Contents - Previous - Next

This is the old United Nations University website. Visit the new site at http://unu.edu

Coverage rates

Coverage is defined as the rate of participants per target sample. The target samples for the different age groups and study areas were defined prior to the field work. Participants in the follow-up study were defined as subjects for whom data were available for at least one study area.

Table 3 shows the distribution of the 2,169 subjects in the follow-up target sample in the supplemented villages and the 929 in the comparison villages. Overall coverage was 71.7%, ranging from 65.9% to 76.1% in the different villages. The rates were slightly greater for the supplemented villages (72.6%) than for the comparison villages (69.5%). They did not differ significantly between the atole and the fresco villages.

TABLE 3. Overall coverage of the follow-up cohort

Coverage rates were greater for females (74.5%) than for males (68.9%); this pattern was similar in the supplemented villages (females, 799/1,060 = 75.4%; males, 775/1,109 = 69.9%) and the comparison villages (females, 343/473 = 72.5%; males, 303/ 456 = 66.4%).

The coverage rates for migrants differed between the supplemented and the control villages (table 4). Among members of the follow-up cohort living in the villages at the time of the follow-up (non-migrant subjects), the rates were about 10% greater in the supplemented (88.6%) than in the comparison villages (78.0%)—probably as a result of the good rapport built by INCAP over the nine years of the longitudinal study. But the rate of participation for migrants was slightly less for the supplemented villages (40.7%) than the comparison villages (44.7%). This may be due to differences in how the target sample was defined for the two types of villages: The target sample of migrants from the comparison villages was identified using information from the 1987 census. Therefore, only adolescent migrants whose families were still living in the villages at the time of the follow-up census were selected. In contrast, the target sample for the supplemented villages was identified on the basis of records from the longitudinal study and thus included some whose entire families had migrated before the beginning of the follow-up study. Some of these families were located through information provided by neighbours and relatives; in the absence of parents or close relatives still living in the village, however, these target subjects were much more difficult to trace. Therefore, the coverage rates for migrants from the supplemented and the control villages are not equivalent.

TABLE 4. Coverage of the follow-up cohort by migration status Migrants

T = target sample; P = participants; % = coverage.

The coverage rates for migrants were much lower overall than those for non-migrants both because of the difficulty of locating migrants and because the data collection was restricted to three cities. The target samples in table 4 include all migrants, regardless of their location at the time of the follow-up study. However, as already mentioned, the data collection was restricted to those who were known to have migrated to Guatemala City or one of the two towns nearest to the study villages, because of resource restrictions and the fact that information available at the beginning of the study indicated that about 64% of the migrants for whom locations were known lived in one of these three locales. Coverage for migrants to these three places was 62%; for all migrants, it was 42%.

The coverage rates for females were greater than for males among both migrants and non-migrants. For migrants, the rates were 45.6% and 36.9% for males and females respectively; the pattern was similar in both the supplemented villages (females, 176/394 = 44.7%; males, 120/333 = 36.0%) and the comparison villages (females, 651134 = 48.5%; males, 40/101 = 39.6%). Among non-migrants the coverage was 89.7% for females and 81.2% for males; again the rates were greater for females in both the supplemented villages (females, 623/666 = 93.5%; males, 6551776 = 84.4%) and the comparison villages (females, 278/339 = 82.0%; males, 263/255 = 74.1%).

The subjects were classified into four birth cohorts with different ages of exposure to supplementation. Cohort I was made up of children born from March 1974 on; these children were partially exposed to supplementary feeding during their first three years of life, considered the most critical period in terms of the potential effects of supplementary feeding. Cohort II, the group born between March 1969 and February 1974, was fully exposed to supplementation during the critical period from birth to 3 years of age and partially exposed from 4 to 7 years of age. Cohort III, born between January 1966 and February 1969, was partially exposed to supplementation from 1 to 3 years of age and fully exposed between the less critical ages of 4 and 7 years. Finally, cohort IV, born before 1966, was partially exposed to supplementation from 4 to 7 years of age. In general, cohort I has the highest coverage rates, followed by cohorts II, III, and IV (table 5). Younger subjects may have had more time to participate in the various tests and interviews than older subjects.

TABLE 5. Percentage coverage of the follow-up cohort by birth cohorts and sex

See text for definitions of birth cohorts I-IV.

Table 6 presents the coverage rates for the different study domains by village type. The rates were around 70% for most domains. The rates in the comparison villages generally were slightly lower than in the supplemented villages but followed the same patterns. For the work-capacity test, more subjects than originally planned were examined in the comparison villages.

TABLE 6. Coverage of the follow-up cohort by study domain and village type

T = target sample; P = participants; % = coverage.
a. More subjects were examined than originally planned.

The low coverage for blood collection deserves comment. From the early stages of the study, some members of the field team suspected that many subjects who had refused to participate in the study did so as a result of anxiety regarding the procedure. This was substantiated by a team of supervisors with substantial experience in field work who interviewed subjects who had refused to participate and their families. Some subjects felt that the very small amount of blood collected (5 ml) was very large relative to the total blood volume in an adult. In one village, a rumor was circulated that the blood was being sold for a profit. A decision was made to make it clear at the time subjects were invited to participate in the study that blood collection was not essential for participation in the rest of the tests, measurements, and interviews. In addition, the subjects were informed that the blood samples were also used for the diagnosis of anaemia and that treatment would be provided when it was needed. After those messages began to be communicated at the time of the invitation, the number of refusals to participate declined, although refusals to provide a blood sample rose.

Other strategies were also used to increase the coverage rates. For example, the coverage of males was low in one of the villages. A large number of adolescents refused to participate even though they had no apparent time constraints. It was common to see individuals who had refused to participate standing in a park across the street from the centre where the tests, measurements, and interviews took place, teasing those who had agreed to participate. The staff realized that a large number of the refusers played soccer, and so a soccer championship was organized in the village by INCAP, including all teams in the village, and this had a positive effect in raising the coverage rates.

The coverage rate for the anthropometric measurements of the parents of the follow-up cohort was 82.4%, with no difference between the supplemented (809/979 = 82.6%) and the comparison villages (386/ 472 = 81.8%). Similar rates were obtained for the life histories of the mothers (82.7% overall; supplemented, 452/543 = 83.2%; comparison, 207/253 = 81.8%). In contrast, the coverage of the income and wealth questionnaire given to the heads of households was lower (62.0%) but was also similar in the supplemented (62.5%) and the comparison villages (60.7%); these low rates were due in part to the long time required to obtain the information and the fact that most of the heads of households worked in agriculture and were away from home most of the day.

Potential bias due to incomplete coverage

Table 7 presents descriptive statistics of key variables collected during the longitudinal study, when the subjects were between the ages of 0 and 36 months. As noted earlier, this information is available only for the supplemented villages. For all the villages combined, the participants in the follow-up study had somewhat greater birth weights, greater supplement intakes, and lower percentages of time with diarrhoea than non-participants. For all the other variables they were similar. The greater supplement intakes among the participants indicate that the subjects in the follow-up study had higher rates of participation in the supplementary feeding programme during the longitudinal study. The heavier birth weights among the participants, although not statistically significant, suggest that their mothers also participated more in the supplementary feeding programme, because maternal supplementation during pregnancy was related to birth weight. The fact that the participants had lower prevalences of diarrhoea, which was shown to be unrelated to supplementation, suggests that the sanitary conditions during the longitudinal study were worse for non-participants than for the participants of the follow-up study.

TABLE 7. Early childhood characteristics of participants and non-participants in the follow-up study from the supplemented villages born March 1969-February 1974

Examination of patterns in both the atole and the fresco villages indicates that supplement intake during the longitudinal study was higher in the participants in the follow-up study than in non-participants, although the energy intakes from fresco were small. The lower prevalence of diarrhoea during early childhood among the participants was restricted to the atole villages; also, the higher birth weights among the participants were more pronounced in the atole villages.

The differences found between participants and non-participants may or may not bias analyses of the effects of supplementation, depending on the specifics of the different domains studied. Therefore, the potential biases should be judged for each particular domain. For domains in which the effect of the supplementation programme is mediated through energy or protein, the fact that those with higher intakes are over-represented in the follow-up sample studied would, in theory, tend to overestimate the effect, because the differences in energy and protein intakes between atole and fresco increased as the volume of

supplement ingested increased (see FIG. 1. Schematic representation of biases resulting from self-selection of participants for different types of outcome). This may also be the case if effects are thought to be mediated through vitamins or minerals. Although the content of these nutrients by volume was similar in both drinks, the volume of atole ingested was on the average two to three times greater than that of fresco during the first three years of life. This fact is represented in the figure by the shorter line corresponding to the fresco group. One way of dealing with differences between participants and non-participants is to apply econometric techniques that adjust for sample selectivity as recommended by Heckman [4].

Concluding remark

The follow-up study was carried out in a timely manner, a tribute to INCAP's ability to plan, staff, and execute field studies. State-of-the-art methods were used, and rates of coverage were equal to or higher than generally obtained for studies of its type. The data have been cleaned and summarized for analysis. The tasks that remain include analyses, interpretation, and dissemination of the results of this ambitious study.

Acknowledgements

Data collection and analyses were supported by NIH grant HD22440. The study was a collaborative effort involving investigators at several institutions: R. Martorell (principal investigator, originally at Stanford University, now at Cornell University), J. Rivera (INCAP, Guatemala), E. Pollitt (University of California at Davis), and J. Haas (Cornell University). Dirk Schroeder provided useful comments and suggestions.

References

1. Ruel MT, Rivera J. Castro H. Habicht J-P, Martorell R. Secular trends in adult and child anthropometry in four Guatemalan villages. Food Nutr Bull 1992;14(3): 246-53.
2. Rivera J. Informe anual de actividades, Junio 1987-Mayo 1988, Proyecto Desnutrición Temprana y sus Efectos en la Juventud. Guatemala City: lNCAP, 1989.
3. Pareja G. Rivera J. Habicht J-P, Castro H. Evaluación de la confiabilidad de las mediciones antropométricas. Arch Latinoam Nutr 1989;39:241-50.
4. Heckman JJ. Sample selection bias as a specification error. Econometrica 1979;47(1):153-61.

Contents - Previous - Next