Survey of malaria indicators in Caprivi Regional Red Cross Society area, Namibia,using cell phone data entry —

Preliminary report, 7 February 2012

RAMP (Rapid Mobile Phone-based) Survey

NaemiHeita (1), Laimi Onesmus (1), HilmaKambundu (1), Polly Helmut (2), David Simataa (2), Michael Charles (3), Bong Duke (4), Jenny Cervinskas (5), Jason Peat (5), Mac Otten (6)

  1. Namibia Red Cross Society, Windhoek Namibia
  2. Namibia Red Cross Society, Caprivi Region, Namibia
  3. International Federation of the Red Cross, Johannesburg, South Africa
  4. Consultant, International Federation of the Red Cross, Canada
  5. Consultant, International Federation of the Red Cross, Nigeria
  6. International Federation of the Red Cross, Geneva, Switzerland
  7. Consultant, International Federation of the Red Cross, Atlanta, USA

Executive Summary

Background. The Namibia Red Cross sponsored a baseline survey for the second phase of a multi-year project among a population of 46,727 that lived in the CapriviRegion during 30 January to 3 February2012 to examine important malaria indicators using several innovative data collection methods.

Methods: The survey used probability-proportional-to-estimated-size (PPES) sampling of primary sampling units (PSUs) and PPES to select one segment from unequal-sized segments. Households were chosen from the segment using simple random sampling. The sample included 30 PSUs, 10 households per PSU, and 1283 persons. Cell phones were used by Namibia Red Cross volunteers to conduct the household interviews and enter survey data using EpiSurveyor software in real-time.

Results: The gap in the number of LLINs needed to achieve universal coverage was 11,816 (48% of sleeping spaces). Sincean estimated 4,824 ITNs were 36 month old and need to be replaced, the total ITNs needed to be distributed in 2012 is16,640. An estimated 44% (95% confidence interval [CI] 36-51%) of persons of all ages slept under an ITN during the night before the survey and 94% of nets were used the previous night. Of children <5 years old with fever in the two weeks before the survey, 29% received an ACT and 22% received an ACT within 24 hours. Seventy-two percent of households received IRS in the previous year.

Conclusion: The number of ITNs available to households needs to be doubled to achieve universal coverage (all persons sleeping under an ITN). Approximately one-quarter of households still need to be reached with IRS. The innovativesurvey methodologies provided valuable health and malaria data rapidly.

Background

The government of the Republic of Namibia (GRN) and partners are striving to reduce the number of cases and deaths from malaria by 50% by 2010 and by 75% by 2015 in line with Millennium Development Goals, World Health Assembly, Roll Back Malaria partners, and GRN goals. The Ministry of Health and Social Services (MOHSS) in Namibia is now stressing the importance of universal coverage of persons of all ages (100% of persons using an insecticide-treated bed net in endemic areas)as advocated by the World Health Organization (WHO) to achieve the disease-reduction goals. The two most important indicators of universal coverage with ITNs are: 1) the percentage of persons that had access to ITNs in the household (assuming that 1 ITN covers two persons), and 2) the percentage of all persons using ITNs the previous night. For treatment, all persons with malaria are supposed to receive an appropriate treatment within 24 hours, especially children <5 years old, the highest risk group for malaria-related mortality. In late 2009, WHO advocated parasite-based testing of all suspected malaria cases (including the use of rapid diagnostic tests—RDTs), even in high-burden African countries.

Malaria continues to be a major public health problem in Namibia. The disease was the leading cause of illness and death from 1999 to 2002 and still remains one of the top five diseases of public health concern in the country. Malaria is endemic in Caprivi, Kavango, Kunene, Ohangwena, Omusati, Oshana, Oshikoto, and part of Otjozondjupa and Omaheke regions, where 65 percent of the Namibian population live and are at risk of malaria[a]. The prevalence of malaria is highest between September to December. Preliminary MOHSS data shows that malaria cases reported from health facilities in the Caprivi Region declined from approximately 4000 cases in 2006 to 300 cases in 2010.

This survey was carried out in Caprivi region[b], in the northeast of Namibia. Divided into the Kongola, Linyanti, Sibanda, KatimaMulilo Urban and Rural, and Kabbe constituencies[c], the region covers a total area of 14.528 km and accounts for 1.8% of the total land area of Namibia. It shares borders with four countries, being Angola and Zambia in the north, Botswana in the south, and Zimbabwe to the east. The administrative centre of the region is KatimaMulilo, the only town in the region. Six settlement areas serve as local administrative centers. These are Bukoalo, Chinchimani, Linyanti, Mafuta, Ngoma, and Omega III. Eastern Caprivi is subject to seasonal flooding and the Kabbe constituency comprises the eastern flood plains of the region, and is subject to seasonal flooding. There is an annual flood season with numerous flood-prone villages. In 2011, the flooding was widespread, and more than 1000 people had been “relocated permanently” from 32 flood-prone villages as the region prepared to meet a Zambezi river swollen to record-size for this time of the year as a result of torrential rains upstream (ref: accessed March 3, 2011). In 2011, more than 20 settlement camps were established on higher ground, with entire villages being relocated to these camps. A number of organizations, including the MOHSS and the Namibia Red Cross, provided services in the camps.

In 2001, Caprivi Region had a population of 79,826 people (with a household population of 78,785 and 16,839 households), representing 4.4% of Namibia’s total (Population Census 2001). Twenty-eight percent of the population is urban and 72% rural. The average household size in Caprivi region is 4.7 persons compared to 5.1 persons nationally, and well below the average rural household size of 5.7. Female-headed households account for 49% of households, and 46% of households have orphans or children under the age of 18. The average number of children per women is 3.8 and 13% of the population is under five years. More than 95% of the populations speak one of several dialects of what is commonly lumped together as Caprivian. The main dialects are Sifwe, Subiya, Totela, and Yeyi. The 2006-07 Namibia Demographic and Health Survey showed that Caprivi was one of four regions in Namibia[d] that have the highest proportion of the population in the lowest wealth quintile of the wealth index and only a small proportion in the highest quintile.

Insecticide-treated bed nets (ITNs) are distributed in Caprivi through the MOHSSroutine health services. A pregnant woman is eligible to receive one free bednet during antenatal care services, and children under age five years that attend a health facility are also eligible to receive a free LLIN. Over the past few years the MOHSSand the Namibia Red Cross have distributed LLINs in the settlement camps established for those affected by the flooding of the Zambezi and the Chobe Rivers through mass distribution to targeted vulnerable groups (pregnant women, children under age five years, and elderly women aged 60 years and over). There has only been one region-wide mass campaign in Caprivi, with LLINs distributed in 2008 as part of the Zambezi River Basin Initiative. Since 2009, the Namibia Red Cross (NRC) has been distributing LLINs in the camps.

In addition to the MOHSS and the NRC, the NGO SMA (Social Marketing Associates) has been distributing LLINs in villages in some constituencies. SFH (Society for Family Health) sells LLINs through their social marketing program, and also distributes LLINs to the elderly when they receive their monthly government pension benefit.

This malaria indicator survey will serve as a baseline for the next phase of the Namibia Red Crossproject Communities Fighting Malariathat aims to improve the health and well-being and improve malaria control for persons in four of the constituencies in Caprivi Region where the Namibia Red Cross works--Kabbe, KatimaMulilo Rural, Kongola and Linyanti. In this project, the NRC, in collaboration with the village heads, will recruit 51 community volunteers who will serve as the project’s supervisors in the villages. Each supervisor will have the responsibility of supervising an average of about 10 community-based volunteers. A total of more than 500 volunteers will be trained by mid-2012 by these supervisors. In turn, each community-based RC volunteer is responsible for about 100 households.

In addition to serving as the baseline for the next phase of the CFM project, this survey is the endline for the first phase of the project. A previous survey was carried out by the NRC and IFRC in May 2011. That survey was also carried out using the RAMP survey methodology. It is expected that the final evaluation will be done in November 2012.

The sampling frame for the May 2011 RAMP malaria survey was not identical to the 2012 RAMPmalaria survey because in the May 2011 survey, the Caprivi Region was devastated by floods, with over 90% of the Kabbe constituency flooded. More than 9,000 people were affected by the floods, and a total of 20 temporary camps and two permanent resettlements camps (Choi in Linyanti and Izwe in Kongola) had been established (Namibia Red Cross, personal communications). Of the 30 clusters selected, 12 of them were in the settlement camps.

While the RAMP survey does not collect data on all the key indicators of the Communities Fighting Malaria project, it does collect data on many of the core indicators related to bed net ownership and usage, and to the treatment of fever in under-five year olds and the diagnosis of malaria. These include indicators that are aligned to the MOHSS Malaria Strategic Plan (2010-2016) and the MOHSS National Monitoring and Evaluation Plan (2010-2016). In addition, the survey collects data on indicators that measure universal coverage, a goal that the GRN is moving toward.

The results of this survey can be used to adjust the MFC project and the malaria prevention and control program, if necessary, and to advocate for needed changes to support the provision of LLINs, and the provision of services for treating fever in children under age five years, to communities within the sample constituencies.Other investigations will be done by the NRC in 2012 to explore in more depth aspects of the CFM project that are related to behavior change and communications.

The International Federation of the Red Cross (IFRC), with the support of DataDyne, WHO, and epidemiologists have been working for several years on a “management survey” concept that uses cell phones for data entry for surveys, primarily on the survey methodology and operations. In 2011, the management survey concept was successfully field tested in three African countries: Kenya in January, Namibia in May, and Nigeria in June. This concept includes use of cell phones and freely-availablesoftware to conduct health surveys rapidly, simply, at low cost, with minimal external technical assistance. The survey sampling method avoids the main potential bias of the Expanded Programme on Immunization (EPI) cluster survey method (initial selection of the first household using a random direction from the center of the PSU). This survey approach is now called the RAMP (Rapid Mobile Phone-based) survey. IFRC will soon publish a technical reference manual and a training manual that will be readily accessible for those interested in carrying out a RAMP survey.

This RAMP survey in Caprivi had the following innovations: 1) 30 clusters (like the EPI cluster survey)[1], 2) total of 300 households in the sample (10 households per cluster), 3) use of the Episurveyor web-based tool to collaboratively design model malaria questionnaires (including responses and skip patterns) that can be easily adapted to local surveys and translated into local languages, 4) use of cell phones to enter data during the interview, 5) daily upload of data to an internet-based database, 6) daily data cleaning of uploaded data, 7) daily review and feedback of data quality issues to interviewers and team supervisors 8) daily analysis of uploaded data, and 9) completion of preliminary results bulletin within 24 hours of the last interview.

Methods

The survey was conducted during 30 January to 3 February 2012 in the four constituencies in Caprivi Region where the Namibia Red Cross works--Kabbe, KatimaMulilo Rural, Kongola and Linyanti. The sampling frame was a list of communal lands, settlements, and mixed areas, from the 2008 sampling frame provided by the Namibian Bureau of Statistics, based on the latest census data (2001 national census) released by the GRN. The 2008 census estimation of population in the area was 46,727. Thirty primary sampling units (PSUs) were selected using probability proportional to estimated size. Using maps of each PSU that were obtained from the Namibia Central Bureau of Statistics, the selected PSUs in the settlements and communal landswere mapped and divided into 2-10 segmentsusing natural boundaries. Once a segment was selected by PPES, all households were listed or mapped and 10 households were chosen by simple random sampling. Additional households were chosen in case members of selected households could not be reached. Data was collected on all persons sleeping in the household (“sharing a common cooking pot”) the previous night. The design resulted in an equal probability survey.

Three questionnaires were developed online using the Episurveyor web-based questionnaire design tool ( person roster, and bed net roster. Questions were modeled after the Roll Back Malaria (RBM) Malaria Indicator Survey (2005).[2] Principal component analysis was used to create the wealth quintiles index for each household. Analysis was performed in STATA version 11 (College Station, Texas, USA),taking into account the design of the survey. “Access” to an ITN was defined as the population that could have been covered by ITNs present in the households at the time of the survey. The ratio of persons that could have been covered per ITN was calculated from the net roster data (1.88persons/ITN). The crude estimate of access was the total number of ITNs in the households times 1.88persons/ITN divided by total population. The ratio of persons to ITN (1.88) for this calculation was similar to theratio of 2.0 used by WHO in its World Malaria Report 2010 to calculate access to ITNs.[3]

We estimated the number of ITNs needed for universal access, the number of ITNs currently in the whole survey domain, and the gap to be filled. The number of ITNs needed for universal access was calculated by dividing the sampling frame population by the average number of persons sleeping under ITNs during the survey (1.88). The number of ITNs currently present was estimated by multiplying the number of ITNs found in households by the survey weight (36.42). The total number of LLINs that need to be distributed in 2012 to achieve universal coverage needs to include nets that need to be replaced due to age (36 months or old nets).

There are two new MERG indicators of access—by household and for populations. The indicator by households is the persons of households with sufficient nets to cover all inhabitants based on a ratio of 2 persons for one ITN. The indicators for populations is the number of persons with access to an ITN assuming that 2 persons can use each ITN.

The indicator about protection of households by ITNs or IRS assumed that any ITN (even just one) in the household protected all inhabitants. In households with many people, a single ITN may provide only partial protection from malaria. Therefore, this indicator may over-estimate protection.

There was significant "heaping" of responses of the age of nets in months at 12 and 36 months. We counted "12" to be in the 12-23 month category and "36" to be in the 36-47 month category.

The questionnaires (see annex for a copy, exported from the EpiSurveyor form to Word format) on the mobile phones were in English, and the majority of the interviews were carried out in Silosi, the local language in Caprivi.

Survey operations. The survey operations were led by the Namibia Red Cross. Training was provided for the 12 interviewers and six team supervisors during four days (23-27 January 2012). Survey field work took 5 days (30 January to 3 February 2012). Nokia-brand cell phones--Nokia 2730 ($80USD, no keyboard, no GPS) and Nokia C-02 ($105 USD, no keyboard, no GPS) --were used to run DataDyne’s freely-available cell-phone-based EpiSurveyor software ( and Survey data were immediately entered into the cell-phone database during the interview for the household questionnaire. Immediately before administering the person roster and the net roster questionnaires, a paper line list of persons that slept in the household the previous night was created by asking about each person who slept in the household the previous night so that the person line number/identifier was available during questions in the person roster, and for questions in the net roster about who slept under each net. Uploading of data on the cell-phone to the internet-based database using EpiSurveyor software required a 2G/GPRS cell-network connection. The field supervisors uploaded data at the end of the day upon their return to KatimaMulilo. Interviewers (and cell-phone data entry persons) were the Namibian Red Cross volunteers who serve as supervisors in the four constituencies included in the Communities Fighting Malaria project: Kabbe, Katima Rural, Kongola and Linyanti. The team supervisors were Namibian Red Cross supervisors that were involved in the Communities Fighting Malaria and other NRC projects.