8/13/2019 International Journal of Epidemiology

1/3

-International Journal of EpidemiologyInternational EpidemiologicalAssociation 1943

Vol. 22. No. 2Printed in Great Britain

Risk Factors for Clinical Marasmus:A Case-Control Study ofBang adeshi ChildrenFITZROY J HENRY,* * ANDR BRIEND,*.'VlNCENT FAUVEAU,* SHARON R A HUITLY .fMOHAMMED YUNUS* AND JYOTSNAMOY CHAKRABORTYIHenry F J (Harvard Institute for International Development, Cambridge, MA02138, USA), Briend A, FauveauV, HUnlYS A, Yunus M and ChakrabortyJ. Riskfactors or clinical marasmus: A case-control t u d y of Bangladeshi children.International Journalof Epidemiology 1993; 22:278-283.A cade-control study of risk factors of clinical marasmus was undertaken to guide intervention efforts in ruralBangladesh. Cases were children whose mid-upper arm circumference measured 120 mm. Between June 1988 and June 1989, 164such pairs ofchildten aged 1-4years were studied. The effects of various demographic, socioeconomic, environmental, and healthfactors, reported by mothers, were investigated in a multivariate analysis using conditional logistic regression. Resultsshowed an increased risk of marasmus among children from families with other children under 5 years of age (oddsratid IORI = 2.51; 95 confidenceinterval [CI]: 1.33-4.741, and children who consumed formula foods (OR = 16.41,95 CI 3.39-79.36). Higher maternal education was associated with reduced risk of marasmus, compared with noeducation, the OR for

8/13/2019 International Journal of Epidemiology

2/3

280 INTERNATIONAL OURNALOF EPIDEMIOLOGYtheir plausibility and utility as ndicators of marasmus,their role as confounding factors, and their distribu-tion in the population. Fourteen risk factors wereselected for inclusion in the final analyses, these arelisted n Tables 2 and 3. Although some of these 14 fac-tors are interrelated, no pair of factors was highly cor-related maximum correlation 0.46 and most were

8/13/2019 International Journal of Epidemiology

3/3

282 INTERNATIONAL JOURNAL OF EPIDEMIOLOGYascertainm ent also conducted he interview. We founda high level of concordan ce between the sample ofsupervisors nterview reports (performed blind) whencompared with those of the community healthworkers. This was true for both cases and controls.lso arguing against bias was the fact th at most of the

variables were objective and verifiable or were col-lected directly fro m the record-keeping system of theMatlab demographic surveillance system. That onlyone of the o ther matern al-report ed variables wasstatistically significant suggests that such selectivemisreporting would be most unlikely.

The results show that th e presence in the family of asibling