What Is the EGSF?

The Guatemalan Survey of Family Health, known as EGSF from its name in Spanish (Encuesta Guatemalteca de Salud Familiar), was designed to examine the way in which rural Guatemalan families and individuals cope with childhood illness and pregnancy, and the role of ethnicity, poverty, social support and health beliefs in this process. It is part of a larger study designed and carried out by Noreen Goldman (Princeton University) and Anne Pebley (RAND), in collaboration with the Nutritional Institute of Central America and Panama (INCAP) in Guatemala, with funding from the National Institute of Child Health and Human Development.

The EGSF sampling frame includes the vast majority of rural communities in four of Guatemala's 22 departments. In each of these four departments the survey provides a self- weighting sample of communities with fewer than 10,000 inhabitants. While the survey is not based on a nationally representative sample, the four departments included in the study were selected to encompass a wide range of social, cultural, economic and environmental conditions in rural Guatemala. Household interviews were conducted in 4,792 households and individual interviews with 2,872 women ages 18 to 35. In each of the 60 sampled communities, we also carried out a community survey in which three key informants and a sample of biomedical and nonbiomedical health care providers were interviewed. The EGSF fieldwork was carried out between May and October 1995. The fieldwork took place at the same time as the fieldwork for the 1995 Guatemalan Demographic and Health Survey (DHS) which was a multipurpose survey based on a national probability sample, but collected more limited information on maternal and child health. The conduct of the two surveys in the same time period provides researchers opportunities for comparisons between the two surveys.

Several features of the EGSF make it different from previous RAND FLS surveys and from many other surveys in developing countries. First, unlike most FLS surveys, the EGSF was not designed as a multipurpose survey, but rather to meet the needs of a specific research project. For this reason, the data collected in the survey are focused on pregnancy-related health for women and on the health of children under age five. Furthermore, the EGSF individual survey interviewed only women ages 18 to 35, rather than all women of reproductive age, because younger Guatemalan women are more likely to have recent pregnancies and young children. The narrower focus allowed the EGSF to collect considerably more detailed data about women's and children's health and health care than is available in most general purpose surveys. Thus, it is ideally suited for researchers who wish to analyze topics such as: children's illness patterns, the frequency and nature of morbidity during pregnancy and delivery, determinants of illness, social support during spells of illness, treatment patterns, and determinants of treatment.

Second, collection of EGSF morbidity information in the survey is based on an innovative calendar design that provides a day-by-day (in the case of children's illness) or a month-by-month (in the case of pregnancy) description of health problems that women and children experience and of treatment behavior. The calendars were constructed to accommodate the different ways in which women think about the process of experiencing health problems and receiving treatment and advice. They provide researchers with much more detailed data on the timing of morbidity and behavioral responses than are available from most surveys. Previous research in Guatemala and other poor countries shows that families often take a sequential approach to coping with illness, e.g., first visiting a less expensive curer, then trying a pharmacist, and later, if the illness continues, consulting a doctor. The EGSF calendars allow researchers to trace these sequences of illness and treatment.

Third, several features of the EGSF were designed to facilitate the estimation of multi-level models including both family and community effects. The EGSF sample is intentionally more highly clustered than those in most sample surveys: an average of 50 households were interviewed in each of the 60 selected communities. The EGSF individual questionnaire also includes questions pertaining to the respondent's last two live births since January, 1990 and her two youngest co-resident children born since January, 1990. The EGSF also interviewed all women aged 18 to 35 in each sampled household. However, because the great majority of households in rural Guatemala are nuclear family households, there are relatively few households with multiple individual respondents.

Finally, the EGSF was designed to elicit information in a manner which would approximate as closely as possible the way that rural Guatemalan respondents think about health, treatment, family relations and economic variables. Consistent with this objective, the EGSF combines open-ended survey methods with pre-coded responses to questions. The responses to most questions were precoded, but interviewers were trained to write in the response in an "other" category if the respondent's answer did not fit into any of the precoded categories. The advantage of this approach is two fold: (1) a considerably broader array of responses than could have been anticipated were given to many items in the survey, and (2) the original Spanish text (and English translations) are provided for many responses. In addition to the textual response, the "other category" responses have been coded into detailed categories in order to preserve the richness of detail in the text. The availability of the original Spanish text, the English translations, and the coded responses for these variables considerably enhances the analytic options for researchers, particularly researchers familiar with Central America.

EGSF Sample

The sample for the EGSF was drawn from rural areas of the following four departments:

Chimaltenango, Totonicapán, Suchitepequez and Jalapa. The sample was restricted to four departments because a national sample would have necessitated the use of more than 21 indigenous languages spoken in Guatemala. These departments were selected on the basis of social, economic, and environmental diversity, and ethnic composition: one primarily ladino (Jalapa), two predominantly indigenous (Chimaltenango and Totonicapán) and one of mixed ladino/indigenous population (Suchitepequez). The language groups included, K'iche and Kaqchikel, are two of the largest indigenous language groups in Guatemala.

The sampling plan was based on a target of interviewing approximately 3000 women ages 18 to 35, living in 60 rural communities within the four departments. In the first stage, a total of 15 communities in each of the four departments were chosen from a list of all communities containing between 100 and 1800 households (approximately 200 and 10,000 inhabitants). Communities of this size were stratified within each department based on the predominant language spoken (Spanish or one of two indigenous languages, K'iche or Kaqchikel). Suchitepequez and Chimaltenango included a Spanish and an indigenous language (Kaqchikel) stratum. All communities in Totonicapán were indigenous (K'iche) while almost all in Jalapa were Spanish-speaking; the very few indigenous language speaking communities in Jalapa were subsequently excluded from the sampling frame. Within each stratum, communities were selected at random, with probabilities proportional to size (number of households) so that the sample would be self-weighting within each department. However, the sample is not self-weighting across departments and sampling weights are provided to adjust for different relative population sizes of the four departments.

In the second stage, 100 households were randomly selected based on a thorough mapping of each community. For each selected household, an adult member of the household provided information on each household member and his/her characteristics and on residents who had left or joined the household in the past year. Survey teams interviewed all women in each household who were reported to be between 18 and 35 years old at the time of interview, using the individual survey questionnaire. In addition, anthropometric measurement was carried out on all sampled women and on all their children born subsequent to January 1990.

The community survey conducted in each sampled community began with an interview with three key informants. The key informants included the mayor or auxiliary mayor, one female community leader, and one other community member who was not part of the formal community leadership (such as a store owner). Key informants answered detailed questions about the community and listed all health care providers (both biomedical and traditional) and all health care facilities that are used by community residents. The lists of health care providers for all three key informants were combined into a consolidated list. Sampling of health care providers was based on this consolidated list and included: one private doctor, one midwife, and two other providers. Other providers included curers, bone setters, spiritists, and other non-biomedical health practitioners. In addition, the survey team interviewed the head of the local government health post or center which served the sampled community.

EGSF Questionnaires

The EGSF used a total of 7 questionnaires for different types of respondents. The questionnaires, questionnaire content, and sample sizes of respondents to each questionnaire are shown in Table 1. Questionnaires for the household survey included the household roster, the anthropometry questionnaire and the individual woman questionnaire. The community survey questionnaires included the key informant questionnaire, the questionnaire for Ministry of Health health posts and centers, the private doctors questionnaire and the questionnaire for midwives and other providers. Spanish, K'iche and Kaqchikel versions of the household roster and individual questionnaire were used during the fieldwork. All other questionnaires were produced only in Spanish.

Table 1

Summary of Sample Survey Data Collected, EGSF (1995)

Questionnaire Information Collected Number of Interviews
Household Roster Listing of household members, relation to head, age, education 4792
Anthropometry Height and weight of all children born since 1/90,height and weight of mother 2668 women
3270 children
Individual Women See Table 2. 2872
Key Informants Economic activities, wages in agriculture & industry, banking, services, transport, water, sanitation, important events, costs of products, migration, census of providers w/i 20 km. 181
Health Centers and Posts Types & training of employees, hours, languages, availability of lab and pharmacy, electricity, water & sanitation, types of patients, fees and payment, services provided and cost, referrals by type of illness, supplies & medicines available, earnings 48
Private Doctors and Private Clinics Similar to questionnaire for health centers and posts 31
Other Providers Type of provider, training, languages, time spent treating, facilities, electricity, water, type of patients, fees & payment, how often list of treatments/care given (for pregnant women and others), frequency with which list of problems presented, referrals, earnings 169

EGSF Individual Questionnaire

Table 2 gives a more detailed description of the individual questionnaire. Sections A through C collect basic social and demographic data about the household and respondents, including a full live birth history. Section D uses a monthly calendar to collect detailed information on morbidity and health care during the respondent's last two pregnancies, for live births occurring since January, 1990. Section E uses a daily calendar to collect information on acute respiratory illness (ARI) and diarrhea for the past two weeks for the two youngest co-resident children, born since January, 1990. Because contraceptive use is relatively uncommon among women in rural Guatemala, section F collects information only on ever use and current use of contraceptive methods. Sections G and H gather information on a woman's marital and relationship history and on social support. Section J collects information on health beliefs including several vignettes to elicit beliefs about the causes of illness. Section K collects information on the respondent's involvement in and contacts with community organizations and hierarchy. Section L obtains information on the economic status of the household. Economic status questions include a consumption module, information on the respondent's and her husband's work and other activities in the past two weeks, and types of economic activities for household members in the past year.

Table 2 Brief Description of the EGSF Individual Questionnaire

Section Information Collected
A. Household and Dwelling Characteristics of dwelling unit
Household possessions
Access to water and transportation
B. Background Data Ethnicity/language
Previous residence
C. Birth History Full live birth/child mortality history
Current pregnancy status
Fetal loss/recent stillbirth history
Providers/health care during current and stillbirth pregnancies
D. Prenatal Care/ Assistance at Delivery For 2 most recent live births since 1/90:
Calendar of problems/providers/persons seen/home remedies during pregnancy
Detailed data on providers during pregnancy
Problems/care at delivery and post-partum
Neonatal problems/care
Causes of death
E. Child Health For 2 youngest children born since 1/90 living in household:
2-week calendar of diarrhea and resp. sympt./providers/persons/home remedies
Detailed data on providers/persons seen/home remedies
Causes of illness
General health status
F. Contraceptive Use Ever and current use of methods
G. Marital History Marital status
Dates of first and most recent union
Education/literacy and ethnicity of partner
H. Social Support Contact with relatives (parents, in-laws, siblings, and siblings-in-law)
Assistance received from relatives
Decision making with partner
Assistance received from partner
J. Health Beliefs Beliefs about causes/treatment of illness
Actual and potential use of different providers/facilities
K. Community Structure Perceptions of economic status of community
Participation of respondent and family members in community
Activities during past 5 years
L. Economic Situation Health insurance for household members
Respondent and partner work history/earnings for past 2 weeks
Econ.activities - past year for respondent, partner and other household members
Home/land ownership
HH consump.of staples (past 7 days) and other expenses (past month and year)
M. Contact Information Information to assist contacting respondent in case of reinterview
N. Interviewer Notes Characteristics of dwelling
Presence of others during interview
Assessment of quality of responses/difficulties encountered