Close
Explore more Centres, Projects and Groups
Welcome
Welcome Banner
An enumerator from the RAMMPS fieldwork team in the Democratic Republic of the Congo making calls for the mobile-phone mortality survey

Rapid Mortality Mobile Phone Surveys (RaMMPS)

Population-based mortality statistics are essential components of an effective public health response to epidemic outbreaks or other crisis situations. The RaMMPS consortium is developing and implementing an innovative mobile-phone based approach to generate national mortality estimates, including excess mortality, where civil registration is incomplete and face-to-face data collection is interrupted or hindered.

Bottom Content
RAMMPS paragraph
Paragraph
Logos: Bill and Melinda Gates Foundation, Burkina Faso, International Centre for Diarrhoeal Disease Research Bangladesh , Institute of Epidemiology Disease Control and Research, Instituto Nacional de Saudi, Institut national de la statistique et de la démographie, Institute of Public Opinion and Research, JiVitA, Johns Hopkins University, Universide de Kinshasa, LSHTM, Malawi Epidemiology Intervention Research Unit, New York University (Abu Dhabi), UCLouvain, UKAid, USAID
About
About RAMMPS 2 columns
About RAMMPS 2 columns left paragraph
Paragraph

Population-based mortality statistics are a key element informing public health policies and the response to epidemic outbreaks or other crisis situations. Ideally, mortality statistics are derived from civil registration and vital statistics systems (CRVS), but in many low- and lower-middle income countries, these are insufficient to fulfil that role. In these settings, mortality estimates are usually based on surveys and censuses, but these are organised infrequently and often interrupted for the same reasons that cause mortality to rise. The RaMMPS project seeks to address this data gap through the development and piloting of methods for conducting mortality surveillance via mobile phone surveys.

The RaMMPS project was launched in December 2020 and data are being collected in Bangladesh, Burkina Faso, the Democratic Republic of Congo, Malawi and Mozambique.

RaMMPS objectives

The RaMMPS project is implemented as a series of five country case studies (Figure 1) with two components:

(i) a national RaMMPS that is implemented as a series of repeated cross-sectional surveys implemented (in most cases spread over a 12-month period), and

(ii) targeted sub-studies focused on populations covered by other data sources for validation purposes. The overall aim of the project is to evaluate the use of mortality mobile phone surveys for estimating (excess) mortality during the COVID-19 pandemic in five LLMICs. More specific objectives of this project are to:

  • Develop and validate RaMMPS instruments and methods for mortality estimation
  • Generate mortality estimates during COVID-19 by age (with a focus on overall, old age and under-five mortality), sex and regional locality, and compare them to a pre-COVID-19 baseline.
  • To evaluate RaMMPS as a cost-effective approach for mortality surveillance beyond COVID-19.
Partners

The RaMMPS project is implemented through collaboration between universities, governmental and non-governmental organizations (NGOs) in the USA, the UK, Belgium, Abu Dhabi and each of the five countries where RaMMPS will be implemented. Figure 1 illustrates the geographic spread of the RaMMPS country case studies. 

Map showing the geographic spread of the RAMMPS country case studies
Figure 1
Funders
Bill and Melinda Gates, UKAID, and USAID logos

The RaMMPS project is supported by grant funding to the US Fund for UNICEF from the Bill & Melinda Gates Foundation (through an agreement with UNICEF-USA) with supplementary funding from the Foreign, Commonwealth & Development Office for the DRC RaMMPS, USAID for the Mozambique RaMMPS and institutional funding from NYU for the Malawi RaMMPS.  

The RaMMPS project entails collaboration between 13 universities, governmental and non-governmental organisations in 5 countries. Some of the coordinating functions are performed by LSHTM, and Ties Boerma chairs the RaMMPS Strategic Advisory Committee (SAC). 

Methods
Methods RAMMPS 2 columns
Methods RAMMPS 2 columns left paragraph
Paragraph

The RaMMPS project is implemented as a series of five country case studies that share a focus on mortality measurement via mobile phone surveys, but they differ in terms of the instruments used, the sampling approach and fieldwork implementation. Because mortality surveillance via mobile phone interviews have not been systematically used, the RaMMPS project includes a number of (nested) validation studies.

Survey instruments

Household Deaths

The purpose of the household deaths module is to provide a rapid assessment of the short-term fluctuations in the overall crude death rate, and the death rate in children aged 0-4 years. Collecting detailed data on household composition can be time consuming. Instead, a short form questionnaire is used to collect age-aggregated mortality data, i.e., the number of household members up to/above age five; number of deaths up to/above age 5 over a specified time period.

Full/Truncated Birth or Pregnancy Histories

Although the mortality risks from COVID-19 mortality in children is low, it is hypothesized that the indirect impact can be substantial. Full Birth Histories (FBH) are routinely used in DHS and MICS to collect data on U5M. Truncated birth or pregnancy histories (TBH), a shorter version of FBH, are used to collect data on U5M among children born in the last five or seven years only. In the RaMMPS studies, the TBH instrument from the questionnaire has been adapted for use from the Malaria Indicator Surveys.

Parental Survival Histories

Parental Survival Histories (PSH) are well-suited for monitoring COVID-19 mortality through RaMMPS as they require limited time to collect and produce mortality estimates for an age range where individuals are most at risk.

Sibling Survival Histories

Sibling Survival Histories (SSH) are routinely collected in the DHS and can be adapted or shortened for the purpose of the RaMMPS. Even though direct COVID-19 mortality is probably modest in this age range, the SSH offer an opportunity to elicit information on mortality among the pregnant and postpartum women.

COVID Vaccination

Most RaMMPS includes a few questions to assess COVID-19 vaccine coverage and vaccine hesitancy among those not yet vaccinated.

Validation studies

RaMMPS questionnaires are adapted from standard instruments for estimating mortality in surveys and censuses. As mobile phone surveys have not yet been used for measuring mortality, this approach requires validation. This entails:

1. a priori validation of RaMMPS survey instruments using other data sources (e.g. comparison of estimates using short form versus full survey instruments; evaluation of bias associated with mobile phone ownership). This work is led by Bruno Masquelier at UC Louvain.

2. ex post-facto validation of RaMMPS mortality estimates against external data sources. Validation of mortality data generated via RaMMPS can be done at the individual level (using individually-linked RaMMPS and reference datasets such as COMSA or HDSS) or at the aggregated level (using aggregated mortality estimates from other sources, including COMSA, HDSS and DHS).

3. Proxy Birth Histories (PBH) validation studies: PBHs are a tool for collecting birth histories through proxy respondents. More specifically, it involves asking women of reproductive age (WRA) to not only report their own FBH but also the birth histories of their reproductive aged sisters or confidants. PBH are appealing because it increases the effective sample size and thus has the potential to improve the cost-effectiveness of mobile phone interviews for estimating (excess) U5M. Further, PBH are collected irrespective of the survival status of the mother, and this could reduce bias in mortality estimates in settings where the survival probability of mothers and children are correlated (e.g., due to HIV or possibly COVID-19). However, the validity of PBH is poorly understood. PBH validation studies are nested in the Bangladesh and Malawi RaMMPS. PBH is administered in the validation samples and reports of births and deaths among sisters of reproductive age will be compared to the reference datasets through individual-level comparisons using an approach that was previously used for validating FBH. These validation studies are led by Dr. Li Liu (JHU).

Countries
Countries RAMMPS 2 columns
Countries RAMMPS 2 columns left paragraph
Paragraph

Each of the five RaMMPS country case studies share broader objectives and several methodological features but also vary in terms of their sampling strategy, the survey instruments used, nested validation studies and so forth. This heterogeneity is intentional to best leverage context-specific attributes of each RaMMPS and to better understand strengths and weaknesses of different mortality mobile phone surveys methods.

Bangladesh
IEDCR, ICDDRB, JiViTa, Johns Hopins University and NYU University, Abu Dhabi logos

RaMMPS Bangladesh uses mobile phone-based interview methods to validate and quantify reports of mortality and pregnancy loss in Bangladesh. The study leverages the strengths of multiple in-country institutions to implement this work, including the Institute of Epidemiology Disease Control and Research (IEDCR), a government institute that has conducted telephone interviews for maternal, child and reproductive health and noncommunicable diseases for the past five years; JiVitA, one of the largest population health and nutrition intervention research organizations in South Asia; and the International Center for Diarrheal Diseases, Bangladesh (icddr,b) who coordinate the Matlab HDSS, an open cohort initiated in 1966 covering 142 villages in a sub-district located 55 kms south-east of Dhaka. Johns Hopkins University (JHU) and NYU University, Abu Dhabi (NYU Abu Dhabi) offer technical support.

The JiVitA Project validates two mobile phone survey methodologies for mortality and pregnancy loss against known mortality and loss events that occurred during previous trials at the site:

  • Aim 1: Validate Rapid Mortality Mobile Phone Surveys (RaMMPS) at the JiVitA Project, using a sample of known, prospectively collected events of infant death, still birth and miscarriage as the basis for this validation. These events were collected prospectively as part of the mCARE-II randomized controlled trial. A subset of these participants (6,600) is being called for the RaMMPS validation. Women’s report of pregnancy history during these calls is being compared to previously collected pregnancy history and outcome status.
  • Aim 2: Validate a Proxy Pregnancy History reporting methodology that uses reporting by closest confidants to ascertain the (index) woman’s pregnancy history, in a subsample of the RaMMPS surveys conducted as part of Aim 1. The implementation validation of proxy pregnancy history methodology is being tested in 3,000 of these phone interviews. This involves asking women of reproductive age to not only report their own full pregnancy history, but also the pregnancy histories of their closest confidant of reproductive age to validate index pregnancies.

IEDCR leverages an existing call center and years of expertise at conducting phone-based disease surveillance to conduct a national level mortality survey using computer assisted telephone interviewing (CATI) systems. The study estimates excess mortality during the COVID-19 pandemic among 25,000 participants constituted through random digit dialing. A series of 5 mortality surveys with representative samples of phone users are being conducted. These surveys cover crude mortality rates, selected age-specific mortality rates as well as a few other health-related topics (e.g., COVID-19).

Work led by icddr,b builds on more than 50 years of demographic observation and record-keeping in Matlab upazilas, a largely rural area located 2 hours southeast of Dhaka. In the Matlab Health and Demographic Surveillance System (HDSS), the icddr,b team will validate data on mortality at adult ages generated via phone surveys. To do so, the team has interviewed more than 2,000 residents of the Matlab HDSS by mobile phone, asking questions primarily about parental and sibling survival, and the registration of recent deaths among these relatives. The reported demographic data has then been linked, at the individual level, with precise HDSS records. This study design thus allows an in-depth ascertainment of errors and biases in survey data collected by mobile phone. To ensure sufficient statistical power, the sample included in this validation study includes an oversample of HDSS residents who recently experienced a death among their relatives. It also includes a sub-sample of respondents who previously participated in an NIH-funded study of the measurement of adult mortality, thus allowing the comparison of different methods of survey data collection (i.e., in-person vs. mobile-based interviews).

The JiVitA Project validates two mobile phone survey methodologies for mortality and pregnancy loss against known mortality and loss events that occurred during previous trials at the site:

  • Aim 1: Validate Rapid Mortality Mobile Phone Surveys (RaMMPS) at the JiVitA Project, using a sample of known, prospectively collected events of infant death, still birth and miscarriage as the basis for this validation. These events were collected prospectively as part of the mCARE-II randomized controlled trial. A subset of these participants (6,600) is being called for the RaMMPS validation. Women’s report of pregnancy history during these calls is being compared to previously collected pregnancy history and outcome status.
  • Aim 2: Validate a Proxy Pregnancy History reporting methodology that uses reporting by closest confidants to ascertain the (index) woman’s pregnancy history, in a subsample of the RaMMPS surveys conducted as part of Aim 1. The implementation validation of proxy pregnancy history methodology is being tested in 3,000 of these phone interviews. This involves asking women of reproductive age to not only report their own full pregnancy history, but also the pregnancy histories of their closest confidant of reproductive age to validate index pregnancies.

Description of the RaMMPS software at JiVitA

The RaMMPS software at JiVitA is developed through the mjivita platform, which tests the flow of the calling system between index women and their closest confidants. The system collects information on recipients and systematically schedules 4 calls with the index women and their closest confidants. The system is programmed to manage various call scenarios and set up automatic rescheduling of calls if needed. For example, the system can store incomplete calls information in the database and color code different call statuses according to interview priority (see figure).

Screenshot of mjivita platform, showing that the system can prioritize interviews based on call statuses

 

Burkina Faso
ISSP, INSD, LSHTM and UCLouvain logos

The Burkina Faso RaMMPS is led by the Institut Supérieur des Sciences de la Population (ISSP) in collaboration with the Institut National de la Statistique et de la Démographie (INSD), the London School of Hygiene and Tropical Medicine (LSHTM) and the Université Catholique de Louvain (UCLouvain).

Democratic Republic of the Congo
University of Kinshasa and LSHTM logos

The DRC RaMMPS is conducted as a collaboration between the University of Kinshasa School of Public Health (UNIKIN) and LSHTM. UNIKIN also hosts the PMA study in the DRC. 

Study population

The study population for DRC RaMMPS is comprised of a targeted 15,000 adults from 18 to 65 years of age living in Kinshasa or Nord Kivu provinces.

Sampling

Initial samples of phone numbers were created using a short Interactive Voice Response (IVR) survey prior to the CATI interview. The intention was to filter out non-functional numbers and elicit basic demographic information. Soon after data collection began, the sample of phone numbers was made up of numbers that had been active in Kinshasa or Nord Kivu in the last 14 days. This approach was facilitated through collaboration with Feroxus, a company based in DRC.

Mozambique
INS and Johns Hopkins University logos

RaMMPS Mozambique aims to test the use of mobile phone interviews to collect data to measure childhood mortality, crude death rate, and selected other indicators at national level. The project is implemented by the Instituto Nacional de Saúde (INS) with technical assistance from Johns Hopkins Bloomberg School of Public Health (BSPH). In addition to RaMMPS, INS and BSPH collaborate on the Countrywide Mortality Surveillance for Action (COMSA), a sample registration system that collects continuous data on pregnancies, pregnancy outcomes for mortality monitoring, and mortality data that is collect through verbal and social autopsy (VA) to ascertain causes of death and the associated factors.

Study population

RaMMPS Mozambique targets women, aged 15-49, to collect data on child mortality, select maternal health coverage indicators, COVID-19 vaccine hesitancy, household deaths, COVID-19 VA among household deaths, and questions on phone ownership and decision-making behavior for seeking healthcare.

Implementation

RaMMPS Mozambique will implement two sets of data collection based on two sampling strategies:

  1. Leveraging the existing COMSA data collection platform with available telephone numbers;
  2. Random digit dialing (RDD), where a randomly generated list of telephone numbers is constructed based on Mozambique’s telephone network structure.

For each component, a total of 15,000 completed interviews is expected.

The RDD sampling will include an interactive voice response (IVR) survey to screen the phone numbers to see if they are active, and potential participant’s willingness to participate in the RAMMPS survey. Potential eligible respondents identified through the IVR will receive a follow-up computer assisted telephone interview (CATI). The Mozambique RAMMPS project is working with a third company web-based platform that specializes in IVR, and automated phone calls and tracking. 

Prior to initiating the main surveys, a pilot was completed in two provinces for the COMSA-based sample and in a small sample for the RDD. The pilot tested four versions of child mortality tools:

  1. Full pregnancy history (FPH);
  2. Truncated pregnancy history (TPH);
  3. Full birth history (FBH);
  4. Truncated birth history (TBH).

The pilot helped assess the feasibility of the different approaches, response rates, and potential cost to help chose the optimal child mortality tool for full scale up. Based on the pilot results, the FPH tool was selected for the main survey. A qualitative study is also planned to explore women’s challenges with mobile phone surveys and effects of gender and power relations on women’s ability to participate in mobile phone interviews.

Malawi
MEIRU, IPOR, LSHTM and NY University, Abu Dhabi logos

The Malawi RaMMPS is implemented by Malawi Epidemiology and Interventions Research Unit (MEIRU), and the Institute for Public Opinion Research (IPOR), with technical support from LSHTM and NYU. MEIRU hosts the Karonga Health and Demographic Surveillance System that will be used as a validation data resource.  

Study population

The study population for the national RaMMPS survey in Malawi will be made up of 20,000 men and women between the ages of 18 and 65 years.

The Karonga Health and Demographic Surveillance System in northern Malawi will serve as the platform for the RaMMPS validation study. Mortality estimates generated through RaMMPS will be compared with HDSS estimates among a sample of 2,500 residents.

Sampling

The sample of numbers given to the CATI survey team are narrowed down before being transferred. This is done through collaboration with an external company called Sample Solutions, who:

  1. will draw a simple random sample based on the numbering structure of the mobile phone providers, then will
  2. verify that the phone number is active using the Home Location Register (HLR), which contains data regarding authorized subscribers using a global system for mobile communication (GSM) core network. The sample constituted in this way is not disaggregated by any demographic characteristics or region. We have pre-defined sample quota to ensure that the sample is representative of different sub-populations.  
Outputs
Outputs RAMMPS 2 columns
Outputs RAMMPS 2 columns left paragraphs
Paragraph
Data dashboards
Presentations and publications

Publications

Presentations

News
News RAMMPS 2 columns
News RAMMPS 2 columns left paragraph
Paragraph

Call for papers

Collaborators in the RaMMPS project are preparing a special issue for Demographic Research on the measurement of adult mortality in settings where civil registration and vital statistics (CRVS) systems are insufficient. This collection aims to bring together articles describing innovations in questionnaire design, data collection and statistical models to measure mortality over age 15 from survey and census data.

The deadline for extended abstract submissions is 15 January 2023.

Find out more information about the special issue in the call for papers.