Peru’s Continuous Demographic and Health Survey

In Peru, the Continuous Demographic and Health Survey (DHS) provides annual data on the initiation of breastfeeding, exclusive breastfeeding under six months, solid foods introduction, and median breastfeeding duration at the national, urban-rural, and departmental (regional) levels of Peru. Critical to the success of the Peru program was establishing a permanent DHS unit within the national statistics agency (INEI) from the beginning of the project and incorporating the interests of stakeholders and helping them navigate the information produced from the continuous survey. The Peruvian government now relies on the continuous survey data, which they use to help decide health program budget allocations for Peru’s 24 regional departments. Breastfeeding outcomes data are broken down into departmental levels facilitating the data to be used to inform strategy in various national initiatives. Peru has one of the highest breastfeeding prevalence in Latin America. The most recently-published data results from the continuous DHS survey are the 2014 which show Peru has relatively good breastfeeding outcomes - 68.2% of infants under 6 months old are exclusively breastfed and the median duration of breastfeeding is 21.2 months.

Description and Context

In Peru, the Continuous Demographic and Health Survey (DHS), sponsored by USAID, provides annual data on the initiation of breastfeeding, exclusive breastfeeding under six months, solid foods introduction, and median
breastfeeding duration at the national, urban-rural, and departmental (regional) levels of Peru (1).

The DHS was initiated in Peru in the 1980s, however, the standard DHS was conducted typically at five-year intervals. Peru realized the need for a continuous survey; firstly, to facilitate the capacity of in-country teams to carry out the surveys–by undertaking the exercise annually, Peru was able to build up their own expertise through regular practice despite staff turnover. Secondly, there was a need to have data available more frequently than every five years in order to better monitor health and population programs. At the time, Peru was decentralizing their government functions and transitioning to results-based budgeting, which required timely health and population information. Thus, in 2004 Peru began its first cycle of the continuous DHS. In this format, DHS data is collected and reported on annually by a permanently maintained DHS office and field staff (1). In 2009, Peru began to report at the departmental level to better understand their health outcomes on a small scale to help target health interventions in the future.

Critical to the success of the Peru program was establishing a permanent DHS unit within the national statistics agency (INEI) from the beginning of the project. This reduced staff turnover as there were long-term job prospects, in contrast to hiring and training new staff every five years for the DHS. Additionally, permanent staff led to better data quality from the field and allowed for closer supervision of performance due to having smaller teams compared to a standard DHS team. (It should be noted that the INEI already had adequate government funding to support its staff and to carry regular DHS surveys). Over time, the external USAID staff trained the Peruvian staff, allowing for a gradual transfer of responsibilities. The survey is now wholly funded by the Peruvian government, with virtually no external technical assistance (1).

Another successful aspect of the program was incorporating the interests of stakeholders and helping them navigate the information produced from the continuous survey; special modules were added to the continuous DHS over the years to address stakeholder interests, for instance, the 2007 cycle and cycles since have included an ethnicity module, requested by the Ministry of the Woman and Vulnerable Populations and designed by GRADE, a Peruvian social research institution. Other additions include child labor and early childhood education indicators in 2009 (taken from the UNICEF Multiple Indicator Cluster Survey) requested by the Ministry of Education; chronic disease and trauma indicators in 2010 requested by the Ministry of Health; even a test for residual chlorine since 2008 requested by the Ministry of Health. In addition, the DHS project team developed more user-friendly reports and data files, ran seminars for policy makers, and data analysis workshops to help stakeholders read and interpret the results. Given the amount of data continuous surveying generates, it is key that countries have systems in place to effectively disseminate and utilize the information (1).

One of the challenges Peru faced were the resources available to foster a greater understanding and use of the data at a regional level and to improve the level of analysis undertaken. Also, using information from one or two cycles of data collection (in Peru, the cycles are semesterly) to understand annual trends had some limitations and they found that they needed to pool many cycles of data to get reliable information, usually 2 to 3 years worth. This was especially the case at the sub-national level and with certain health indicators that have a slow pace of change. In response to this, in 2008 they expanded the sample size from 6,300 to 23,000 households and then increased to 29,000 in subsequent cycles. In 2015 the number jumped to 35,000 households, about six times the original annual sample size. The demand for solid annual data led to increased sample sizes and, in turn, increased costs (1).

The Peruvian government now relies on the continuous survey data, which they use to help decide health program budget allocations for Peru’s 24 regional departments (1). The data is also used to inform strategy in various national initiatives, one of which is the National Strategy for Development and Social Inclusion by the Peruvian Ministry of Development and Social Inclusion (4). Using data from the continuous DHS (ENDES in Spanish), they have developed goals and strategies to improve infant nutrition, for instance, giving local and regional government as well as social programs such as Cuna Más the responsibility to train parents and care givers in exclusive breastfeeding and the proper introduction of complementary foods (4).

The most recently-published data results from the continuous DHS survey are the 2014 results, which show Peru has relatively good breastfeeding outcomes but clearly still has room for improvement (2); Nationally, 55.1% of infants are
breastfed within the first hour of birth, 93% within the first day (2). 68.2% of infants under 6 months old are exclusively breastfed (2). Furthermore, the median duration of breastfeeding is 21.2 months and 83.2% of infants 6-9 months
of age receive complementary foods (2). Outcomes are also broken down into departmental levels.

Main Components

The following components delineate some of the design and implementation processes that create Peru’s continuous DHS (1):

Sample:

The sample was chosen from previous clusters in the Peru 2000 DHS survey in order to reduce sampling variance across cycles and to be able to monitor trends stemming back to 2000. The continuous DHS took 10 clusters in each Peruvian department (24) except for Lima in which more clusters were taken. 20 households were surveyed per cluster. Total, each cycle of the Peru CS was to cover about 6,000 households annually. The CS design also adjusted for population growth and migration, nonresponse, and any variability in the sub-samples. In order to reach sub-national levels even further, beginning with the 2009 cycle the CS sample was selected annually in two stages, two semesters of the year.

The first sampling stage consisted of the typical departmental clusters, and the second consisted of dwellings within each of the selected clusters. The domains of the sample were the country’s 24 departments, and there were four levels of urban-rural stratification: large cities, small cities and towns, semi-urban areas, rural areas. In addition, clusters in large cities were stratified by wealth using census data.

Survey Contents:

Core indicators included in the sample survey population structure (age, household structure), education, fertility, contraception, fertility desires, mortality, maternal health, child health, nutrition (which includes breastfeeding indicators), women’s status, domestic violence, and HIV/AIDS. Special modules were added to address stakeholder interests, for instance, the 2007 cycle and cycles since have included an ethnicity module, requested by the Ministry of the Woman and Vulnerable Populations and designed by GRADE, a Peruvian social research institution. Other additions include child labor and early childhood education indicators in 2009 (taken from the UNICEF Multiple Indicator Cluster Survey) requested by the Ministry of Education; chronic disease and trauma indicators in 2010 requested by the Ministry of Health; even a test for residual chlorine since 2008 requested by the Ministry of Health.

Staffing:

27 teams of two interviewers and one supervisor each make up the fieldwork staff of the Peru continuous DHS. One biomarker technician per team is hired for the time of fieldwork (performs anemia testing ie). Headquarters staff consists of the following members working full time:

• Survey director
• Population and health specialist
• Statistical technician
• Data entry supervisor
• Data entry specialist
• Administrative specialist
• Field work assistant
• Driver

Regional supervision was provided by the directors of the departmental offices of INEI (Peru’s National Statistics Institute) for the four departments where the regional field staff were based. These four national supervisors visit the various teams during fieldwork, observe and enforce proper procedures, and answer any questions that may have arisen.

When the continuous DHS was initiated in 2004, 60% of the staff employed had worked in the 2000 DHS, providing strong experience. All supervisors and interviewers were college graduates from the social sciences and the biomarker technicians were public health nurses in 2004. Since 2009, all continuous survey staff have been INEI employees since 2009 rather than temporary project hires, with all the benefits (health insurance, retirement, sick leave) that are part of government employment.

Interviewing:

All selected households are visited to obtain in-person interviews, and call backs are made if the household individuals cannot be interviewed on the first visit. The team stays in the cluster between four and six days.

Data Processing:

Since 2006, the Peru continuous DHS has used computer-assisted personal interviewing (CAPI). CAPI alerts interviewers to potential errors including out-of range responses, errors in following skip instructions, and inconsistencies between responses, for example, in dates recorded in the birth history. It allows many errors to be corrected at the point of interview instead of at the headquarters during data entry, thus, it improves the quality of the information collected and reduces time costs associated with correcting data.

Data Dissemination:

Four types of reports are being produced annually for the most recent Peru continuous DHS cycles:

• A preliminary report at the national level using the representative first-half sample data. It is produced in July in order to inform the Peruvian Congress’ budget discussions and for use in the president’s Peruvian Independence Day speech (late July).
• A second preliminary report based on the full sample. It is produced in January from the previous year’s data collection (which ended in November or early December) for the evidenced-based budgeting section of the Ministry of Economics to assist in the setting of state-level health program budgets.
• A main survey report. It is presented at a national seminar at the beginning of May.

State level reports are produced for each of the 24 states later in the year. INEI has available the reports of all the Peru CS cycles as well as earlier DHS surveys through its website.

Evidence of Implementation Strategy

The report on which much of this case study investigates, was written by USAID, who clearly indicate the Peru continuous DHS experience was very successful. They claim that most of the initial objectives for the continuous DHS have been achieved in Peru: the survey has been institutionalized, with the Government of Peru funding all survey operations, and a DHS unit was established within the INEI, the government’s national statistical agency. Currently the unit’s field staff are permanent INEI employees, and the unit is operating without external technical support. The USAID report also states that staff retention and morale are high within the unit. Finally, both the Peruvian legislative and executive branches rely upon the data generated by the CS, which are integral to the process of determining the health program budgets for Peru’s 24 departments.

Evidence of the continuation of the program is provided by the INEI. The INEI website has continuous DHS reports and other documents such as interview documents for that cycle (3). The most recently-published results from a continuous DHS survey are the 2014 results, which show Peru has relatively good breastfeeding outcomes but clearly still has room for improvement (2); Nationally, 55.1% of infants are breastfed within the first hour of birth, 93% within the first day (2). 68.2% of infants under 6 months old are exclusively breastfed (2). Furthermore, the median duration of breastfeeding is 21.2 months and 83.2% of infants 6-9 months of age receive complementary foods (2). Outcomes are also broken down into departmental levels.

Cost and Cost-Effectiveness

The USAID report cites the cost-effective nature of the continuous DHS–a permanent survey was seen as more cost-effective than funding two full-scale DHS surveys. Even if the five-year cost of the continuous survey was about the same as the two DHS surveys or one DHS and one interim survey, a continuous DHS survey would provide annual national estimates and subnational estimates on a two to three year basis versus every five for the standard DHS survey. The continuous DHS survey also was seen as easier to fund as a line in the national annual budget.
In addition, permanent staff and infrastructure would be in place instead of re-establishing and re-hiring staff every DHS survey.

The INEI has absorbed the role of the continuous DHS, and is funded directly by the Peruvian government. The budget in 2016 was $13.1 million USD (5).

Perceptions and Experiences of Interested People

USAID have indicated that they consider the Peruvian continuous DHS experience to be a success. Most of the initial objectives were achieved, the unit’s field staff are permanent INEI employees, and the unit is operating without external technical support. Staff retention and morale are high within the unit. Finally, both the Peruvian legislative and executive branches rely upon the data generated by the CS, which are integral to the process of determining the
health program budgets for Peru’s 24 departments.

They also report challenges faced, such as resources available to foster greater understanding and use of the data at a regional level and resources to improve the level of analysis undertaken. Also, using information from one or two cycles of data collection (in Peru, the cycles are six monthly) to understand annual trends had some limitations and they found that they needed to pool many cycles of data to get reliable information, usually 2 to 3 years worth. This was especially the case at the sub-national level and with certain health indicators that have a slow pace of change. In response to this, in 2008 they expanded the sample size from 6,300 to 23,000 households and then increased to 29,000 in subsequent cycles. In 2015 the number jumped to 35,000 households, about six times the original annual sample size. The demand for solid annual data, led to increased sample sizes and, in turn, increased costs.

Benefits and Potential Damages and Risks

• Adequate funds for implementing the national survey are necessary or there is a risk of not obtaining truly accurate data due to budget or technical constraints. The design of a continuous survey program should ensure all segments/regions of the population are reached and estimate proper funding.
• There is a risk that if an external donor, such as USAID initiates the survey, it will not be continued by a national agency. It is crucial that, if external stakeholders are involved, they integrate with an in-country organization to develop a sustainable, long-term national survey in order to monitor trends over time.

Scaling Up Considerations

• It is key that the implementing organization of the national survey have a strong statistical staff with considerable expertise in general survey implementation and, specifically, in DHS implementation if this is the survey to be adopted nationally. The organization must also be adequately funded by the government. INEI in Peru was an excellent implementing organization; they had adequate government funding to support its staff and to carry out its regular statistical operations.
• It is important that the national survey take into account and adjust for population growth and migration, nonresponse, and any variability in the sub-samples in order to be able to reduce sampling variance and to be able to accurately monitor trends throughout survey cycles. Peru ensured this was done as they implemented a post-stratification scheme and also calculated these adjustment factors for each cycle.
• The ability to add new indicators to the national survey based on stakeholder interests and country context is extremely important in order to include new, relevant information for policy makers.
• It is crucial that the survey questions are appropriately formulated/written in order to gather accurate answers. National surveys should be designed on previously health surveys. The DHS is a well-established, proven survey that Peru adopted with the help of USAID.

Barriers to Implement

• Human and financial capacity and infrastructure is a barrier to implementing a national survey, which requires extensive resources to provide an accurate summary of the country’s health and population.
• Difficult field conditions may prevent field staff from obtaining interviews. For instance, the USAID report cites that the strong seasonality of malaria in some African countries may require significant adjustments to the length of fieldwork and the number of survey teams each year if malaria testing (one of the biomarkers taken as part of the national survey) is to be done during the high period (1). This will burden the field staff significantly.
• Uploading the continuous health surveys in a timely manner for public viewing is a barrier; while the Peru international statistics agency has survey and methodology documents from the most recent continuous DHS surveys available, the latest full report uploaded onto its official site was from 2014. This impedes the circulation of current breastfeeding data and can lead to misconceptions about the current national breastfeeding situation.

Equity Considerations

• Human and financial capacity and infrastructure is a barrier to implementing a national survey, which requires extensive resources to provide an accurate summary of the country’s health and population.
• Difficult field conditions may prevent field staff from obtaining interviews. For instance, the USAID report cites that the strong seasonality of malaria in some African countries may require significant adjustments to the length of fieldwork and the number of survey teams each year if malaria testing (one of the biomarkers taken as part of the national survey) is to be done during the high period (1). This will burden the field staff significantly.
• Uploading the continuous health surveys in a timely manner for public viewing is a barrier; while the Peru international statistics agency has survey and methodology documents from the most recent continuous DHS surveys available, the latest full report uploaded onto its official site was from 2014. This impedes the circulation of current breastfeeding data and can lead to misconceptions about the current national breastfeeding situation.

References:

1. Shea O., Way, Ann. (2014). The Peru Continuous DHS Experience. DHS Occasional Papers No. 8. Rockville, Maryland, USA.

2. Instituto Nacional de Estadistica de Informacion (2015). National Demographic and Health Survey 2014.

3. Instituto Nacional de Estadistica de Informacion. Bases de Datos.

4. Ministry of Development and Social Inclusion. (2014). National Strategy for Development and Social Inclusion.

5. El Peruano. (2016). Authorize Transfer of Items in the Public Sector Budget for Fiscal Year 2016 in favor of the National Statistics and Information Technology Institute and National Institute of Radio and Television of Peru – IRTP.

Could it work?

This is an example where a country built up its infrastructure and capacity to implement a continuous survey within the existing setup, i.e., the national statistics agency. The survey was then adapted over time to fulfill the needs of multiple stakeholders, demonstrating the flexibility that can and should be built into a continuous survey mechanism. This intervention is flexible to be adapted to other countries.

Will it work?

USAID evaluation of the continuous survey found it to be more cost-effective than running separate full-scale DHS surveys, especially where a permanent team is in place, rather than the expense of rehiring every 5 years. It was also found to be easier to have a line in the national budget for surveys when needed annually. Potential field difficulties (weather, disease) can be overcome by planning survey cycles accordingly.

Is it worth it?

The risks involved are underfunding the effort but as the continuous survey was found to be more cost effective than running separate surveys every five years, perhaps the financial burden would not prove to be a barrier in the middle to long term. This case study evidences that this type of data collection is certainly worth it and the potential to gather information on multiple issues would make it an attractive option.