Cover Image
close this bookCommunity-Based Longitudinal Nutrition and Health Studies: Classical Examples from Guatemala, Haiti and Mexico (INFDC, 1995, 184 p.)
close this folder5. A prospective study of community health and nutrition in rural Haiti from 1968 to 1993
View the document(introduction...)
View the documentIntroduction
View the documentBackground on Haiti
View the documentMaterials and methods
View the documentResults
View the documentDiscussion
View the documentAcknowledgements
View the documentReferences
View the documentBibliography
View the documentNotes


Evolution of Community Health in the Defined Population at the Hal Albert Schweitzer

In the HAS experience, communities can be mobilized to combat the most common killing and disabling diseases once they participate in the community "diagnosis" and understand appropriate action in concert with a health team. Community volunteers (about one per 100 families) helped the technical service outreach team to set up "rally posts" where quarterly preventive activities could take place. During the 1970s, this included immunization, training in ORT and family planning, growth monitoring/counseling, and nutrition intervention in the form of temporary targeted supplementary feeding for children with growth faltering. Because tuberculosis emerged as a major problem, the outreach teams also collected sputum specimens and made smears which were brought back to HAS for laboratory diagnosis.

Community health workers and professional consultants recommended that primary health care services be made geographically closer to the communities served. Therefore during the 1980s, HAS recruited, trained, and supervised medical auxiliaries who manned six outlying dispensaries and developed liaisons with other private voluntary organizations (PVOs) who also had dispensaries. These proved valuable in bringing curative aspects of primary health care closer to needy villagers, while at the same time reducing the case load at the hospital. The total number of outpatient consultations doubled; by 1992, HAS was seeing 6,000 outpatients per month, with an equal number being seen at outlying dispensaries. Thus, in the sense of providing earlier diagnosis and treatment, the dispensaries brought improvement. But dispensaries proved disappointing as an answer to preventive aspects of primary health care; medical auxiliaries were instead overwhelmed with curative care.

Immunization, growth monitoring coverage, and village-level nutrition intervention dropped during the 1980s because of the reasons described above and also because HAS redefined its RHVs as paid Agents de SantI> in keeping with government norms. Community volunteers were forgotten; Agents de SantI> were not required to be residents of the communities they serviced. The Agent de SantI> was supposed to cover a population of 2,000-3,000, or about 500 families, beginning with family enrollment and followed by home visits. They were supposed to set up "rally posts" for immunization and growth monitoring/counseling without community volunteers, but with the help of a technical backup team for immunizations. The new approach resulted in a decrease in preventive coverage in the 1980s.

The lesson learned was that the volunteer RHV, chosen by his or her own community to act as a liaison to health services, is essential. In HAS experience, women volunteers have been found to be an essential ingredient in the process, as are paid community health worker/supervisors and technical support outreach teams. A new model in the Plassac region is under way. Here the community health department mobilized women volunteers (1:15 families) to reach their own neighborhoods. Several of them cooperate to help the Agent de SantI> set up assembly posts in their own neighborhood for immunization, growth monitoring/counseling, and mini-antenatal and family planning consultations. The nearby dispensary auxiliary assists in some consultations. Prescribed medications are then provided by community pharmacies maintained by the community health committee who help finance the venture.

Neither outlying dispensaries nor paid community health workers (Agents de SantI>) work well without community mobilization and (in Haiti's case) women volunteers. There are young mothers in most rural communities who can and will give some time and effort to help create "rally posts" or "under-five" clinics at the neighborhood level. With such input, immunization and growth monitoring/promotion activities are far more likely to reach the majority of families. This experience is not unique. A World Vision project on Haiti's Ile de la Gonave, serving an extremely poor population of 80,000, recently completely immunized 80% of children under two through the use of community volunteers and the assistance of local churches (World Vision, 1993).

In rural Haiti, volunteers are also essential to vital event reporting, especially that related to pregnancy outcome. Newborns who survive only a few hours or days as well as other births and deaths are usually noted by local neighborhood women who, even though illiterate, can and will report them to paid health workers. Without this level of reporting, HAS had no data with which to calculate age-specific cause-specific death rates or birth rates for the 1 980s, which impairs community diagnosis. This necessitated a retrospective study to fill in the gap during the past 20 years.

The results discussed here are first and foremost those of the HAS study near Deschapelles, a unique study with 25-year follow-up results. Other studies will be brought into the Results section whenever pertinent.

Population Growth

In 1980, estimated overall population density exceeded 540 per square kilometer, with 390 per square kilometer of arable land (Institut Haitien de Statistiques, 1980). The HAS census tract of 23 villages did not double as expected over the intervening 22-year period; but an overall population growth of 47% was documented. Upland villages, however, more than doubled, increasing by 105% (see Table 1). Lowland villages are closer to public transportation; the communities are less traditional and more mobile. Here out-migration played a role, since the rate of natural increase would be expected to be greater than 2% per year. In the PISP project near Petit Goave, the natural rate of growth fluctuated slightly, but overall growth was around 2% per year. The population did not increase at this rate, due to net out-migration (Paisible and Berggren, 1984).

TABLE 1 Evolution of the Population of 23 Rural Villages Near Deschapelles Over the Period 1968-1990 Preliminary Data Based on Census Update by Community Health "Agents de Sant/B>






Growth rate (over 22 years)

Upland Villages







Lowland Villages













Over a four-year period (1968-1972), the HAS census tract documented a crude out-migration rate of 97.6/1,000 which was almost balanced by an immigration rate of 92.0/1,000, giving a net migration rate of -5.6/1,000 over a four-year period of observation. A year of food shortage (1970) greatly increased the propensity to migrate. In the subsequent year, over one-third (34%) of males aged 20 to 24 were involved in migration. The rate for females in the same age group was about half as much; many females in this age group were burdened with children and tended to remain behind (G Berggren et al., 1980).

Mobile Children, Dissolving Conjugal Unions, and Child Health

The HAS and PISP studies provide insights into the determinants and consequences of high mobility and its relationship to family building in rural Haiti. Dissolving Haitian conjugal unions are a cause of mobility from one household or one village to another; there is a propensity to displace children from their biologic mothers in the process. Even children under the age of two were separated from their biologic mothers. Some such children appeared to be unwanted when the mother moved in order to enter into a new union. Others were simply left behind with relatives, usually a grandparent, when the mother moved to gain financial advantage.

In the PISP study, 38% of women in the reproductive age group stated that their male consort had fathered children elsewhere, often in what appeared to be a pattern of serial placage. One woman proudly pointed to the four stepchildren she had raised, each from a different placage union of the one man who had been her male consort. In each case, she had taken over a child left behind when the biologic mother moved. The risk of child displacement and some of its consequences to the child were explored (W Berggren et al., 1981).

The PISP study used a multiple decrement life table to study the risk of a child under two being "orphaned" by being separated from his or her biologic mother. In each year of the study, cohorts of children were followed to calculate such risk. In the first year of the study, by the time a child survived to 24 months of age, he or she had a 20% risk of being separated from the biologic mother either from being orphaned or from being left behind by a move or a dissolving union. In the second year of the study, the risk dropped slightly for unexplained reasons. The reason for this separation was rarely death of the mother; usually the child was left behind with other relatives or another wife in the process of a dissolving union. Such children fared poorly; for children under one, 13.5% died within the first six months of the separation from the mother (Paisible and Berggren, 1984, p. 1-13).

An anthropologist studied the determinants and consequences of child relocation in the census tract of HAS. His preliminary analysis of residence patterns showed that 17.5% of children under 10 years of age did not live with one or both natural parents. He conducted an in-depth analysis in one village. Concentrating on children under 12, he found that of 165 such children born to the couples in that community, 26 had residency established outside the community at the time of the study. However, another group of 26 children had migrated into the community. He studied the circumstances of 35 children who had left their natal households and one or both parents. Twenty-four of the 35 had left disrupted conjugal unions, accompanying one or the other parent at the time of the breakup. Children accompanying a parent were usually of the same sex as the parent (Rawson and Berggren, 1973). Eleven relocated children who had not left a dissolving conjugal union had been moved by their parents into the homes of close relatives. Nine lived with grandparents and two with an aunt or uncle. More than two-thirds of the latter were from homes with four or more surviving children, leading the anthropologist to conclude that overcrowding in small Haitian homes is a factor in child relocation.

A significant finding of this study was the poor nutritional status of displaced or relocated children. Of such children under the age of six, more than one-half had already been screened into nutrition rehabilitation centers, indicating that the HAS nutrition surveillance program had discovered their growth faltering. Half of these malnourished tots were living with only one parent or with another relative. Further analysis revealed a statistically significant association between malnutrition and child-parent separation in this community. The finding was no surprise to the itinerant technical support team who helped weigh and measure the children monthly. Their questions to the caretakers of growth-faltering children often revealed that well-meaning parents had displaced a child into another household, usually with the idea of better providing for such a child. Most children not nutritionally deficient lived within a complete nuclear family (Rawson and Berggren, 1973). Such families were often smaller and the parents were less burdened.

In the HAS census tract, migration rates for girls 10 to 14 years old exceeded those for boys, with a doubling of the out-migration rate for girls (158/1,000) compared to boys (86/1,000) in 1971. The surprise was how often such children were migrating alone, separated from one or both biologic parents. Children born to polygamous common-law unions may be passed from one wife to another or left behind with relatives when a couple separates or moves on; this is especially true for young girls, who are seen as babysitters and as carriers of water and wood. Young girls recorded as migrating alone were usually found on investigation to have moved to "take care of" someone. The move was related to the unpaid labor they could provide. Several parents confessed that someone from the capital, not necessarily a relative, had come and collected the child, promising to take care of her but in fact inducting her into long hours of arduous labor. In at least three instances, such children returned home in poor physical condition; one died in Port-au-Prince, apparently from tuberculosis. Interviews with families revealed that they had falsely hoped their children would have the chance for schooling in Port-au-Prince in return for their labor.

Major Findings in Mortality Reduction

Infant mortality rates

Retrospective reproductive histories revealed an infant mortality rate of 126/1,000 live births in the HAS catchment area just before the initiation of the community health activities described above. Infant mortality rates in the community health impact areas of all the longitudinal studies appeared to have been cut by two-thirds in the first three to five years of the projects (see Table 2). Yet national estimates for the same time period show that the infant mortality rate for Haiti was 100 or more. In 1972, for example, when HAS documented an infant mortality rate of 34, the national estimate was 150 (Allman and May, 1979). The Institut Haitien de l'Enfance in 1987 estimated the national infant mortality rate at or slightly below 100/1,000 live births/year (Cayemittes and Chahnazarian, 1989).

Given the high infant mortality rates of Haiti (at or near 100 or more), the reductions in Table 2 are statistically significant (Reinke et al., 1993). Although there has been a slow decline in national infant mortality rates (Cayemittes and Chahnazarian, 1989), all the programs reviewed here achieved declines much more rapidly than expected. The most salient feature in accompanying deaths-by-cause studies was the disappearance of tetanus of the newborn. Deaths from diarrhea were also reduced (W Berggren et al., 1981; Paisible and Berggren, 1984).

Reduction of death rates due to tetanus of the newborn following immunization of all women is a finding in all the projects. Retrospective fertility histories in the Artibonite Valley revealed a neonatal tetanus fatality rate of 136.9/1,000 live births prior to immunization of mothers against tetanus during 1956-1962 (W Berggren et al., 1981). This rate was reduced to 78.9 with immunization of pregnant women and finally dropped to zero in the census tract by 1971, when all women had completed an immunization schedule of three doses of tetanus toxoid. By 1991, there were no reported cases of neonatal tetanus in the entire resident population of 180,000 served by the hospital. In the PISP project, prospective reporting of deaths by cause and the use of verbal autopsy methods documented neonatal tetanus death rates of 32.9/1,000 during the first six months of the project in Trou Chouchou where activities began. After an immunization campaign in which 85% of women had received three or more doses of tetanus toxoid, the death rate due to tetanus was reduced to 1/1,000 live births (Paisible and Berggren, 1984).

TABLE 2 Reduction in Infant Mortality Rates in Four Defined Rural Haitian Populations with Census-Based, Community-Oriented Health Services

Name of Study:







Petit Goave



Years Covered:





Infant Mortality Rates

Baseline or Year I





Year 3




Year 5





Year 7





Year 25


(follow-up study)

% reduction in first 5 - 7 years






aW Berggren et al., 1981, 1993; bPaisible and Berggren, 1984; cMenager and Tamari, 1990;
Berggren et al., 1993; dFarmer, 1992; Harvard Center for Population and Development Studies,

e From retrospective fertility histories using World Fertility Survey methods to determine vital rates in preceding years.

f These declines are statistically significant using the formula of Reinke (Johns Hopkins University School of Public Health) for longitudinal small area studies, as follows (Reinke, 1993).

E = PBDY (1)


P = Population size
B = Crude birth rate
D = Infant mortality rate
Y = Years of experience (data)
E = Expected (baseline) infant deaths
S = Observed % reduction in D needed for significance

In 1991-1992, however, Haiti experienced a measles epidemic which affected children under the age of one, and it is expected that 1992 data may reflect an upturn in infant mortality rates due to measles and to an increasing number of children born to mothers who may have AIDS.

One- to Four-Year Mortality Rates

Mortality rates in children one to four years old have been documented only by the HAS, PISP, and SCF/Maissade projects (see Table 3). HAS baseline data revealed a one- to four-year age-specific mortality rate of 14/ 1,000 before preventive, community-based interventions. This rate dropped to half of what it had been and remained at around 7/1,000/year or below during the next 15 years, according to the follow-up study. During this time, the national rate was estimated by the Haitian Institute of Statistics at 23 to 26/1,000 (W Berggren et al., 1981).

TABLE 3 One- to Four-Year-Old Age-Specific Death Rates in Rural Haiti Longitudinal Studies of Community-Based Primary Health Care

Name of Study:






Petit Goave


Years Covered:




Baseline or Year 1




(natl. 26)

(natl. 26)

Year 3




Year 5




(natl. 23)

Year 7




Year 25




(follow-up study)

% reduction in first 5-7 years





aW Berggren et al.,1981, 1993; bPaisible and Berggren, 1984; cMenager and Tamari, 1990; Harvard Center for Population and Development Studies, 1993.

In the PISP project, the one- to four-year-old mortality rate remained at 26/1,000/year in a nonintervened comparison (control) area, whereas in an area that had received intervention with health surveillance and child survival activities only, it dropped to 14/1,000/year.

The SCF/Maissade project has seen similar results, but has had an excess of deaths in the most recent year due to a measles epidemic (Harvard Center for Population and Development Studies, 1993).

What can explain the drop in death rates for the one- to four-year-old age group other than immunization, ORT, and earlier diagnosis and referral? It must be noted that nutrition interventions have been a salient feature of the projects described above. Over the time periods covered, as many as 6% of toddlers suffer Gomez third-degree malnutrition, many with the edema typical of kwashiorkor (that is, they weigh less than 60% of the international standard median weight for age). In the HAS, PISP, and SCF/Maissade projects, mothers assisted in rehabilitating their own children using locally available, inexpensive weaning foods. Underlying malnutrition has been shown to be a predictor of mortality in a number of studies and no doubt plays a role here (Scrimshaw and Hurtado, 1987).

It is important to note the relationship between being separated from one's biologic mother and the risk of death in rural Haitian children. The PISP project documented this risk (Paisible and Berggren, 1984), as shown in Table 4. Table 4 has strong implications for health education in community-based health projects in rural Haiti. Little is said to warn parents about separating children from their mothers, a not uncommon practice and one that is growing as the socioeconomic situation worsens.

Fertility Reduction

The HAS retrospective study showed that age-specific fertility rates have changed very little over the past 25 years, despite the continuing availability of intrauterine devices, pills, and condoms at the hospital and its outlying dispensaries (see Table 5). Reasons for resistance to family planning were studied in the 1970s by HAS investigators who noted that the "child-survival hypothesis" as a predecessor to family planning acceptance did not seem to hold for rural Haitians in the Artibonite. However, researchers noted that the crude birth rate in the HAS catchment area was only 36/1,000/year, lower than other developing countries during the 1970s. Brakes on fertility studied in the PISP and HAS projects included prolonged interpregnancy intervals due to breast-feeding (average of 18 months), frequent dissolution of conjugal unions, and a delayed age at first childbearing, documented at an average of 21 years in the PISP project area and 21.3 years in the HAS studies (Paisible and Berggren, 1984; Harvard School of Public Health, 1993). The age at onset of menses was also delayed, averaging about 15.5 years in the PISP project areas (Paisible and Berggren, 1984).

TABLE 4 Percentage of Rural Haitian Children Separating From Their Biologic Mothers in the First 24 Months of Life*

Risk of being separated from mother for all who survive to that month:


Month of Life

Trou Chouchou

Grand Goave


















Risk of Death,:

Percent dying within 6 months of separation




* From: Projet Inte Santt de Population, Prospective Longitudinal Studies in Three Defined Populations Near Petit Goave, Haiti, 1974-1978
Source: Multiple Decrement Life Table using person-months of follow-up for all children followed in three census tracts (Paisible and Berggren, 1984).

TABLE 5 Age-Specific Fertility Rates by Period (per 1,000) in the Artibonite Valley of Rural Haiti HAS Census Tracts, 1993 Retrospective Study

Time period

Age group
































In this study, the total fertility rate for 1977- 1991 varied by census tract from 4.5 in the villages nearest the HAS to 6.8 and 6.1 in the two more remote census tracts.

Mobility affected fertility rates. Reasons for in-migration given by adult females induded "contiguity" (21%), that is, the woman moved to stay with a migrating household. Only 17% of moves were made to form a union, and another 24% were made to join a relative, most often in the process of a dissolving union.

As women grow older, they are more likely to live without a male consort in residence, so that more than half of women over 40 are in this situation, often burdened with grandchildren. The situation of being "husbandless" is not new to them; on census day, in both studies, one-third of women who had ever entered into a union were found to be without a male consort.

Nutrition Results

HAS studies showed that rural children fall behind in growth at about the sixth month of life, suffering wasting and stunting to the degree that their 50th centile is almost identical with the US (Boston standard) third centile. Where nutrition intervention occurred in the form of demonstration-education for mothers who rehabilitated their own malnourished children, stunting was prevented in the lower 10th centile, and admissions for severe forms of malnutrition dropped (King et al., 1978).

The HAS studies showed that nutrition intervention apparently prevented stunting (G Berggren et al., 1985). Weight-for-height and height-for-age date were compared for preschool-age Haitian children enrolled in the HAS nutrition intervention program and children measured in the Haiti National Nutrition Survey in 1978. Cross sections of the longitudinal data of the intervention program corresponding to the season when the national survey was conducted (May to September) were chosen for the three years of available program data (1969,1970, and 1971). Significantly less stunting was found in the children in the 1970 and 1971 intervention groups than in the children covered by the national survey. Tests of trends also showed that the height status of children in the intervention group improved from 1969 to 1971. Wasting, or low weight status, was in general not significantly different in any of the comparisons. Nevertheless, the data were more favorable for children in the intervention program even in 1970, a year of food shortages.

However, the reasons the HAS success were not immediately accepted by the Haitian Bureau of Nutrition included the following:

1. Demonstration-education and rehabilitation units are best accomplished in the context of growth monitoring/counseling systems where mothers can weigh their children nearby and are allowed to hand carry and keep at home the Road-to-Health weight-age graph.

2. HAS nutrition rehabilitation centers were kept temporary and itinerant across a zone, thus avoiding the absenteeism occurring in those of the Bureau of Nutrition, where mothers had to come from farther and farther away as new "promotions" were identified.

3. HAS professionals used ongoing aggregate data from the village-level growth monitoring system to focus on those communities with highest rates of severe malnutrition.

Government Bureau of Nutrition units instead tended to become fixed "day care centers" with high rates of absenteeism among participating mothers. The "fixed center" catchment area grew ever wider to accommodate mothers with children at risk, with the result that distances for some mothers became burdensome, as did their required contribution (mothers were expected to help with wood, water, and fresh greens for the nutrition education sessions). Thus, Haiti's Bureau of Nutrition mothercraft centers appeared to be costly and ineffective, and funding for them ceased during the 1970s.

The PISP studies showed that mothers who had been through nutrition demonstration education and who had rehabilitated a child were able to prevent death significantly more often in the younger siblings of children who had been malnourished (G Berggren et al., 1984). These mothers were taught to feed children more frequently with a more calorie-dense diet, using locally available, inexpensive foods. Demonstration education lowered the mortality rate of malnourished children to 68% of the mortality rate experienced by those whose mothers had growth monitoring and counseling services but did not receive the benefits of demonstration education. Younger siblings of malnourished children were less apt to become malnourished and had significantly lower death rates than did the younger siblings of malnourished children whose mothers had not participated in demonstration education.

Emerging evidence from current information systems reveals the vulnerability of the poorest families to malnutrition, tuberculosis, and HIV. These facts lend new impetus to link economic assistance, improvement in economic conditions, and income-generating projects to the needed nutrition and health interventions (Farmer et al., 1991).

Tuberculosis Results

Tuberculosis was a major cause of death in the HAS, PISP, and Cange projects. All deaths suspected to be caused by tuberculosis were reviewed by physicians. At HAS, hospital records or autopsy confirmed the diagnosis; in the PISP project, a local tuberculosis treatment project had records on most cases. Of particular interest is the fact that prospective death reporting revealed that tuberculosis outstripped maternal causes as a major killer of women in their reproductive years. Those women had an overall mortality rate of about 80/10,000 in the first year of the project. In the PISP project, the women's age-specific, cause-specific death rate for tuberculosis was 17/10,000, whereas death from maternal causes accounted for 13/10,000 in Trou Chouchou, 4/10,000 in Grand Goave, and 6.7/10,000 in Meilleur.

The initial HAS census tract survey (N=7,369) showed that 2.3% of the population suffered active tuberculosis, as determined by repeated chest X-rays after skin testing. Half of these were pediatric cases in which sputum examination was difficult. A follow-up study was carried out in 1972 on all patients who were originally diagnosed and treated as having tuberculosis (255 cases). Of these, 183 were confirmed by X-ray and/or sputum examination, all cases followed and treated by qualified pediatricians and internists at HAS. The investigators concluded that 2.1 % of the originally surveyed population had tuberculosis (Casey et al., 1971). Findings compared well to a study carried out in the Jeremie area of Haiti during the same decade, in which 2.3% of that population were thought to have active tuberculosis.

Age-specific, cause-specific death rates from tuberculosis dropped continually over the five-year period when health surveillance activities were continuously followed and documented at HAS.

Farmer's prospective, population-based study at Cange showed that giving free tuberculosis medications plus financial aid to buy food resulted in significantly higher cure rates among rural Haitian (non-HIV-positive) tuberculosis patients. Case fatality and morbidity rates were significantly higher in a group that received free tuberculosis medications without financial aid.

It is doubtful that tuberculosis statistics have improved since the advent of AIDS in Haiti beginning in 1981. A current survey for HIV positivity in the Artibonite Valley is being carried out. Preliminary results reveal that as many as 6% of the rural population taking advantage of HAS services may be HIV positive. Estimates for HIV are lower for the PVS/Cange area, where rates are likely to be around 1% (Farmer et al., 1991).