Data collection tools for maternal and child health in humanitarian emergencies: a systematic review

Abstract Objective To describe tools used for the assessment of maternal and child health issues in humanitarian emergency settings. Methods We systematically searched MEDLINE, Web of Knowledge and POPLINE databases for studies published between January 2000 and June 2014. We also searched the websites of organizations active in humanitarian emergencies. We included studies reporting the development or use of data collection tools concerning the health of women and children in humanitarian emergencies. We used narrative synthesis to summarize the studies. Findings We identified 100 studies: 80 reported on conflict situations and 20 followed natural disasters. Most studies (76/100) focused on the health status of the affected population while 24 focused on the availability and coverage of health services. Of 17 different data collection tools identified, 14 focused on sexual and reproductive health, nine concerned maternal, newborn and child health and four were used to collect information on sexual or gender-based violence. Sixty-nine studies were done for monitoring and evaluation purposes, 18 for advocacy, seven for operational research and six for needs assessment. Conclusion Practical and effective means of data collection are needed to inform life-saving actions in humanitarian emergencies. There are a wide variety of tools available, not all of which have been used in the field. A simplified, standardized tool should be developed for assessment of health issues in the early stages of humanitarian emergencies. A cluster approach is recommended, in partnership with operational researchers and humanitarian agencies, coordinated by the World Health Organization.


Introduction
Humanitarian emergencies are natural disasters, man-made events or a combination of both that represent critical threats to the health, safety, security or wellbeing of a community. 1 Humanitarian emergencies resulting from conflict, natural disasters, famine or communicable disease outbreaks have important health implications. Currently, there are approximately 39 million people displaced by conflict or violence. 2 Every year, millions are displaced due to weather-related or geophysical disasters. 3 Women and children are generally the worst affected -representing over three-quarters of the estimated 80 million people in need of humanitarian assistance in 2014. 4,5 Moreover, many countries with high maternal, newborn and child mortality rates are affected by humanitarian emergencies.
Humanitarian emergencies are frequently characterized by the collapse of basic health services. For better decisionmaking, coordination and response in such emergencies, humanitarian actors need access to appropriate information. 4,6,7 Studies have reported that during humanitarian emergencies, there can be either a shortage or, conversely, an overload of information. Both situations impair provision of effective humanitarian assistance. 8 Sexual and reproductive health has historically been neglected in humanitarian emergency settings. 9 Health services provided for women and children vary depending on location, climate, culture, existing infrastructure, population health and type of humanitarian crisis. The types of response also vary, with multiple governments and humanitarian agencies involved. Efficient, easy to use, comprehensive data collection tools are needed to aid situation analysis, decision-making and coordination of responses to humanitarian crises. 10 We review tools for collection of data concerning the health of women and children in humanitarian emergencies. We identify which tools are available and where they have been used. For each study, we describe the setting and purpose of the study, the types of data collected and the tools used to collect the data.

Search strategy
We conducted a systematic review according to current guidelines. 11 We searched MEDLINE, Web of Knowledge and POPLINE databases for studies in English published between 1 January 2000 and 30 June 2014. Searches incorporated medical subject heading terms, keywords and free text using the following search terms: "reproductive health", "sexual", "maternal", "newborn", "child/child health service*", "pregnan*", "neonat*" under one search string and "disaster", "post conflict", "war", "humanitarian", "refugee", "internally displaced" under another string. The Boolean operator "OR" was used for the terms under each search string and "AND" was used to combine the two strings. The detailed search strategy is available from the authors.
Through a snowballing process, we identified organizations known for their work in humanitarian emergencies and searched the websites of these organizations -including CARE International, the Centers for Disease Control and Prevention, Harvard Humanitarian Initiative, the Inter-Agency Standing Committee, the International Federation of Red Cross and Red Crescent Societies (IFRC), the Joint United Nations

Inclusion and exclusion criteria
Studies were included if they reported the development or use of data collection tools concerning the health of women and children in a humanitarian emergency. We included studies, even when tools for data collection were not specified or the method was not described (Fig. 1).
Two aut hors i nd e p e nd e nt ly searched databases and websites. The titles and abstracts of identified studies were screened and excluded if not meeting the inclusion criteria. Full texts of remaining studies were assessed for eligibility. When it was not clear if a study should be included or not, two reviewers discussed the study and if consensus was not reached, a third reviewer was consulted. The reviewers summarized information on tools used, type of data collected and the purpose of the study. Data were classified into four categories, based upon the continuum of care: (i) sexual and reproductive health including sexual/gender-based violence and family planning; (ii) maternal and neonatal health; (iii) infant and child health; and (iv) sexually transmitted infections, including HIV/AIDS.
Studies that met the inclusion criteria were summarized using textual narrative synthesis. 10 First, we developed a commentary report on the type and characteristics of the included studies, context and findings using a standard matrix. The reviewers then looked for similarities and differences among studies to discuss and draw conclusion across the studies.

Results
We identified 2227 studies: 2109 publications from databases and 118 studies from websites. After removal of duplicates, the titles and abstract of 1593 studies were screened and of these, 225 studies were identified as eligible for full text review. Of these, 112 were not specific to humanitarian or emergency settings and 13 were not relevant (Fig. 1).
Of the 100 studies identified, 69 studies described the number of people affected. The population consisted of 677 568 individuals; 65 971 were identified as women and 57 427 children; 37 660 (57%) of children were younger than five years (Table 1, available at: http://www.who.int/bulletin/ volumes/93/9/14-148429). Studies ranged in sample size from seven (in case studies of survivors of sexual violence) 63 to 179 172 (in a rapid assessment of micronutrient deficiency following drought). 71 Eighty studies reported on conflict situations, while 20 studies reported on situations following a natural disaster (tsunami, hurricane or drought). Nineteen studies reported on the timing of data collection: three studies collected data within one week, 70,72,79 five within three months, 7,19,49,51,52 and 11 studies collected data six months to one year after the onset of the humanitarian emergency. 21,36,38,46,55,60,73,76,81,86,87 Data were collected from refugee populations in the recovery phase. Our review did not identify any studies that collected data during the disaster preparedness phase, which is defined by UNFPA as, "the period preceding a humanitarian crisis -use of early warning signals to avert crises or prepare response". 111 Seventy-six studies examined the health status of the population affected, while 24 examined the availability and coverage of health services, usually measured using the minimum initial service package. 60 A variety of indicators were collected with some studies using specific toolkits for field settings ( Table 2).
Data were collected for monitoring and evaluation purposes in 69 studies. In 18 studies, data were collected for the purpose of advocacy; seven studies were operational research and six studies described a needs assessment. No studies that we identified had the primary aim of collecting data to support a funding request.

Infant and child health
No description of specific tools used.
No description of specific tools used.   4 were used in one study each. The remaining 79 studies did not specify which tools had been used to collect the data.
Of the 17 toolkits identified (Table 3), 14 could be used to collect data on sexual and reproductive health, eight on maternal and newborn health, four on child health and seven on sexually transmitted infections and HIV. Some of the tools were designed to collect more than one category of data (e.g. Twine). Of the 14 tools used for data collection on sexual and reproductive health, four were specifically designed for gender-based violence. A further 13 studies also collected data on genderbased violence, but no data collection tool was identified.
Similarly, there was no specific tool to collect child health data, but four toolkits had questionnaires that included the collection of some data on child health data. Twine contains a specific section for child health data collection, including nutrition. 4 Refugee health: an approach to emergency situations 113 is designed to collect data on children for diseases under surveillance, nutritional status and common communicable diseases. The Sphere handbook 128 has rapid assessment tools to collect health service assessment data as well as sample surveillance reporting forms. These can be used to collect information on chil-dren younger than five years and provide outbreak alerts for this age group. These tools incorporate early warning and response network surveillance for early detection of epidemic-prone diseases in emergency settings. We did not identify specific tools for sexually transmitted infections and HIV, but relevant data are collected as part of seven of the more general sexual and reproductive health toolkits. 130

Discussion
Our review provides an overview of the data collection tools available as well as the published experience of the use of these tools. We advocate the use and harmonization of existing tools rather than the development of new tools. As we could not identify any studies reporting on data collection for disaster preparedness or disaster response, there is a need to adapt existing tools or develop new tools to facilitate data collection specifically for these phases. We excluded tools used primarily in non-humanitarian settings and may not have captured all available tools or data collected in humanitarian emergency settings.
Most of the tools specify which methods are needed to collect the required data, including both quantitative and qualitative methods in specific contexts. The methods used depend upon the purpose of data collection, the available resources and the nature of the information sought. Table 4 summarizes commonly reported methods to collect data during an emergency. 130 Of the 100 studies included in this review, only 19 described the data collection tools used and only six commented on their applicability in field settings. Authors may not be aware of the existence of a wide range of toolkits, or the importance of documenting their experiences.
To improve the response to humanitarian emergencies, target groups need to be identified and their specific needs understood. For sexual, reproductive, maternal, newborn and child health the underlying contexts which prevent or enable access to services also need to be considered. 130 The international humanitarian community continues to highlight the importance of documenting and addressing the problem of sexual and gender-based violence. 37 A central repository of data collected during a humanitarian emergency, where a core set of indicators is agreed on, would be useful. The repository would allow any user to submit or explore data to inform decision-making and enable comparisons between and across settings.
Only eight studies were conducted within the first six months of a humanitarian emergency. The majority of studies (69/100) and data collected were used to monitor and evaluate ongoing interventions. This may reflect the necessity of providing immediate life saving measures during the early stages of humanitarian emergencies. Rapid assessments are vital in the early stages of humanitarian emergencies. Information is required to highlight changing needs to inform appropriate provision of relief and urgent medical assistance. Most importantly, rapid assessment tools need to be simple to use. 131 It is encouraging to note that the tools developed so far seem to have used a cluster approach for data collection. Introduced in 2006 as part of the UN Humanitarian Response, a cluster is defined as: "a group of agencies that gather to work together towards common objectives within a particular set of emergency response". 132 The approach aims to improve the effectiveness of humanitarian assistance by improving predictability and timeliness of a response process through a coordinated effort. 111 The cluster approach can strengthen accountability among key actors and enhance the complementary nature of different organizations involved in providing humanitarian assistance. Although the health and nutrition clusters are critical for maternal, newborn and child health, the available tools consider other clusters as crosscutting areas including protection, water and sanitation, camp coordination and management. 132

Conclusion
There is a need to evaluate, standardize and harmonize existing data collection toolkits and to develop others that can be used in the response phase of humanitarian emergencies. Information is needed on the applicability of existing tools in relation to the types of populations and the emergency situations in which they are used. It would be useful to develop shortened versions of existing tools adapted specifically to use in the response phase, together with a more comprehensive version for the later phases of an emergency. Humanitarian assistance reports should include analyses of the lessons learnt when using data collection toolkits. This information can assist modification of existing tools and development of new tools. Whenever new toolkits are developed by interagency working groups, it is important to take the perspectives of field users into account. Wider dissemination of the availability of data collection tools among humanitarian workers can be achieved by educating staff at headquarters and country offices of humanitarian organizations, or by including the toolkits in disaster risk reduction training.
To plan and evaluate interventions and actions that will save lives in humanitarian emergencies, appropriate data are needed. To ensure that tools used to obtain such data are easy to use and comprehensive, it is essential that both individuals involved in field operations and in operations research continue to work together. New standardized tools should be developed and existing ones adapted based upon standards for data collection in emergencies with inputs from humanitarian agencies. 111  estudios (76/100) se centraban en el estado de la salud de la población afectada, mientras que 24 lo hacían en la disponibilidad y cobertura de los servicios de salud. De las 17 herramientas de recopilación de datos diferentes identificadas, 14 se centraban en la salud reproductiva y sexual, nueve trataban sobre salud maternal, neonatal e infantil y cuatro se utilizaban para recopilar información sobre violencia sexual o basada en el género. 69 estudios se habían realizado con fines de supervisión y evaluación, 18 para promoción, siete para investigaciones operacionales y seis para la evaluación de necesidades.
Maternal and child health in emergencies Thidar Pyone et al.