Tamil Nadu and child nutrition : a new assessment

NUMBER 87 E9,E Precis Operations Evaluation Department April 1995 Tamil Nadu and Child Nutrition: A New Assessment India's Tamil Nadu Integrated Nutri- 9 Food taken home was shared with can be important. Growth monitor- tion Project, covering a rural popula- other family members, reducing the ing-monthly weighing of all chil- tion of more than 13 million, is one of impact on beneficiaries. dren 6-36 months old-was a critical the world's largest projects for nutri- - The food given was too coarse and element of this strategy. It was the tion education and targeted supple- bulky for very young children to eat. way in which interventions were tar- men tary feeding. Though it did not o Not enough emphasis was given geted only to problem cases, thereby achieve all its goals, it is one of the to nutrition education for mothers, controlling program costs. It was most successful efforts to date to reduce nor to the complementary health care also an important educational tool, to severe malnutrition. While much has interventions needed to improve explain to mothers why one child been written about this project, an im- their nutritional status, was receiving food and another not, pact evaluation by OED extends previ- and to provide them with objective ous analyses in three respects.* It re- Project goals, approach feedback about how they were car- views progress since 1986, the date of ing for their children. Indeed, this the last large-scale evaluation; it uses a As a result, the government project was the first large-scale use of large representative sample of indi- in 1980 launched the Tamil Nadu In- growth monitoring for this purpose. vidual beneficiaries for its analysis; and tegrated Nutrition Project it distinguishes the influences of other (TINP I). The project sought better The project's supplementary feed- determinants of nutrition and health targeted and more cost-effective ing component was also innovative. in the area. ways to improve maternal and child It focused on feeding very young nutrition and health. It covered the children-who are the most vulner- By the second half of the 1970s, In- rural areas of those districts with the able to poor nutrition-for relatively dia had a variety of public feeding worst nutritional status-about half brief periods to help them recover programs. Malnutrition was particu- the state, and a rural population of their growth, in contrast to the more larly severe in the state of Tamil about 9 million. Its total cost of $81 common approach of prolonged Nadu, despite the 25 different nutri- million (originally estimated at $66 feeding of older children. It relied tion programs operating there which million) was supported by an IDA heavily on local nutrition workers, were costing the government about credit of $32 million. The credit be- $9 million annually. Evaluation stud- came effective in 1980 and closed in ies showed that these programs were 1989, two years later than planned. Tact au aten ni- reaching only a small fraction of the most vulnerable groups: TINP I sought to improve the nu- t ec r 4 Ron N . 37, tritional and health status of pre- Rdcer OED bmac RevaldatGo Because the programs were not school children, primarily those rert ae imact e an targeted on the basis of nutritional 6-36 months old, and pregnant and e aresan tank criteria, they did not reach their in- nursing women. A central tenet of exomutie Irers andumtaff tended beneficiaries. Children most the project was that most malnutri- frmt nte nal c nts at risk were not identified. tion is the result of inappropriate mnitn fromcRegCnan o * Feeding on site tended to replace child care practices, and not of in- the public from the Bank's meals that beneficiaries would other- come, famine, or unpreventable wise have eaten at home, health problems, though these factors growth was found to be faltering, TINP II education for diarrhea management, administration of Vitamin A, periodic The goals of TINP II include more attention to coordination deworming, and supplementary feed- reducing severe malnutrition between health and nutrition service ing of a limited number of women. among 0-36 month old children- delivery. by 50 percent in new areas and by Growth monitoring and nutrition in- 35 percent in TINP I areas--and Most of these changes are positive, puts were adequate. Although the en- helping to reduce infant mortality to resulting directly from lsons rollment of 6-36 month old children 55 per thousand live births and to learned in TN? L But two are ques- was less than desired (about 77 per- halve the incidence of low birth tionable. The most problematic is the weights. relaxation of entry and exit criteria for cnt lely bece fili in outcy- supplementary feeding. This modifi- inge hamle weiren' ificlttrah), Design features that distinguish cation was apparently a t o nce erle chilre's m n TINP II from TINP I include: the limited success of TINP I in re- ducing moderate malnutrition, systematic. In 1986, they reached * extension of the target age group to Though that apparent lack of Success 82 percent of the possible maximum, 0-60 months (from 6-36 months in may have been due to overly ambi- a figure that compares very favorably TINP I); tious project goals, it is not clear that with other efforts to use growth moni- * introduction of a new mother- the best way to deal with moderate toring. Other indicators suggest that linked child health card to establish a malnutrition is to provide more the procedures laid down-for ex- link beh% een births and TINP enroll- food-better health services or more ample for beginning and ending ment; intensive education might be more supplementary feeding, providing * maternal growth monitoring; cost effective. Investigating the ta educational inputs, encouraging cor- * inclusion of non-formal early sons for the shifts in nutritional status munity participation, and keeping ac- childhood education; and undertaking operations reerh curate records-were carefully fol- * expansion of services to to determine the best way to correct lowed. Sharing of supplemental food more children in different stages the situation (assuming it needed cor- with non-participants and substitu- of malnutrition; rection) might have been preferable tion for food received at home was probably small. locally trained, working in conjunc- used a large representative sample, tion with local women's groups. attempted to control for other factors, Health interventions worked much less and quantitatively investigated ques- well. Except for immunization, the An eight-year extension of the tions such as who benefitted and project did not achieve its target levels project (TINP II) is now in progress, which inputs were crucial to the out- of implementation. Record keeping supported by an IDA credit of $96 come, in order to understand why was less reliable than for the nutrition million approved in 1991. TINP 11 this program has succeeded where interventions and procedures were seeks to reach most of the remaining others have failed. not followed as systematically. rural population in Tamil Nadu. (Box 1.) The impact evaluation focuses The data were analyzed for three mainly on the nutrition and health points in time-1982, 1986, and 1990. Why another study? impacts on children. It relies mainly While all service delivery indicators on statistical analyses of a large rep- improved between 1982 and 1986, Many have doubted the practical- resentative sample of service records several of them deteriorated between ity of a nutrition and health program for 1982-90 from some 9,000 commu- 1986 and 1990. This peak and decline based on growth monitoring and nar- nity nutrition centers created by the pattern most likely reflects a decline rowly targeted short-term supple- project. Project monitoring data, a in performance resulting from low mental feeding. Before TINP, there field visit, interviews, and a review of morale, which was created by uncer- had been no successful, large, sus- the literature were also used. tainties in 1989 and 1990 about tained application of these principles. whether and in what form the project That is why a careful assessment of Findings would continue. Once the follow-on TINP, based on detailed quantitative project, TINP 11, started, performance documentation of its achievements, is Service delivery was partial but appeared to pick up. so important. highly beneficial Malnutrition levels fell significantly Before this impact evaluation, such TINP I provided a package of ser- documentation was not available. vices: nutrition education, primary The data indicate a statistically There have been good qualitative as- health care, supplementary on-site significant improvement in weight- sessments and limited quantitative feeding of children who were se- for-age during 1982-90. They also studies, but none has simultaneously verely malnourished or whose indicate a steady drop in malnutri- April 1995 tion rates for all ages included (an administration unit covering a TINP increased its capacity to enroll in the sample. These improvements population of about 100,000), and the most needy. were continuous over time; they information from other large-scale did not follow the peak and decline nutrition interventions. Who benefitted? pattern found in the service deliv- ery indicators. In addition, an estimate of the ef- Nutrition status in Tamil Nadu fect of income changes on nutrition improved across the board, regard- Much of the change can be attributed was made for the TINP areas, using less of gender, caste, age groups, to TINP an elasticity measure estimated from or income levels. There is some other studies. In each case the com- indication that lower caste individu- The nutritional improvements parisons are favorable to TINP. als and higher income groups observed can be attributed to TINP benefitted more. if they can be distinguished from Roughly averaging the figures the effects of other factors that together suggests that half to three Some inputs contributed more than oth- might also explain the improve- fourths of the decline in malnutrition ers to improved nutrition ments. This was attempted in two in TINP areas was due to TINP stages, first by comparing changes and other nutrition programs in Multivariate analysis suggests that occurred within TINP districts those areas. Although the NMP that frequency of weighing, younger with changes in non-TINP areas, operated in all districts in which age of enrollment, and immunization and second, by considering the dif- TINP operated, NMP does not ac- were statistically significant while ferential impact of other programs count for much of the observed im- deworming and Vitamin A inputs operating within TINP areas-the provement in nutrition within the were not. one of interest here being the Nutri- 6-36 month age group. One reason tious Meals Program (NMP). is that only children 24 months The project was cost effective and older are eligible. Unfortunately, there are no pure Available evidence suggests that control areas that could be compared Who participated? TTNP I had lower costs per capita and with TINP areas. Instead, the evaluation per beneficiary than nutrition projects relied on aggregate data on nutrition Slightly more boys participated in that are not narrowly targeted. status gathered for other purposes in TINP than girls, the participation of nearby districts, data for TINP areas for caste children increased from 37 to 42 The recurrent cost per beneficiary the period before TINP was imple- percent, and the average age at en- of TINP has been estimated at about mented, baseline data available at the rollment declined over time. The last $9.50 a year. This compares favorably start of TINP in each set of blocks two of these findings suggest that with the cost of the nutrition compo- Changes in nutrition profile of children in TINP areas, 1982-90 Figure 1 shows malnutrition prevalence rates as standard deviations from the reference medians of weight-for-age maintained by the Indian National Center for Health Statistics. Figure 2 shows the same data in terms of percentage deviations from these standards. Both figures show that the percentage of children seriously malnourished has declined while the percentage modestly malnourished and normal has increased. 1. Percent children above/below -2 standard deviations 2. Percent children above/below 75 percent of median (SD) of median weight-for-age (woa) weight-for-age (w/a) 1000_ ins TIPara0asdetoTN S60 60 40- 40- 20 20 0 0 1982 1986 1990 1982 1986 1990 Year in TTNP program Year in TINP program o below 60% of NCHS median w/a 0 60-75% of NCHS median w/a below -25D of median wfa o] above -2oD w/a E] above 75% of NCH median w/a OD Prcis nent of the national Integrated Child phased out without loss of nutri- planned-keeping costs down, Development Services (ICDS) pro- tional and health status? Only a few reducing dependence on feeding, gram. TINP's costs per village signs are available-knowledge and and, along with growth monitor- dweller were a little over half the attitude surveys, extent of breast- ing, serving as a powerful educa- costs of ICDS (excluding this feeding, and percentage of children tional tool. program's education components) requiring feeding-but they are all and TINP had roughly twice the ef- moving in favorable directions. All this can be accomplished fect on severe malnutrition. Com- with acceptably small leakages and paring program costs per benefi- Evidence from a TINP Il baseline costs. ciary (women and children), ICDS study shows that mothers who par- 9 Universal feeding is not neces- cost 25 percent more than TINP for ticipated in TINP I are significantly sary to achieve nutritional and half the benefit. These estimates more aware of good nutritional and health gains. have been questioned because they health practices than those who have * Most important, the education were derived from a study that used not participated. Signs of behavioral provided can induce permanent a small sample, but they are the best changes among these mothers in- changes in mothers' behavior that available. It should be no surprise dude their longer/higher rates of positively affect their children's that a program that helps only the breast feeding and lower percentage health and nutritional status. This most needy is cheaper and more of children requiring feeding. It is may take more time than originally effective than one that aims to help also encouraging to observe, in a vil- thought, but once made the all regardless of need. lage that was one of the earliest to es- progress is unlikely to be reversed. tablish a community nutrition center, Is the project sustainable? that former participants in the pro- Importance of processes gram have formed an "adolescent Financial sustainability seems as- working group". The key to TINP's success sured. Under the TINP II agreement, has been the great care exercised the government of Tamil Nadu is fi- Of more solid significance would in planning and executing its nancing all the operating costs of the be evidence of improvements in the processes: program in TINP I areas. These nutritional and health status of costs were estimated to be less than younger siblings of children who careful selection and training of 5 percent of the Tamil Nadu were once in the program, and in the community nutrition workers; government's expenditures on nu- status of new mothers and their chil- o detailed work routines; trition in 1988-89. dren. While some field data now * heavy emphasis on supportive exist to explore these possibilities, supervision and on the job training; Institutional sustainability is more time and budget constraints pre- 9 efforts to gain community sup- difficult to assess. The project appears cluded such analysis within the port; to be operating efficiently. The pro- scope of this impact evaluation. e emphasis on accurate monitor- gram took less time to achieve full ing; and operational status in new areas un- Conclusions * use of the data gathered in der TINP II than under TINP I. trouble-shooting and feedback. Moreover, some dimensions of the The findings of this impact evalua- project are now being operated with tion confirm the practicality of large- Such attention to detail is not more permanent institutional ar- scale nutrition and health programs typical in such projects and may re- rangements than before. The main based on growth monitoring and flect the fact that TINP I was owned, case in point is the establishment of narrowly targeted short-term supple- developed, and executed by local a permanent Communication and mental feeding. They confirm that: authorities, rather than superim- Training Center for the project. posed from afar. The areas where e Women can be induced to bring TINP I failed-implementation of The most important question is their children in for weighing on a an effective health program and in- whether people have permanently regular basis. tegration of the nutrition and health changed their behavior. Are women * Short-term feeding based on nar- components-were precisely those internalizing TINP's messages so row targeting with clear entry and where less attention was paid to that the program can eventually be exit rules can be made to work as process issues. OED Preis is produced by the Operations Evaluation Department of the World Bank to help disseminate recent evaluation findings to development professionals within and outside the World Bank. The views here are those of the Operations Evaluation staff and should not be attributed to the World Bank or its affiliated organizations. Please address comments and enquiries to the managing editor, Rachel Weaving, G-7137, World Bank, telephone 473-1719. Internet: rweaving@world bank.org April 1995