standard
Africa CDC 2024 guide	Surveillance approaches
WHO Health Cluster guidance	Emergency thresholds and indicators
UNHCR HIS toolkit	Data flow in refugee settings
Academic research partnerships	Emerging methods (e.g. mobile phone surveys)

Toolkit Structure

Mortality Survey and Surveillance Toolkit v1.0
Creative Commons BY 4.0 | Mortality Estimation Initiative

0_Sampling_Tools/
    Sample_Size_Calculator.R   # incl. fieldwork team planner
    Sample_Size_Calculator.xlsx # incl. fieldwork team planner
    PPS_Cluster_Selection.R
    PPS_Cluster_Selection.xlsx

1_Technical_Guidance/
    Technical_Guidance_v1.pdf          # 60-80 pages, 10 chapters

2_Planning_Templates/
    Concept_Note_Template.docx
    Technical_Protocol_Template.docx

3_Data_Collection/
    Survey/
        Paper/
            Household_Listing.pdf
            Mortality_Questionnaire.pdf
        XLSForm/
            household_listing_v1.xlsx      # Pre-survey cluster listing
            mortality_survey_v1.xlsx       # Retrospective HH survey
    Surveillance/
        Paper/
            Death_Notification.pdf
        XLSForm/
            death_notification_v1.xlsx     # CHW continuous reporting
    Shared/
        Paper/
            Verbal_Autopsy.pdf
        XLSForm/
            verbal_autopsy_v1.xlsx         # VA module (both pathways)

4_Analysis/
    Excel/
        CDR_U5DR_Calculator.xlsx
        Data_Cleaning_Template.xlsx
    R/
        01_data_cleaning.R
        02_cdr_calculation.R
        03_confidence_intervals.R
        shiny_app/app.R                # Point-and-click interface

5_Reporting/
    Factsheet_Template.pptx
    Short_Technical_Report.docx
    Preliminary_Findings_Slides.pptx

6_Training/
    Manager/   # 2-day training package
    Enumerator/ # 3-day training package
    Community_Engagement/

7_Quick_Reference/
    Indicator_Definitions.pdf
    Emergency_Thresholds.pdf
    Local_Event_Calendar_Template.xlsx

Chapter Outline

Ch.	Title	Content	Pages
1	Introduction	Why mortality estimation matters. Emergency thresholds. How to use this toolkit.	4-5
2	Choosing Your Approach	Survey vs surveillance: when each is appropriate. Hybrid approaches.	4-6
3	Survey Design	Target population, two-stage cluster sampling with PPS, sample size, recall period, indicators (CDR, U5DR, neonatal, maternal, CSMF).	12-15
4	Surveillance Design	Death notification (facility and community), VA integration, denominator estimation, completeness.	8-10
5	Data Collection	Paper and XLSForm instruments. Household listing. Mortality questionnaire. VA administration.	8-10
6	Data Management	Transfer, backup, minimum cleaning standards, common errors.	4-6
7	Analysis	CDR/U5DR calculation with CIs, design effect, CSMF, threshold interpretation, bias documentation.	8-10
8	Reporting	Factsheets for decision-makers. Technical reports. Data sharing and archiving.	4-5
9	Training	Manager and enumerator training overviews. Community engagement.	4-5
10	Ethics	Informed consent, distress management, data protection, "do no harm", VA ethics.	4-6

Tone: Short paragraphs. Decision trees and flowcharts. Worked examples. "If you are in this situation, do X." The survey manager is the primary audience.

Concept Note Template

2-3 page document for internal approval

Section	Content
Background	Why mortality data is needed in this context
Objectives	Primary: estimate CDR and U5DR. Secondary indicators.
Methodology	Survey vs surveillance. Sampling approach. Recall period.
Timeline	Planning, training, fieldwork, analysis, reporting
Budget	Personnel, transport, devices, per diem
Ethics	IRB requirements. Consent. Data protection.
Expected use	How will findings inform programming decisions?

Technical Protocol Template

8-12 page document for technical review and IRB

Section	Content
Study design	Cross-sectional, retrospective mortality survey
Target population	Definition, inclusion/exclusion criteria
Sampling design	Two-stage cluster sampling with PPS. Sample size calculation with assumptions.
Recall period	Duration, anchor event, calendar tool
Data collection	Instruments, deployment platform, language
Quality assurance	Training plan, daily checks, supervision structure
Analysis plan	CDR/U5DR, CIs, DEFF, disaggregation
Ethical review	Consent, distress, anonymisation, data retention
Dissemination	Reporting products, timeline, audience
Limitations	Survival bias, recall bias, denominator uncertainty

Two Pathways, Shared Standards

The toolkit provides separate data collection instruments for survey and surveillance. Each pathway has its own forms, deployment model, and data flow. Verbal Autopsy is shared across both.

Survey Pathway

One-off retrospective household survey. Cluster sampling, fixed recall period. Two forms: Household Listing (pre-survey) and Mortality Survey (main instrument).

Surveillance Pathway

Continuous death reporting by CHWs or facility staff. No sampling frame needed. One form: Death Notification (short mobile form, ongoing).

Shared

Verbal Autopsy (WHO 2022) can be triggered from either pathway when cause-of-death data is needed for under-5, maternal, or unknown-cause deaths.

Mortality Survey

Household Listing

Death Notification

Verbal Autopsy

SURVEY PATHWAY

Mortality Survey Questionnaire (XLSForm)

Bilingual (English/French). Skip logic, constraints, hint text, encryption-at-rest. Deployed to tablets for retrospective household interviews during a cluster survey.

type	name	label::English	label::French	relevant	constraint	hint
begin_group	metadata	Survey Metadata	Metadonnees
select_one cluster	cluster_id	Cluster	Grappe			Select from assigned clusters
integer	hh_number	Household number	Numero du menage		.>0	Sequential within cluster
geopoint	hh_gps	GPS location	Position GPS
date	interview_date	Date of interview	Date de l'entretien
end_group
begin_group	household	Household Information	Informations sur le menage
integer	hh_size	Current household size	Taille actuelle du menage		.>0 and .<=50	All persons sleeping here
integer	hh_under5	Children under 5	Enfants de moins de 5 ans		.>=0 and .<=${hh_size}
integer	births	Births since [ANCHOR]	Naissances depuis [REPERE]		.>=0
integer	arrivals	Arrivals since [ANCHOR]	Arrivees depuis [REPERE]		.>=0	People who joined the HH
integer	departures	Departures since [ANCHOR]	Departs depuis [REPERE]		.>=0	People who left the HH
integer	num_deaths	Deaths since [ANCHOR]	Deces depuis [REPERE]		.>=0	Include all ages
end_group
begin_repeat	death_detail	Death Details	Details du deces	${num_deaths}>0
text	deceased_name	Name of deceased	Nom du defunt
select_one sex	deceased_sex	Sex	Sexe
integer	deceased_age	Age at death (years)	Age au deces		.>=0 and .<=120
date	death_date	Date of death	Date du deces		>=${recall_start}	Must be in recall period
select_one cause	probable_cause	Probable cause	Cause probable
select_one yn	sought_care	Sought care?	A cherche des soins?
note	va_flag	VA may be needed	Autopsie verbale requise	${deceased_age}<5 or ${probable_cause}='maternal'
end_repeat

Key XLSForm Features

Skip logic: Death details only appear when deaths > 0. VA flag triggers on under-5 or maternal deaths.

Constraints: Age 0-120, dates within recall period, household size > 0, under-5 count cannot exceed total.

Encryption: Configured for encryption-at-rest on KoBoToolbox/ODK. Sensitive mortality data protected on device.

SURVEY PATHWAY

Household Listing Form

Used during the pre-survey phase to create a complete list of households in each selected cluster. From this list, the required number of households is randomly selected.

Field	Type	Purpose
Cluster ID	select_one	Assigned cluster from PPS selection
Household number	integer	Sequential listing number
GPS	geopoint	Location for mapping and verification
Head of household	text	Identification for sampling
Household size	integer	Number of current members
Structure type	select_one	Dwelling classification
Occupied status	select_one	Occupied / Empty / Destroyed

The "Occupied status" field is critical for documenting survival bias: destroyed or empty dwellings may indicate households that died or fled.

SHARED - BOTH PATHWAYS

Verbal Autopsy Form (WHO 2022 aligned)

Optional module triggered from either survey or surveillance when cause-of-death data is needed. Aligned with the WHO 2022 VA instrument covering 66 causes mapped to ICD-10.

Section	Fields	Duration
Identification	Deceased demographics, respondent relationship, date of death	3-5 min
Signs and symptoms	Duration of illness, specific symptoms (fever, cough, diarrhoea, bleeding, etc.), progression	10-15 min
Medical history	Known conditions, previous treatment, pregnancies (for women 12-49)	5-8 min
Care-seeking	Health facility visits, treatments received, barriers to care	5-7 min
Circumstances	Injury details (if applicable), place of death, final hours	5-8 min
Open narrative	Respondent's own account of the illness and death	5-10 min

Coding: PCVA (physician review, gold standard) or algorithmic (SmartVA, InterVA, InSilicoVA). The toolkit provides guidance on both approaches.

SURVEILLANCE PATHWAY

Community Death Notification Form

For community health workers to report deaths as they occur. Short form designed for mobile reporting. Continuous data collection, no recall period or sampling frame.

Field	Type	Notes
Reporting CHW	select_one	Pre-populated from CHW register
Community	cascading select	Region > District > Village
Date of death	date	Constrained to last 30 days
Deceased age	integer	Years (0 for neonates)
Deceased sex	select_one
Place of death	select_one	Home / Facility / Transit / Other
Probable cause	select_one	Simplified cause list (8-10 categories)
Maternal death flag	select_one yn	Triggers MPDSR notification
VA referral needed	calculate	Auto-flagged for under-5, maternal, or unknown cause

Planning Tools

Before data collection begins: determine sample size and plan fieldwork logistics.

⬇ Download Sample Size + Fieldwork Planner

R Shiny App

Excel Calculator

R Scripts

R Shiny Application - Point-and-Click Interface

The user never sees R code. Upload CSV, set parameters, click Generate.

Input Panel

Upload survey data (CSV)

Recall period (days)

Mid-period population

Emergency CDR threshold

Output Panel

Crude Death Rate

1.24 / 10,000 / day

95% CI: 0.82 - 1.66

Emergency threshold: 1.0 | Point estimate: ABOVE

Under-5 Death Rate

1.87 / 10,000 / day

95% CI: 1.12 - 2.62

Emergency threshold: 2.0 | Point estimate: BELOW

Design effect (DEFF): 1.78 | Clusters: 28 | Households: 625

The interactive CDR/U5DR Calculator below demonstrates the live calculation logic. The full R Shiny app adds CSV upload, disaggregation, cluster-level charts, and PDF report download.

▶ Try the Live R Shiny App ⬇ Download Demo Data to Upload Download the demo CSV, then upload it to the Shiny app to see it in action.

Excel CDR/U5DR Calculator

For teams with no R capacity. Named ranges, protected formulas, no macros.

Tab	Content	User action
1. Data Entry	Cluster-level summary: cluster ID, households, person-days, deaths, under-5 deaths	Paste from KoBoToolbox export
2. CDR Calculation	CDR formula with confidence interval. Threshold comparison with conditional formatting.	Automatic
3. U5DR Calculation	Under-5 death rate with CI. Emergency threshold: 2.0/10,000/day.	Automatic
4. Disaggregation	By sex, age group, geographic area. Pivot table format.	Select disaggregation
5. Dashboard	Visual summary: CDR, U5DR, threshold comparison. Copy-paste ready.	Copy to report

⬇ Download CDR/U5DR Calculator ⬇ Download Sample Size + Fieldwork Planner

Documented R Scripts

For teams with R capacity who want full control over the analysis.

# 01_data_cleaning.R
# Load raw export, check for:
#   - Missing cluster/household IDs
#   - Household size inconsistencies
#   - Death dates outside recall period
#   - Age and sex completeness for deaths
# Output: survey_data_clean.csv

# 02_cdr_calculation.R
# Calculate person-time per household:
#   mid_period_hh_size * recall_days
# Aggregate to cluster level
# CDR = (total deaths / total person-days) * 10000
# Flag: above/below 1.0 emergency threshold

# 03_confidence_intervals.R
library(survey)
design <- svydesign(
  ids = ~cluster_id,
  data = hh_data,
  weights = ~1
)
cdr_mean <- svymean(~death_rate_10k, design)
confint(cdr_mean)   # 95% CI
deff(cdr_mean)      # Design effect

Mortality Survey Factsheet (1 page)

Decision-makers read the top third. Methodologists read the bottom. Structured around: Is this an emergency? Is it getting worse? Where should we concentrate?

Mortality Survey Results

Jonglei State, South Sudan | March 2026 | 90-day recall period

1.24

CDR /10k/day

ABOVE THRESHOLD

1.87

U5DR /10k/day

BELOW THRESHOLD

625

Households

28 clusters

What This Means

The crude death rate exceeds the emergency threshold of 1.0/10,000/day (95% CI: 0.82-1.66), indicating elevated mortality. Under-5 mortality is below the 2.0 threshold but requires close monitoring.

Leading Causes of Death (Verbal Autopsy, n=47)

Malaria

28%

Diarrhoea

19%

Pneumonia

14%

Injury/violence

9%

Other/unknown

30%

Methodology

Two-stage cluster sampling (PPS), 28 clusters, 625 households. 90-day recall anchored to the start of the dry season. DEFF: 1.78. Limitations: survival bias in areas affected by displacement; 8 destroyed dwellings recorded during household listing.

Training Outline

Day 1, Morning: Survey Design

Sampling methodology. Two-stage cluster sampling with PPS. Using the sample size calculator. Adapting for context: camps, dispersed populations, mixed settings. Recall period selection and anchoring.

Day 1, Afternoon: Data Management

Setting up KoBoToolbox project. Deploying XLSForm instruments. Monitoring submissions in real-time. Running daily quality checks: completeness, duration outliers, GPS consistency.

Day 2, Morning: Analysis and Interpretation

Using the Excel calculator or R Shiny app. Calculating CDR and U5DR. Understanding confidence intervals. Design effect. Interpreting results against emergency thresholds.

Day 2, Afternoon: Reporting and Ethics

Using reporting templates. Presenting to decision-makers. Documenting limitations (survival bias, recall bias). Ethical responsibilities: consent, distress, data protection.

Package Contents

Material	Format	Pages/Slides
Training slide deck	PPTX	60-80 slides
Facilitator guide	PDF	15-20 pages
Manager reference guide	PDF	8-10 pages
Exercises and worked examples	PDF + Excel	5-8 exercises
Daily QC checklist	PDF (laminated)	1 page

Training Outline

Day 1, Morning: Introduction

Why we measure mortality. What CDR means. How the data will be used. The role of the enumerator in the data chain.

Day 1, Afternoon: Consent and Ethics

Informed consent script (practised until fluent). Respondent distress management: recognising distress, pausing interviews, referral pathways. "Do no harm." When to stop an interview.

Day 2, Morning: The Questionnaire

Household listing: who counts as a household member. Mortality module: recall period, anchor event, recording deaths. VA module (if applicable). Walk through every question.

Day 2, Afternoon: Tablet Training

Using KoBoToolbox/ODK. Navigating the XLSForm. Skip logic in practice. Uploading data. Troubleshooting common issues. GPS recording.

Day 3, Morning: Practice Interviews

Role-play with fellow trainees. Then practice with real households in the local area (pilot). Debriefing after each practice session.

Day 3, Afternoon: Data Quality and Logistics

Daily checks: what supervisors look for. Common errors. Daily schedule. Transport. Communication with supervisors. Safety and security protocols.

Package Contents

Material	Format
Training slide deck	PPTX (40-50 slides)
Enumerator field guide	PDF (laminated pocket card + 8-page booklet)
Consent script	DOCX (adaptable to context and language)
Practice exercises	PDF (role-play scenarios)
Troubleshooting guide	1-page reference

Guidance Structure

Phase	Activities	Key messages
Before the survey	Meet community leaders. Explain purpose, process, and what will happen with the data. Address concerns about asking about deaths.	"We are collecting this information to understand the health situation and improve services. Participation is voluntary."
During the survey	Ensure enumerators identify themselves clearly. Allow community observers if requested. Maintain sensitivity around bereaved households.	"If at any point you feel uncomfortable, you can stop the interview. This will not affect your access to services."
After the survey	Share summary findings with community leaders. Explain what actions will follow. Thank the community for participation.	"The information you shared is being used to [specific action]. Here is what we found and what happens next."

Specific Considerations

Verbal autopsy sensitivity: Asking bereaved families to recount death circumstances. Informed consent must acknowledge this explicitly. Distress management protocols are mandatory, not optional.

Conflict settings: Identifying which households experienced deaths can expose respondents to risk. Discreet approaches to household identification. No household-level data shared externally.

Draws on published research on community participation in MPDSR, and the WHO MDSR "no shame, no blame" principle.

Quality Assurance Checklist

All XLSForms validated on KoBoToolbox and ODK	✓
R Shiny app tested with simulated data	✓
Excel calculator tested with edge cases	✓
Technical guidance peer-reviewed by TAG	✓
"Think Aloud" usability testing completed	✓
Multi-language columns verified (EN/FR)	✓
Creative Commons BY 4.0 licensing applied	✓
README and version control documentation	✓
Ready for digital hub integration	✓

Sustainability Design

Principle	How
Open source	R, XLSForm (open standard), Excel. No proprietary dependencies.
Modular	VA module updated independently of sampling guidance. Each component self-standing.
Creative Commons	BY 4.0 licence. Anyone can adapt, translate, redistribute.
TAG custody	Designed for technically competent PH professionals to maintain.
Digital hub	Structured for the READY Initiative model. Multi-lingual, downloadable, web-hosted.

📱

KoBoToolbox Collection

➔

📄

CSV Export

➔

🔎

Data Cleaning

➔

📈

CDR/U5DR Analysis

➔

📋

Factsheet / Dashboard

KoBoToolbox Collection: Household listing and mortality survey forms deployed to tablets. Data syncs automatically when connectivity is available. Encryption-at-rest protects sensitive mortality data.

CSV Export: Download from KoBoToolbox or ODK Central as CSV. The demo data below simulates a 28-cluster mortality survey export with 676 household records.

Data Cleaning: Check for missing cluster/household IDs, household size inconsistencies, death dates outside recall period, age/sex completeness. The Excel template or R script handles these checks.

CDR/U5DR Analysis: Calculate person-time per household, aggregate to cluster level, compute CDR and U5DR with confidence intervals. Compare against emergency thresholds.

Factsheet / Dashboard: One-page factsheet for decision-makers: Is this an emergency? Is it getting worse? Structured for Health Cluster coordination meetings.

Demo Data

First 10 rows of a simulated 28-cluster, 676-household KoBoToolbox export.

_id	cluster_id	cluster_name	household_id	hh_size	hh_under5	births	arrivals	departures	num_deaths
1	C01	Bor Town	C01-001	6	2	0	1	0	0
2	C01	Bor Town	C01-002	4	1	1	0	0	0
3	C01	Bor Town	C01-003	7	2	0	0	1	1
4	C02	Makuach	C02-001	5	1	0	0	0	0
5	C02	Makuach	C02-002	3	0	0	1	0	0
6	C03	Anyidi	C03-001	8	3	1	0	0	1
7	C03	Anyidi	C03-002	5	1	0	0	1	0
8	C04	Kolnyang	C04-001	4	1	0	0	0	0
9	C04	Kolnyang	C04-002	6	2	0	1	0	1
10	C05	Pariak	C05-001	5	1	0	0	0	0

Download Demo Data (CSV)

Aggregated Demo Summary

Totals across all 28 clusters and 676 households in the demo dataset.

3,474

Total Population

709

Under-5 Population

31

Total Deaths

10

Under-5 Deaths

90

Recall Days

28

Clusters

How this works

Before a mortality survey, you need to know how many households to visit. Too few and your estimate is too imprecise to tell whether the population is in crisis. Too many and you waste time and resources in an emergency. This calculator determines the minimum sample size needed to produce a CDR estimate with enough precision to compare against the emergency threshold (1.0/10,000/day).

The formula starts with the number of person-days of observation needed for a given precision, then adjusts upward for cluster sampling (design effect) and non-response, and finally converts from persons to households.

Parameter	What it means	How to decide
Expected CDR	Your best guess of the death rate before the survey. This is the rate you are designing the survey to detect. Lower expected rates require larger samples because deaths are rarer events.	Use 0.5 (sub-Saharan Africa baseline) if no prior data. Use a higher value if you suspect a crisis.
Desired precision	How close your estimate needs to be to the true rate. A precision of +/-0.3 means if the true CDR is 1.0, your survey result will fall between 0.7 and 1.3. Tighter precision = larger sample.	0.3 is standard. Use 0.5 for rapid assessments when speed matters more than precision.
Recall period	How far back you ask households about deaths. Longer recall captures more deaths (reducing the sample needed) but introduces recall bias - people forget or misdate events.	90 days is the standard for SMART surveys. Use 60 days in acute emergencies; up to 120 if population is stable.
Household size	Average number of people per household. This converts the number of persons needed into a number of households to visit.	Obtain from census data, prior surveys, or humanitarian registration. Typically 4-7 in sub-Saharan Africa.
Design effect (DEFF)	A multiplier that accounts for cluster sampling. Households within the same village tend to have similar mortality (shared exposures, same health facility, same conflict events). This correlation means you need more households than simple random sampling would require.	Use 2.0 as default. If prior surveys in similar settings report DEFF, use that. Range: 1.5 (low correlation) to 3.0 (high, e.g. famine pockets).
Non-response rate	Percentage of selected households you expect to be unable to interview - absent, refused, inaccessible. The sample is inflated to compensate.	5% in stable settings, 10% standard, up to 20% in active conflict or displacement.
Number of clusters	How many geographic locations (villages, camps, wards) you visit. More clusters with fewer households per cluster is statistically better than fewer clusters with more households, because it reduces the design effect's impact on your confidence interval.	25-30 is the accepted standard (SMART, EPI). Fewer than 20 produces unreliable confidence intervals. Go higher if logistically feasible.

Fieldwork Team Planning (below the calculator) converts your sample size into an operational staffing plan: how many enumerators, supervisors, teams, and days you need - accounting for travel between clusters, training time, and buffer days.

Parameters

Expected CDR (/10,000/day)

Your best estimate of the current death rate. Baseline sub-Saharan Africa: ~0.5

Desired precision (+/-)

How narrow the confidence interval should be. Standard: 0.3

Recall period (days)

How far back you ask about deaths. Standard: 90 days

Average household size

Persons per household. Obtain from census or prior surveys

Design effect (DEFF)

Cluster sampling multiplier. Default 2.0, range 1.5-3.0

Non-response rate (%)

Expected % of households you cannot interview. Standard: 10%

Number of clusters

Geographic locations to visit. Standard: 25-30 (SMART/EPI)

Results

Adjust parameters and click Calculate.

Reading the results

SRS person-days needed - the theoretical number of person-days of observation required under simple random sampling. This is the statistical starting point.

SRS persons needed - person-days divided by the recall period. The number of individuals you would need if you could randomly select from the entire population.

After DEFF adjustment - multiplied by the design effect because cluster sampling is less efficient than simple random sampling. This is the real number of persons you need.

After non-response - inflated to account for households you cannot interview.

Households - persons divided by average household size. This is what your field teams actually count.

Per cluster / Total - households distributed evenly across clusters, rounded up to a whole number per cluster, then multiplied back. The final total is always a round number (clusters x households per cluster).

Fieldwork Team Planning

Once you know the sample size, you need to plan the field team. How many enumerators? How many supervisors? How long will fieldwork take? This section converts your sample size into an operational staffing and logistics plan.

Interviews per enumerator per day

Mortality surveys typically achieve 4-6 per day. Depends on terrain, household density, and interview length (~30-45 min each).

Travel days between clusters

Days needed to move the team from one cluster to the next. 0.5 if clusters are close; 1-2 in dispersed, conflict-affected or mountainous settings.

Enumerators per team

Typically 4-6 enumerators per team. Larger teams finish faster but are harder to supervise and transport.

Supervisor ratio (enumerators per supervisor)

1 supervisor per 4-6 enumerators is standard. Supervisors run quality checks, manage logistics, and handle community entry.

Number of field teams

More teams = shorter fieldwork but higher cost (vehicles, supervisors, coordination). 2-3 teams is typical.

Training days

3 days for enumerator training is standard. Add 1-2 days for pilot testing in the field.

Buffer days (weather, access, rest)

Allow for unexpected delays: rain, road closures, security incidents, rest days. 1-3 days depending on context.

Fieldwork Plan

Results will update automatically based on sample size and fieldwork parameters.

Cluster Population Data

Demo data: Bor South County, Jonglei State

Cluster	Name	Population	Cumulative

Clusters to select

Selection Results

Click "Run PPS Selection" to see results.

Analysis Mode

The core CDR/U5DR formula is the same for both survey and surveillance. The difference is in how the denominator (person-time) is estimated.

Survey mode: Person-time derived from sampled households. Mid-period population estimated from current household size adjusted for births, deaths, arrivals, and departures. Design effect applied. Cluster-level confidence intervals.

Survey Data

Total population surveyed

Recall period (days)

Total deaths (all ages)

Under-5 population

Under-5 deaths

Number of clusters

Results

Enter survey data and click Calculate.

Toolkit Dashboard

The 10 Deliverables

Finalised Toolkit Outline

Technical Guidance Document

Concept Note and Protocol Templates

Data Collection Tools

Analysis Resources

Validation Findings Report

Reporting Templates

Manager Training Materials

Enumerator Training Materials

Community Engagement Guidance

Final Toolkit Package

Timeline

Six Standards We Build On

Deliverable #1: Toolkit Outline

Toolkit Structure

Deliverable #2: Technical Guidance Document

Chapter Outline

Deliverable #3: Concept Note and Protocol Templates

Concept Note Template

Technical Protocol Template

Deliverable #4: Data Collection Tools

Two Pathways, Shared Standards

Survey Pathway

Surveillance Pathway

Shared

Mortality Survey Questionnaire (XLSForm)

Key XLSForm Features

Household Listing Form

Verbal Autopsy Form (WHO 2022 aligned)

Community Death Notification Form

Planning Tools

Deliverable #5: Analysis Resources

R Shiny Application - Point-and-Click Interface

Input Panel

Output Panel

Excel CDR/U5DR Calculator

Documented R Scripts

Deliverable #6: Reporting Templates

Mortality Survey Factsheet (1 page)

Mortality Survey Results

What This Means

Leading Causes of Death (Verbal Autopsy, n=47)

Methodology

Deliverable #7: Manager Training Materials

Training Outline

Day 1, Morning: Survey Design

Day 1, Afternoon: Data Management

Day 2, Morning: Analysis and Interpretation

Day 2, Afternoon: Reporting and Ethics

Package Contents

Deliverable #8: Enumerator Training Materials

Training Outline

Day 1, Morning: Introduction

Day 1, Afternoon: Consent and Ethics

Day 2, Morning: The Questionnaire

Day 2, Afternoon: Tablet Training

Day 3, Morning: Practice Interviews

Day 3, Afternoon: Data Quality and Logistics

Package Contents

Deliverable #9: Community Engagement Guidance

Guidance Structure

Specific Considerations

Deliverable #10: Final Toolkit Package

Quality Assurance Checklist

Sustainability Design

Data Pipeline: KoBoToolbox to Dashboard

Demo Data

Aggregated Demo Summary

Sample Size Calculator

How this works

Parameters

Results

Reading the results

Fieldwork Team Planning

Fieldwork Plan

PPS Cluster Selection

Cluster Population Data

Selection Results