Nutrition Surveys and Surveillance Activities in Russia and the Newly Independent States: A Review of USAID-Sponsored Activities (1994)

◆ ◆ ◆ ◆ ◆

This chapter presents a discussion of the relevant data and the committee 's conclusions concerning each of the three questions posed to the CIN by USAID. With respect to the latter, the committee would like to stress again that its conclusions were based on a relatively small number of studies and problems described in the previous chapters. Nonetheless, the committee hopes that the findings and conclusions presented in this chapter will be helpful to USAID and other decision makers in their work in the NIS and in other regions of the world with similar populations.

The committee approached this question by searching for evidence of an emerging food crisis in Russia at the time of the surveys (1992) or of chronic nutrition problems that existed prior to the surveys and that may still be occurring. The committee's primary goals were to understand the food and nutrition situation in Russia at the time and to examine the methods of data collection for the projects reviewed. The CIN was not asked to critique the specific studies in detail, nor could this be achieved with the information, process, or time available to the committee. In addressing this question, the committee also reviewed data from studies conducted outside of Russia (e.g., the 1993 Anemia

Prevalence Survey in Uzbekistan) to identify issues of at-risk populations that could bear on the situation in Russia.

Three categories of information are useful for assessing whether there is an emerging, acute food and nutrition problem in Russia that represents a deterioration from the previous situation. These are:

1. changes in the prevalence of malnutrition (i.e., poor nutritional status) based on anthropometric, hematological, or biochemical data;

2. evidence of new constraints in the population's ability to purchase food or of new shortages in the food supply; and

3. indications of changes in market food security.

This information must be linked to the options that are available to improve the situation.

The committee defines malnutrition as evidence that an inadequate food supply has had an adverse impact on the anthropometry, hematology, or blood nutrient concentrations of any population group within Russia. Evidence of malnutrition could reflect a preexisting, chronic problem with purchasing power or with the quantity or quality of the food supply, rather than a current, acute crisis. Because the appropriate action would differ depending on whether the problem were acute or chronic, this distinction should be borne in mind throughout the following description of the nutritional situation. Actions to improve the situation will be required not only when there is evidence of malnutrition, but also if nutritional status has worsened between surveys. An example in the Russian population of this latter case would be an observed reduction in body weight even if the population was previously overweight and was still not underweight at the time of the last survey.

In the RLMS, longitudinal data include weight, length or height, arm and head circumference, and midarm skinfold measurements from individuals in 7,200 randomly selected households. Although data were to be collected three times per year, only data for the 17,179 individuals included in Round One (July to October 1992) were provided to the committee. As described in more detail in Chapter 2, the RLMS results must be accepted with caution, because of obvious anomalies in the data and the lack of rigorous statistical analysis or information on sample sizes in the categories presented. Furthermore, it was not possible for the committee to evaluate the anthropometry of infants or young

children (whose anthropometry is usually more sensitive to food shortage) separately, because these were combined for all persons 1–17 years of age.

With these caveats in mind, the general conclusions of the committee concerning Round One of the RLMS were that, for children aged 1– 17 years, there was little evidence of acute, short-term malnutrition (weight or height Z scores ≤ −2)⁶ or chronic malnutrition (height-for-age Z scores ≤ −2). Anthropometric values of the Russian children were similar to those in most developed countries.

The CARE surveys on representative samples of children under 2 years of age, conducted from August to October 1993 in three regions of Russia, showed that current height, weight, and weight-for-age distributions were slightly higher than the international reference population. Retrospective clinic records on the same children produced similar distributions, although the representativeness of the sample, accuracy of these data, and consistency of methods are always of concern in a retrospective study. Mean birthweights were 3,316 ± 501 (SD) g, and only 4.8 percent were < 2,500 g.

Data on birth weight and infant anthropometry were also collected by the Institute of Nutrition in Moscow (Volgarev, 1992), and they are consistent with the RLMS data for children. Prevalence of low birth weight (< 2,500 g) was underestimated to a certain extent, because until 1993, data from nonviable infants weighing < 1 kg were excluded. However, prevalences of low birth weight of 5.8 percent in 1990, 5.9 percent in 1991, and 6.5 percent in 1992 do not suggest either a high or a marked increase in such a prevalence.

With respect to Russian adults, the RLMS data indicate that they tended to be overweight based on their body mass index (BMI, weight/height ²). In most regions, 50–60 percent of individuals aged 35–50 years had a BMI over 25, indicating overweight. Few were underweight; about 1–2 percent had a BMI below 18.5.

These data can only be used to reflect the long-term, chronic food situation in Russia as of 1992. The committee saw no data that were useful for evaluating recent changes in anthropometry (i.e., recent changes in the energy supply) because comparative data from subsequent rounds of the RLMS (planned to be repeated every 4 months) were not available.

Measures of micronutrient status have been collected by the Institute of Nutrition in Moscow between 1983 and 1992 (Volgarev, 1992). The Institute has used small (40–200), nonrandom (convenience) samples of adults (workers from different socioeconomic groups and blood donors), children aged 10–15 years (attending schools in four regions), and infants (attending polyclinics in the Volga and northern Siberian regions). Results of biochemical, but not hemoglobin, measurements are available on the adult sample. On average, plasma

concentrations of vitamin C and B vitamins were low. However, the validity of these prevalence data is currently in doubt for the following reasons: nonrandomness of samples; possible inadequacies in the methods used for sample collection, storage, and analysis; seasonal variability in sampling; criteria used for defining deficiency; and whether reference samples were used. There was no obvious decreasing trend for the biochemical values since 1983, although the number sampled each year was small. No analyses of biochemical measures from children were presented. Data were, however, available for infants. Using a relatively low cutoff for anemia (hemoglobin < 10.2 g/dl compared to the CDC 's recommended cutoff of 11 g/dl), this condition was found in 11 percent of infants in the 1989–1991 Institute of Nutrition surveys in Volga and northern Siberia.

Anemia is generally perceived to be a common condition, but no data were available from the Institute of Nutrition surveys on either anemia or iron status of adults, with the exception of pregnant women. Some data for pregnant women suggested a substantial increase in latent iron deficiency using a (serum ferritin cutoff < 20 µg/liter, substantially higher than the usually accepted cutoff for deficiency of 10–12 µg/liter) in the second and third (but not first) trimester of pregnancy during 1989–1991, but the number of subjects surveyed and the sample were not described. An increase in the prevalence of anemia during that time might be due to a lack of available iron supplements, or to a lack of meat intake of this population in 1992, but this is speculation on the part of the committee.

The CIN was also provided with data from the 1993 Anemia Prevalence Survey in children from Uzbekistan. The prevalence of stunting (height-for-age Z scores ≤ −2) was 16 percent in infants, 32 percent in toddlers, 22 percent in older preschoolers, and 13 percent in school-aged children. The rates of wasting (low weight-for-height) were similar to those expected in the reference population and do not indicate any new, emergency situation with regard to food availability. Because of the cross-sectional nature of the survey, however, it is impossible to determine whether the greater rates of stunting in these Uzbekistani toddlers indicate a recent deterioration in nutritional conditions.

During the Institute of Nutrition inspections and checkups, about 40–50 percent of children were diagnosed as having mild rickets. However, no data on vitamin D status were reported, and the September 1993 CDC Kyrgyzstan Trip Report suggested that the diagnostic criteria used would substantially overestimate the prevalence of rickets. Nevertheless, the northern latitude, possible recent lack of cod liver oil supplements to infants and children (reported in Kyrgyzstan), lack of food fortification with vitamin D, and historical concern about rickets suggest that this condition should be monitored closely.

In general, if energy intake correlates positively with income, this may mean that income is limiting food purchasing ability. Upon inspecting the tabular data from the RLMS presented to the committee, there did appear to be a modest, positive relationship between income and energy intake for adults, but not for children. However, no statistical analysis of this relationship was presented. Data on the adequacy of dietary energy intake by adults and children are difficult to interpret; however, on average, intakes do not appear to be constrained. In addition, as discussed above, more than half the adults in 1992 had an overweight problem.

According to the RLMS, the poorest households (< 50 percent of poverty line) spent about 60 percent of their income on food compared to 52 percent for the wealthiest (> 200 percent above poverty line). This difference was mostly explained by a slightly lower proportion (but presumably a substantially lower amount) of income spent on meat and dairy products. Rural households spent a larger percentage of their food expenditures on cheaper foods such as breads, cereals, and grains and a smaller percentage on meat and milk, compared to their urban counterparts.

Other data from the RLMS on income distribution provide an insight into the magnitude of the problem; 28 percent of the households had incomes below the Russian poverty line as defined in the RLMS, and an additional 54 percent had incomes less than 150 percent of the poverty line. Real incomes may be decreasing for many households in the short- to medium-term, perhaps because of the elimination of food subsidies and escalating food prices. This finding suggested to the committee that a significant proportion of households may be at risk of household food insecurity. One very valuable aspect of the RLMS, provided it is carded out on a regular basis, will be the ability to track households over time to assess the effects of structural adjustment on household food security.

In the CARE surveys on pensioners in Russia, about half of the pensioners surveyed perceived a lack of food security based on their responses to questions such as whether they always have enough money to buy food or their self-perceived weight loss. Reported milk and meat intakes were relatively low for this group in the view of the committee, although the possibility of underreporting must be considered. Some regions received large amounts of food aid (such as dried milk and cereals) in 1992, while others did not. The reasons for these differences could not be discerned from the information provided to the CIN. In the CARE study of Armenian pensioners in Yerevan, these pensioners appeared to be considerably worse off than pensioners in Russia. In part, this was caused by a shortage of food resulting from the blockade and war with Azerbaijan, as well as low pensions, according to the CARE survey.

In the CARE Pensioner Surveys, about half of the respondents reported a weight loss of more than 5 kg in the past 6 months. Although the accuracy of this self-reported weight loss is dubious, it is, nonetheless, indicative of a fairly widespread belief that the food supply was less adequate than it had been previously.

The Goskomstat ongoing survey of 49,000 Russian families showed a decrease in the consumption of meat, fish, dairy products, vegetables, and fruit between 1991 and 1992, which ranged from about 10–30 percent among these food groups. There was no change in bread, oil, or fat intake, but there was a slight increase in sugar intake. A shortage of infant formulas, perhaps to the extent that only half of the demand was met, was mentioned briefly in the Institute of Nutrition report and was attributed to a shortage of quality raw materials (Volgarev, 1992).

Data on the frequency of consumption of foods from five food groups (dairy, fruits, vegetables, grains, meats) were collected during the CARE survey of children less than 2 years of age in Russia, which was conducted from July to October 1993. However, no frequency data were available in the draft report provided to the committee. The information presented was analyzed in terms of an unvalidated “mean adequacy ratio,” which attempted to convert food frequency data into risk of specific nutrient deficiencies. For this reason, this report could not be used to assess food availability.

Market food security refers to the presence of enough food in the market at usual (reasonable) prices. This is also information that is well-understood by most decision makers and may be more likely to trigger action than perceptions of household food insecurity or measurements of changes in food consumption. Extensive data on various aspects of market food security were collected during the RLMS, but they were not presented in the Round One report provided to the committee.

The 1993 Anemia Prevalence Survey in Uzbekistan, which was completed during June and July 1993 and in which a high rate of stunting in growth was noted in children (see previous section), also contained information on the consumption of specific foods during the previous 24 hours. The data were presented as percentage of individuals from each age group who consumed any food from each of eight food groups. No quantitative information on food consumption was obtained. Of the adults studied, approximately 100 percent had consumed staples; 90 percent consumed meats, dairy products, fats or oils, and sweets; 40 percent consumed vegetables; and 20 percent consumed fruits and nuts. Except for the infants, the consumption patterns did not vary substantially by age group. The results are notable in the context of understanding the etiology of anemia because of the high frequency of meat and dairy product

consumption and relatively low rates of fruit and vegetable consumption. Because of the nature of the questions in this cross-sectional survey, it is impossible to determine whether these patterns represent a change from former eating habits.

There is certainly evidence that the nutritional conditions of infants and toddlers could be improved, possibly by provision of enhanced complementary foods or education on the preparation of these foods from available ingredients, appropriate fortification of existing complementary foods, control of common childhood infections, or some combination of these approaches.

Beyond the disruption of food trade, food price subsidies, and the availability of food and medicines that have resulted from the breakup of the Soviet Union and the implementation of economic reforms, it appears that there have been disruptions that affect food fortification with nutrients and nutrient supplementation of vulnerable groups. The committee is very concerned, for example, by anecdotal reports that salt iodination is limited or nonexistent because the importation of iodized salt has been affected by trade barriers and because of a lack of funds to repair iodination equipment. Most of the NIS have endemic iodine deficiency, with historically high rates of cretinism and other consequences of this deficiency.⁷ Failure to ensure iodination of the salt supply could result in serious, long-term health problems, even in the absence of a general shortage of food. Monitoring of the iodination situation must, therefore, be a high priority and can be conducted independently of food and nutrition surveys.

The committee had similar concerns regarding the availability of iron supplements to women and children. Supplements appear to have been used as treatments for anemia, rather than as a preventive measure. Isolated reports of an increasing prevalence of iron deficiency (Volgarev, 1992), and of an inadequate supply of iron supplements, are disturbing. According to the CDC Kyrgyzstan Trip Report of September 1993, there was a general concern about rickets and a statement that cod liver oil was unavailable. Again, a failure in the system that provides these important micronutrients through supplements could have serious health and nutrition consequences that would not necessarily be detected through traditional food consumption surveys.

An additional problem is that the social safety net that contributed directly to maintaining adequate health and nutrition in certain dependent population groups (e.g., orphans and the elderly or disabled who live alone or in public

homes) may have been severely weakened and in some cases may have ceased to function altogether.

Based on the contextual information available to the committee, the evidence was strongest concerning the nutritional risk faced by elderly pensioners as a result of the disintegrating social safety net. In most large Russian cities, there has been a program run by the Department of Social Assistance (a division of SOBES, the Department of Social Welfare) through which social workers provide special assistance to pensioners who live alone, invalids, single mothers, and others who are considered needy (CARE Pensioner Surveys). One important form of assistance has been visits to homebound individuals to help with personal care, shopping, and food preparation.

The CARE Pensioner Surveys highlight the importance of monitoring the level of assistance through the Department of Social Assistance by making it clear what a central role this program has played in protecting the health and nutritional status of dependent pensioners. Although, the surveys did not present direct evidence of deterioration in these services, it seems logical to assume that declining public resources may cause some social workers to lose their jobs, while the erosion of the value of salaries will cause others to seek secondary forms of income generation. Either scenario would lead to less assistance for those who have been dependent on public assistance and would likely exacerbate the nutritional risks directly associated with the declining value of their pensions.

The macroeconomic reforms that have been implemented throughout the NIS are intended to stabilize the economy and, ultimately, to enhance economic growth and alleviate poverty. A basic assumption in this process is that economic growth at the national level translates into improved incomes at the household level. The committee expects that improved incomes will enhance food security, as well as improve health and nutritional status. However, even where macroeconomic reforms result in higher incomes, there is a lag time for this to occur. During this lag time between the initiation of reforms and a newly established equilibrium, certain subgroups of the population may be unable to adapt to structural adjustment, at least in the short- to medium-term. It is, therefore, important to identify those households and individuals that are likely to be affected adversely by structural adjustment and to maintain or initiate appropriate types of safety net programs that could buffer adverse effects of economic constraints on health and nutrition during the transition period.

It should be noted that all surveys in Russia were conducted prior to or during 1992, except for the 1993 CARE study on children under 2 years of age. Thus, the committee's conclusions apply only to the situation up to these times. There were no analyses of longitudinal data from Russia that enabled the committee to evaluate changes in food security or nutritional status during the time periods examined. The CIN made the following conclusions regarding Question 1:

• Due to lack of appropriate qualitative data, it is not certain that the population groups surveyed were those at most risk of food shortage or nutritional problems.

• Based on the available documentation, and taking into consideration the apparently high average BMI of the adult population of the NIS prior to the recent social and economic changes, there appeared to be little evidence of energy deficiency or widespread malnutrition in Russia at the time the surveys were conducted (1992). There was a high prevalence of overweight up to this time; however, this finding does not preclude the possibility that a gradual weight loss due to energy shortage is currently occurring.

• In the CARE survey of Russian children under 2 years of age conducted from July to October 1993, there was no evidence of low weight-for-age, height-for-age, or weight-for-height in any age group. Thus, the committee concludes that there was no objective evidence of undernutrition among the Russian children less than 2 years of age who were included in the sample.

• In the RLMS, anthropometric data were aggregated for children 1–17 years of age, which made their interpretation difficult. With this caveat in mind, this group showed little evidence of undernutrition based on weight or height measures.

• No longitudinal data were available from the CARE or RLMS surveys, so that it was not possible to conclude whether the food and nutrition situation was getting worse at that time. There were some indications, however, that the food system was under stress in late 1992. About half of the pensioners in the CARE surveys answered affirmatively to a question about whether they had a 5-kg weight loss in the 6 months prior to being surveyed. Although the validity of the amount of weight loss is questionable, there is clearly a perceived food security problem among this group. Although average BMI was not low at the time of the CARE surveys, the distribution of BMIs was not provided to the committee. These data possibly may have revealed a certain proportion of individuals with energy deficit, although this is speculative. The 1992 report by Volgarev reveals reductions in meat, fish, dairy product, vegetable, and fruit consumption between 1991 and 1992.

• Some of the populations surveyed showed evidence of chronic micronutrient deficiencies. Anemia is generally perceived to be a common condition in the region, but the only data available in Russia were from pregnant women. It is possible that a relatively high percentage of anemia in this group has been exacerbated by recent events, including a shortage of iron supplements. Other chronic micronutrient deficiencies reported in the survey populations included iodine and vitamin D. Iodine deficiency may be endemic in Russia and the NIS, and iodination of salt may be threatened by new trade barriers and lack of funds to repair equipment. Although the prevalence of mild rickets may be overestimated, there are reasons to believe that women and children are at risk of vitamin D deficiency.

• The relevance to the nutritional situation in Russia of the mortality data presented could not be interpreted, because important potential confounding variables (e.g., availability of medications) were not considered by the researchers.

• Anecdotal evidence suggested that consumption of alcohol in the Russian populations surveyed was rising, but the relevance of this finding to food security or nutritional status was not clear. Future analyses of the RLMS data may shed light on this situation.

The committee first offers a number of suggestions to guide the process that should be used to design and implement future food and nutrition surveys in the NIS. These are followed by a list of specific measuresthat might be included. Because of the CIN's assumptions (see Chapter 3), these comments relate, above all, to information gathering about a relatively rapidly evolving food crisis.

To be useful, the information to be generated from a nutrition monitoring system must give timely and adequate guidance that enables a quick and accurate decision to be made about some course of action. To begin with, one needs to know:

• Whether or not there is an emerging problem with respect to nutritional status about which one should be concerned. This requires knowledge about which problems are usual (chronic) and which are new or changed.

• Some information about the causes of the problem, especially those etiologic factors—be they nutritional or programmatic—that interventions can address.

• Which population groups are likely to be most affected. This means understanding causality well enough to know what households are most vulnerable and how the households allocate the deprivation within the household.

• Options for action. These options define the need for other information especially relative to targeting interventions, monitoring their implementation, and assuring that the chain of causality is affected as presumed. Identifying options for action also permits identification of mechanisms for making decisions about the options, which relates to the next point.

• Who decides on the options of action and how to structure the information generation and communication so that the decision makers are properly informed in a timely fashion.

It is imperative that the courses of action, and the lead times for their implementation, be understood during the design of information generation systems for rapid assessment of emerging situations. In situations of rapid change, decisions about how to proceed are often made from afar or even outside the country or continent. This is also true for deciding how surveys will be conducted. This tendency should be avoided.

Qualitative information is available locally that can give insight into all five of the above points, and it is better suited for this purpose than quantitative data alone (i.e., data without the qualitative context). Previously collected quantitative data, if available, are useful to substantiate insights derived from qualitative data, but they are not essential. Insights are essential for designing a quantitative survey that can be conducted efficiently and interpreted with confidence. This qualitative information can be used to address changes that threaten nutrition, to understand coping strategies, and to discuss intervention options and who will determine them. It may be obtained from interviews with national/regional/local decision makers, key informants, frontline service staff, researchers, and others with an informed interest or stake in the outcome.

The following six questions are designed to elicit information that can inform decision makers about current or potential adverse changes in the nutritional status of the population.

1. What changes in market food supply have occurred that are of concern?

   This question refers not only to reasons for overall food shortages, but also to shortages of specific kinds of foods, such as fortified products intended to prevent previously prevalent micronutrient deficiencies. Information should be obtained on perceptions of dietary changes that may occur or that have occurred as a result of economic stress. It is necessary to know why these changes have occurred and whether they can be reversed to the extent that they affect the food supply and food prices in local markets. This information on market (local) food security predicts where it is important to target further information gathering.

2. What were the usual sources of income and food in this population? What population groups have lost direct access to food and now receive less income relative to food prices?

   This information defines the vulnerable groups in the population that should be targeted for further information gathering. In most situations of nutritional stress, the most vulnerable groups are assumed to be women (especially during pregnancy and lactation), infants, and young children. This assumption is based on the higher nutrient requirements of these groups and the potential for malnutrition to cause irreversible harm to the fetus or child. Another potential at-risk group is the elderly because of their fixed incomes that are vulnerable to depreciation during inflation; health problems that affect food purchasing, intake, or utilization; and social problems such as living alone or dependency on others. Although these population groups were sampled in the NIS surveys supported by USAID, the committee felt that the usual assumptions about vulnerability should be reexamined in the context of the NIS. For example, although children may suffer more than adults as a result of chronic or acute food shortage in developing countries, a recent crisis in a previously well-fed population such as the NIS might result in parents going without food in order to feed their children adequately. Unemployment and other stresses caused by worsening economic and changing social conditions may result in alcoholism, anorexia, and generally poor eating habits in both male and female adults. The CARE surveys showed that female Russian pensioners living in NIS other than Russia appeared to be at particular risk, because they were less likely to have relatives nearby and, perhaps, were victims of rising ethnic tensions. Small communities had a higher proportion of elderly, and a higher proportion of them were living alone.

   To acquire the necessary qualitative insights into who the probable vulnerable groups are, information must be collected about how food was previously and is now distributed to schools and other institutions; to workers and the unemployed; and to individuals such as pensioners, children, infants, and orphans. The committee feels that this information is best acquired by interviews with key informants.

3. Are there increases in nonfood expenditures that threaten food purchasing power?

   The problem of expenditures threatening food purchasing power may occur when costs of other basic needs besides food, such as medicines or housing, are excessive. This information can be obtained by interviewing officials responsible for these human needs (e.g., officials in health or welfare departments).

4. What mechanisms previously prevented destitution and malnutrition, and how are they supposed to be implemented?

   Normative information (i.e., on what should be happening) should be obtained from senior informants in ministries and other agencies. However, questions about the actual situation (e.g., “Are those who are supposed to be implementing the activities at the front line still able to do so?”) should also be asked of those implementing activities directly who may be suffering a range of stresses. They may, for example, have to travel long distances or have to repair critical machinery (e.g., for iodine fortification) under circumstances where money or time may be limited. Such questions may help to determine vulnerable areas or if services are failing for certain groups, thus increasing their vulnerability.

5. Are there changes in determinants of nutrition besides food availability?

   Examples of changes in non-food determinants of nutrition include the loss of the social system that supplied help for food purchasing and preparation among the elderly, or the lack of medicines or supplements. The information described under Question 2 will usually give an indication about those stresses that need to be mentioned.

6. Have new interventions been implemented?

   It is necessary to know what is supposed to be implemented (this can be learned from senior officials) as well as whether it is actually being implemented and whether the effects are what were expected (best learned from frontline implementors).

The following two categories of qualitative information can provide added useful information to decision makers.

1. Understand coping strategies.

   Under food stress, old food strategies are expanded and new ones developed. These coping strategies need to be identified to develop measures of their use and success. Of particular concern is identifying behavior that was previously present, but only becomes a nutrition coping behavior under food stress. Determining the prevalence of such behavior may be meaningless; only changes in prevalence may be informative. For instance, it is hard to judge the relevance of the prevalence of gardening as an indicator of food need (a question asked in the CARE Pensioner Surveys), if it was previously a widespread recreational activity.

2. Discuss intervention options and who will decide them.

   For information to be useful to decision makers, it must be relevant to the decisions to be made—both in terms of whether an intervention is needed and how it should be implemented. This requires knowledge about what kinds of interventions are possible. The CIN assumes that USAID's role in Russia and the other NIS is faciliatory, with the major responsibility for the resources and administration of interventions falling to the respective NIS governments. Key decision makers who might be involved should be identified early on, before a quantitative survey is designed. Long experience using surveillance shows that, for information transfer and use to be timely and efficient, the decision makers must be involved in deciding what information is useful, when it is needed, and how data should be interpreted. In the committee's experience, it is unusual for decision makers to be involved in considerations of study design, and a special effort may be needed to make them or their staff perceive this task as a priority.

The above qualitative information provides guidance about what quantitative information to gather and from whom. The following three items are examples of quantitative information that should be considered in designing and conducting surveys and surveillance.

1. Sampling frame.

   After the vulnerable areas and groups have been identified by the above information, the appropriate sampling frame must be identified. In the surveys reviewed by the CIN, the selection and implementation of sampling frames seemed appropriate except where noted, but this observation requires further investigation to ensure that high-risk groups were not excluded (e.g., lists of recipients of social services may not include recent immigrants). Inspection of the qualitative data described above could be useful to determine whether food problems are relatively universal or more concentrated in specific population groups or regions and, hence, whether widespread or more focused surveys are needed.

2. Review possibilities for data collection, including existing surveys or activities.

   In reviewing possible sample frames for data collection, one should consider the possibility of using different surveys for different target populations or individuals. One reason is that a survey in one vulnerable group (e.g., the elderly) may have a different sampling strategy and require collection of different data from a survey in another vulnerable group (e.g., young children). Another reason is that it may be more efficient to introduce nutritional and food security considerations into other kinds of surveys (e.g., health, welfare) that are almost certain to be considered for implementation in these kinds of situations

and, if combined with a nutrition survey, can permit more informed decisions about integrated interventions. Such surveys may already exist, or be planned or on-going and need little modification. Certain activities that are not surveys also afford opportunities for data collection because they sample the right individuals (e.g., immunization of children in vulnerable age groups).

These piggyback mechanisms should be considered before mounting stand-alone nutrition surveys. Focused nutrition surveys may still be necessary to assess vulnerable groups not adequately sampled by the above surveys, but a new survey to address nutritional concerns is usually uninterpretable if health considerations (especially among infants, toddlers, and the elderly) are not also considered.

3. Kinds of data.

   It is important to differentiate between data that are vital for revealing outcomes of concern (e.g., malnutrition discerned by anthropometry, household food security, or perceptions of change for the worse) from data that illuminate the causes of these outcomes (e.g., economic status or lack of iron supplements). Data needed for interpreting outcomes will determine the way the outcomes are presented by providing the categories for comparisons. For example, if weight loss data are presented by employment category, it may be possible to infer which groups are at greatest risk. These comparisons must produce meaningful information that relates to the interventions that can be instituted. Sometimes they require comparison to outside standards. This is especially true when one is concerned with prevalence, because these are only interpretable if they are based on widely accepted criteria (e.g., BMI < 18.5 for adults or height-for-age Z scores of −2 or −3 for children).

Data collection, and especially data analysis, is much more efficient if the data are restricted to the minimum necessary to make the comparisons required for action. If other data are to be collected, the reasons for their collection need to be clearly stated. For example, certain data may be needed to plan responsive short-term actions, other data may be needed to design longer-term actions to deal with chronic nutrition problems, and still other data may be needed to obtain knowledge that has wider applicability (scientific information). When data are collected for different purposes, an important principle is that the analyses must be carefully defined to meet the short-term needs before the data are collected. Adhering to this principle will ensure that essential data get priority in data management (e.g., coding, cleaning, and compiling), so that these data may be quickly analyzed and interpreted.

Power analyses should be done to ensure that a large enough sample of households or individuals is included in the survey and that the sample is not excessive. In general, power analyses should be based on the magnitude of

change that must be identified. In the committee's opinion, sample sizes in the high hundreds are often sufficient to detect changes important for the kinds of short-term decisions required.

The qualitative information collected will give an idea about the speed with which the emerging situation is likely to evolve. On that basis, monthly information is probably needed in Armenia due to the rapidly changing situation (and is apparently being collected by the government), but in Russia less frequent monitoring might be appropriate. From other sources, one should also have an idea about the relevant actions that might be envisaged to deal with the emerging situation. There is a lag time between availability of information and the decisions to take action; there is also a lag time for decisions to be implemented. Part of the purpose of setting up surveillance (see “Institute Surveillance” below) is to shorten that decision time.

Information from previous surveys, coupled with continuing qualitative information gathering, may indicate a need to monitor the nutrition situation and the evolution of its determinants in a subgroup of the population. It is important that both the sampling frame and the reduced number of variables to be collected be thought through afresh. One approach is to conduct repeated cross-sectional surveys that do not necessarily include the same households or individuals, such as was done in the CARE Pensioner Surveys. These surveys require larger sample sizes to identify changes than do longitudinal surveys of the same individuals. However, longitudinal surveys such as the RLMS are more demanding because of the expense and difficulty of remeasuring a high proportion of the same individuals from survey to survey. Another problem with longitudinal surveys is that the cohort measured often becomes unrepresentative of the population of concern. This problem can be caused by such factors as the increasing age of individuals or change in employment status.

Surveillance is a system that effectively links the data generated for decision making with the actual decision-making process. Again, this process involves knowing and working with those who make the decisions for specific interventions. The committee could not ascertain who the decision makers were for the studies reviewed; the exception was the Armenian government in the case of their ongoing surveillance system.

In considering the cost and utility of a survey, it is important to define the specific information that will be generated. The most appropriate method to accomplish this is to prepare dummy tables of results prepared from Day 1 of the

project, and predict likely inferences from these tables related to food and nutrition policy decisions that will need to be made. Another important issue is the timeliness of information generation and how long it takes to transfer the information to decision makers. Unless the decisions that are going to be made on the basis of information generated from future surveys are directly relevant to interventions or policies and are predefined (a prerequisite for developing the dummy tables and predefining the analysis), and unless the decision makers are identified and involved in the development of the surveillance system, the committee strongly advises against any further surveys or monitoring, unless their purpose is limited to validating assumptions about causality and vulnerable groups.

If surveillance is to be undertaken for any length of time, it is essential that criteria be established to judge its effectiveness and that these criteria be used to evaluate and thus improve the surveillance. These criteria should include both the content of the information (of which data quality is a small but essential part) and the timeliness of its transfer to decision makers. The mechanism to monitor surveillance needs to involve those who use the information to make decisions (i.e., those who would use the information if it were useful and timely).

There is also a need to understand what is going on in the NIS in order to validate assumptions made about mechanisms of, and responses to, change in nutritional status, and to validate who the vulnerable groups are. The committee felt that this information is important not only to understand the situation as it evolves in the near future, but also in the longer term. This tactic requires information gathering that need not be so well predefined or so timely. However, the design of data collection methods and analyses must meet higher criteria of scientific validity relative to causality for the results to be interpretable than is required for the decision making described above. Some present and proposed data collection activities are in this category (e.g., parts of the RLMS that go beyond the purposes mentioned above) and need to be considered on the basis of their scientific, rather than their immediate, decision-making merit. Scientific quality is important because the data obtained must be valid. The best way of assuring this scientific quality is through critical peer review.

The above considerations relate to gathering information that will enable decisions to be made to mitigate or reverse the ill effects on nutrition of socioeconomic dislocations in relatively well-off populations. The more chronic issues of anemia, childhood undernutrition in some of the NIS, alcoholism, and iodine and vitamin D deficiencies also need to be addressed.

Much information can be collected by methods other than quantitative household surveys. It is important to be sure that quantitative data from surveys are really necessary to take action. For example, if it is clear that distribution problems are causing local market food insecurity, these problems can be dealt with without quantitative information from household food surveys.

The specific information to be collected should depend on current circumstances and the possible actions that might be taken. The qualitative information described above must first be collected in the field before the specific study design and appropriate variables are decided upon. With this caveat, the CIN proposes the following three levels of information.

In addition to the preliminary qualitative information described above, the following kinds of socioeconomic, anthropometric, and household food availability information should be collected in a representative, cross-sectional survey. This type of survey should be able to detect a food crisis, or at least a growing risk of one, and determine why it is occurring (e.g., shortage of food in the market and economic problems).

Socioeconomic Data Information should be collected on household demographics, employment, income, ethnicity, language, and other characteristics that are likely to be associated with increased risk of food insecurity and deteriorating or poor nutrition. This information should permit construction of a vulnerability classification that is useful to decision making and decision makers. These characteristics should have been identified by the preceding qualitative data collection.

Measured (not Reported) Weight and Height of Young Children It is usually assumed that children are a vulnerable group, although this must be confirmed based on qualitative or other information. In many situations, the 12- to 24-month age group is most revealing of change in the general nutritional

situation in the past year, because growth faltering usually occurs between the ages of 3 and 18 months. Older children can be included to increase sample size, but there is a strong risk that their growth-faltering occurred some years before. Height is more sensitive to growth faltering than weight. The prevalence of height Z scores ≤ −2 and ≤ −3 should be calculated; any evidence of height Z scores < −3 indicates a serious situation. It is now accepted that weight-for-height of children between about 6 months and 24 months is usually low only in situations of very severe starvation, but it might be useful to monitor changes in the prevalence of weight-for-height Z scores ≤ −2. In all cases, the distribution of Z scores must be presented to permit determination of the prevalence of low values and, if repeated, changes in the distribution over time.

Body Mass Index of Adults Weight and height should be measured and used to calculate BMI in kg/m². The World Health Organization now accepts a BMI < 18.5 to be an indicator of inadequate energy intake. Thus, the distribution of BMIs must be presented.

Food Availability at the Household Level One of the earliest signs of major changes in food availability or in purchasing power is that households and individuals perceive feeling insecure about their food supply. Culturally appropriate questions about these perceptions can be based on examples in the Cornell Hunger Scale,⁷ with the caveat that the scale was developed in the context of a stable economy where most people felt secure about food. Therefore, questions must be designed based on the understanding of recent changes in the food supply and of coping strategies used by specific population groups that were obtained, often quantitatively, before the study is designed. Efforts should be taken to minimize purposeful misreporting to demonstrate neediness (e.g., by careful wording of questions and appropriate training of interviewers), and to maximize the face validity of the indicators in the eyes of decision makers. Extended durations of breast-feeding is another possible indicator of food insecurity that can be elicited easily through interviews.

It may be useful to ask whether there are any backup food stores in the house, or whether there is only food available for that day. In the United States, where food is generally purchased once a week, a lack of food stores indicates an acute food shortage. In the NIS this may not be the case. Therefore, qualitative, cultural information should be collected prior to designing the question. Staples are usually bought over a longer-term period and low stores of these items may be more serious. When possible, visual inspection of food stores is advised to validate respondents' statements.

Purchasing Power Obtaining quantitative information about purchasing power requires an expenditure survey such as that in the RLMS. These types of surveys should be avoided when information is needed rapidly, because collecting data on expenditures can slow down the timely flow of information. Questions about perceived economic resources to purchase food may be more practical.

Biochemical Analyses These methods are expensive and require careful standardization across laboratories in order to be interpretable. They are not useful for short-term surveillance. Hemoglobin concentrations can be measured with a portable monitor if information on anemia is desirable (e.g., in population groups with a suspected high prevalence) or if iron supplementation is possible. Accompanying qualitative information should be collected on possible causes of anemia or changes in its prevalence, especially on iron supplementation practices.

Food Intake If repeated surveys are anticipated in order to monitor changes in the amounts and types of foods consumed over time, information should be collected on one 24-hour recall of foods consumed by the target individuals. This information should be semiquantitative (i.e., include reported amounts of each food consumed). Conversion of food intake data to nutrient intakes is not particularly useful if the purpose is to detect food emergencies.

Prevalence of Rickets, Goiter, and Cretinism Rickets, goiter, and cretinism will increase if there is failure to provide supplements of vitamin D and iodized salt. These conditions represent a serious threat to the long-term health of the population, but they may not, of themselves, indicate a food shortage. Trends in prevalence may be available from national surveys, health practitioners, or clinics. However, the number of individuals affected needs to be sufficiently large if trends in prevalence are to be detected through nutrition surveys.

Infant Feeding Practices and Food Availability These measures include the duration of breast-feeding and the availability of nutritionally adequate, complementary, and weaning foods.

Other kinds of information are also useful as general indicators of the health and nutritional status of the population. These may reflect changes in the health system more than changes in the food supply. Examples include the prevalence

of maternal-infant problems, such as percent low birth weight (< 2,500 g) and infant mortality, and mortality rates of other population groups.

Information that the committee would have found useful in answering Question 2 includes: the extent to which the current food situation is changing in the NIS and the rapidity of this change, the type and size of role that USAID could play in alleviating any acute or chronic food or nutrition problems, and whether there is concern about collecting information or providing assistance that will improve the sustainability of solutions and monitoring. Given the available information, the committee concluded the following:

• Selection of appropriate indicators and methodologies depends very much on the availability of qualitative information, much of which should be collected locally and in advance of planning. In general, such information should provide a sense of the problems that might exist and their likely causes, whether the problem is likely to be acute or chronic, probable vulnerable groups, options for action, who has need of the information, and how quickly the information is needed.

• At a minimum, specific qualitative measures should be elicited in advance of any survey and should ideally provide information on such items as recent changes in market food supply, usual sources of income and food and the population groups for whom access to these have changed, rising costs of competing necessities, the functionality of existing food and other safety nets, knowledge about how household food strategies might change under stress, and what kinds of interventions might be possible in a given population in a given region (which requires talking directly with the potential decision makers).

• It is important to distinguish between the need for data that are vital for revealing undernutrition (e.g., anthropometry) and lack of food security and the data necessary for interpreting outcomes (e.g., economic data). Data collection will be achieved more rapidly and efficiently if restricted to the minimum information necessary to assess the situation and to make comparisons required for decision making.

• At a minimum, survey measures should include: socioeconomic data, measured weight and height of children (preferably 12–24 months old), BMI of adults, food availability at the household level, and market food security. Optional measures include purchasing power; biochemical analyses including hemoglobin; food intake; prevalence of rickets, goiter, and cretinism; infant feeding practices and availability of appropriate foods for infants and young children; and general indicators of nutritional status and health.

• The minimum essential data to be collected and the analyses to be performed must be carefully defined before data collection. Dummy tables of

results should be prepared from the outset of the survey and likely inferences should be made on which further decisions can be based. Steps should then be taken to ensure that these predefined data get priority at all steps of data collection through analysis, so that results can be quickly returned.

• Sample sizes in the high hundreds are often sufficient to detect changes in prevalence of outcomes such as BMI and most other anthropometric measures that are important for the kinds of short-term decisions required. Whether a representative regional survey is needed, rather than one focused on specific population groups or regions, will depend on the initial qualitative information concerning likely vulnerable groups, the universality of the food problem, and possible types of action.

• The need for and timing of repeat surveys depend on the speed with which changes in the situation are likely to evolve.

• A surveillance system requires a minimal lag time between data collection and decision implementation, which should be discussed with decision makers.

• Repeat surveys may benefit from differing sampling frames and changes in the variables collected over time. Decision makers must be identified and involved in the development of any surveillance system for it to achieve effective decisions.

• Links with existing surveys and activities should be considered both to maximize cost-effectiveness and to assist in interpreting findings.

• To detect changes in food or nutrition status, a larger sample size is needed for repeated cross-sectional surveys than for longitudinal surveys (where the same household or individuals are monitored continuously). Longitudinal sampling, however, is often more costly and difficult to implement.

• Criteria must be established to monitor the effectiveness of any surveillance system in terms of information content, timeliness, and usefulness for decision making.

• Chronic preexisting nutrition problems should be identified and monitored, especially if these are likely to be exacerbated by current crises. Examples include anemia and the need for fortification with iodine, vitamin D, and possibly other micronutrients.

As noted earlier, evidence available from the surveys examined does not permit definitive conclusions regarding the impact of the current political dislocations in the NIS on the food supply or nutritional status of vulnerable groups. In part, this gap exists because the surveys did not allow the committee to determine which are the most vulnerable population groups or to understand the causes of any food-related problems. In addition, the committee was provided

with little information about what strategies might be programmatically feasible for USAID or other funding agencies, and therefore its considerations about how the information could be used programmatically were limited. Also, the committee was uncertain how the health policymakers in the NIS view, or are involved in, the USAID decision-making process. Thus, the committee's specific programmatic recommendations to USAID are made cautiously.

It is clear, nonetheless, that certain population groups lacked food security and that there was evidence of infrastructural problems that could precipitate a worsening situation. In addition, there was evidence of preexisting micronutrient deficiencies that will worsen if food fortification and supplementation are jeopardized by the current turmoil. These problems are sufficient to suggest specific actions that should be considered by USAID. These include determination of the most appropriate intervention strategy in each instance, provision of needed foods and food fortification, institution building, and collaboration with, and building upon, other activities in the region that foster population health and well-being.

Several important issues must be considered when designing the appropriate intervention strategy. First, is food lacking nationally or locally, or are the means to purchase food lacking? When food is lacking locally (market food insecurity), food prices rise relative to other goods. Even if foods are found in the market, they are sold at exorbitant prices. High food prices may also be explained by overall inflation. However, some households, whose income does not rise with inflation or who lose their purchasing power, may suffer household insecurity even though there is no market food insecurity. Therefore, household food insecurity can arise from market insecurity and/or from inadequate household food purchasing power.

Direct food aid from the United States to assure market food security should only be considered when the national food supply is inadequate (although this is a complex issue for which political, economic, programmatic, contractual, and diplomatic variables need to be considered). Otherwise, importation will cause a fall in food prices, have a negative impact on agricultural and home food production, and create certain expectations at a household or programmatic level that are difficult to change later. A possible exception is when food is carefully targeted to vulnerable groups who do not consume a major proportion of the food supply. Often the market food insecurity arises because of transportation and distribution problems. Provision by the United States of assistance to alleviate these problems can be more helpful than direct food assistance.

The nutritional status of the NIS population—and of particular population groups such as the elderly, pregnant women, and infants—was formerly strongly dependent on the social service network. Attention should be paid to the extent to which deterioration in this system could explain any deterioration in nutritional status and to the potential benefits of protecting specific services. Similarly, there was strong anecdotal evidence for lack of medical supplies such as vaccines, of nutrient supplements such as iron and vitamin D, of fortified foods such as iodized salt, and of appropriate infant foods. If true, this lack may explain a significant part of any increased prevalence of mortality, morbidity, anemia, and micronutrient deficiency diseases, such as goiter and rickets, and infant growth faltering.

It is the committee's understanding that food assistance is currently being provided by donor agencies to Armenia, Azerbaijan, Tajikistan, and Georgia in the form of commodity grains, vegetable oils, dry milk, and a small amount of blended foods. Such assistance was of value in the short term, given the evidence from the CARE Pensioner Surveys in 1992 that this food reached a high proportion of the elderly, at least in some regions. In the longer term, USAID should explore the possibilities of collaborating with other U.S. groups such as the Food and Drug Administration (FDA) to provide assistance in bringing about structural changes that might have longer-term impact. For example, the examined reports suggest a high prevalence of marginal vitamin D deficiency (an observation that needs to be confirmed), which should be preventable by proper fortification of foods such as milk. The FDA can use its expertise in the food fortification area to assist in implementation of appropriate fortification measures in the NIS. Where such fortification is not possible or sustainable, USAID should provide assistance in facilitating the importation of appropriate fortified foods or supplements for vulnerable groups.

Unlike USAID policy in Africa, Asia, and Latin America, the Bureau of Europe and the NIS indicated during the committee's deliberations (continuing the policy of the earlier NIS Task Force) that long-term development is not a primary goal for the agency in Russia or elsewhere in the NIS. However, in a March 1994 memo to the CIN, USAID's chief of health programs states that “as USAID shifts focus away from short-term emergency responses to health and nutrition problems, the Office of Democratic Initiatives and Health and Humanitarian Resources (DIHHR) is responsible for longer term health

programs which focus on institutional development” (see Appendix A for the complete memorandum).

In view of this conflicting information, the committee based its recommendations on two assumptions. First, the most likely time frame for continuing USAID assistance in the region is another 3–5 years —perhaps through the end of this decade, but not beyond. Second, the primary goal of the agency is to assist in averting food and nutrition emergencies in order to protect the health and welfare of the population and to contribute to social and political stability.

Nonetheless, the committee wishes to express its view that institutional support and institutional development are important goals that should be kept in mind, even within the relatively short time frame and limited mandate that the USAID has established for itself in the region. Given the relatively high level of professional training in the population, the relatively well-developed food processing industry, and the well-developed social support infrastructure in Russia and many of the other NIS prior to 1990, institutional support and development in the area of food and nutrition would not require the same investment in basic training and basic infrastructure that would be required in most other regions where USAID is actively involved in providing assistance. Therefore, while USAID provides short-term assistance to nutrition and health monitoring and surveillance, there will be many opportunities wherein a small amount of additional effort will make an important contribution to leaving behind strengthened institutions without seriously competing with more short-term emergency related goals. These additional efforts might include systematic technical training with detailed written protocols, copying and sharing of recent relevant academic papers and reports, widespread discussion and interpretation of surveillance and monitoring data within an explicit conceptual framework, and the purchasing of equipment that will remain on site.

USAID should make every attempt to leverage other programmatic resources in the region to promote population health and well-being. The preexisting nutrition and food-related social programs and industries in Russia and the other NIS, and the high level of current bilateral and multilateral interest in the region, provide opportunities for USAID to leverage its food and nutrition monitoring assistance. Where milk kitchens, school lunch programs, and the Department of Social Assistance visits to shut-ins continue to function, USAID-assisted nutrition monitoring may provide helpful information for targeting the limited resources of these programs to those populations most in need.

United Nations Children's Fund (UNICEF) is establishing representatives in several NIS. Given UNICEF's commitment to child welfare and its capacity to marshal resources for child survival and development in places where

USAID-assisted nutrition monitoring and surveillance identifies a substantial risk among infants and preschool children, it may be possible to work jointly with UNICEF to design and implement appropriate interventions. Given its commitment to promotion of breast-feeding, UNICEF may also be an appropriate and willing collaborator in studies of possible breast milk contamination in populations exposed to the Chernobyl disaster or other environmental contaminants and in the provision of appropriate education to mothers. This issue is important in that a decline in breast-feeding in a population is not easily reversed when a food crisis arises, which puts infants in that population at greater risk of inadequate feeding and of infection.

As a member of the U.N.'s Administrative Committee on Coordination-Subcommittee on Nutrition (ACC/SCN), USAID could exchange information about their food surveillance and monitoring activities, which would help to coordinate their efforts not only with United Nations' agencies, but across bilateral agencies as well. USAID should investigate which agencies have most leverage in terms of food and nutrition activities in the NIS. This coordination would also be important if USAID is considering a cut or halt in funding of food and nutrition activities.

The World Bank has already shown considerable interest in the value of nutrition monitoring and surveillance data through its initial role in the design and financing of the RLMS and its current support for a similar survey in Kyrgyzstan. Given that the World Bank usually has well-established lines of communication with ministries of finance and planning, as well as health and agriculture, it may be in a particularly good position to encourage adjustments in economic policy to prevent nutritional deterioration during this period of rapid economic adjustment.

The evidence presented to the committee was too out-dated to permit definitive conclusions regarding the current food supply and nutritional status of vulnerable groups or the causes of food-related problems (e.g., a market food shortage versus inadequate purchasing power). There is, however, suggestive evidence of food insecurity, of infrastructural problems that could precipitate a deterioration in conditions, and of preexisting micronutrient deficiencies. The committee makes the following conclusions regarding Question 3:

• The decision to supply direct food aid should be based on evidence that the national food supply is inadequate. The exception is when direct food aid is carefully targeted to vulnerable groups who consume a relatively small proportion of the total food supply.

• Assistance with local food distribution and transportation problems may reduce market food insecurity.

• The nutritional status of the population may be as affected by problems in the delivery of other services as by changes in food supply and purchasing power. These problems might include failures in the social service network; supplies of necessities such as medicines, vaccines, appropriate infant foods, and nutrient supplements; and fortification of food with specific micronutrients.

• Structural changes are needed that may have longer-term impacts on the micronutrient status of the population. These changes include food fortification.

• There are opportunities for the support of NIS institutions that can take on the task of food and nutrition surveillance. Such support should be achievable at relatively low cost and have a high payoff.

• The currently high level of interest of bilateral and multilateral agencies concerning the NIS means that there are opportunities for USAID to leverage its food and nutrition monitoring assistance and to increase communication about NIS problems with other organizations.