Methods of Nutrition and Health Survey ofBadghis Province, AfghanistanMarch 2002 :Methods of Nutrition and Health Survey ofBadghis Province, AfghanistanMarch 2002 Bradley A. Woodruff Meredith Reynolds U.S. Centers for Disease Control and Prevention Felicite Tchibindat Cyridion Ahimana Peter Salama UNICEF – Afghanistan Country Office


Background :Background More than 20 years of civil conflict Three years of drought Extensive population movements Loss of agricultural capacity Some evidence of serious malnutrition in north and west


Survey Objectives (I) :Survey Objectives (I) Assess children age 0-59 months for: Prevalence of acute malnutrition Clinical indicators of micronutrient deficiencies Especially anemia Assess women of reproductive age (15-49 years) for: Prevalence of malnutrition Clinical indicators of micronutrient deficiencies


Survey Objectives (II) :Survey Objectives (II) Assess households and household members for: Coverage of relief food distribution Water source and access Mortality rate Causes of death


Methods - Sample size :Methods - Sample size


Methods - Sample size :Methods - Sample size For nutrition outcomes: Need 534 children total 534 children / 1.3 children per household = 411 households 411 HHs / 0.75 response rate = 548 households 30 clusters of 18 households each


Methods - Sample size :Methods - Sample size


Methods - Sample size :Methods - Sample size For mortality: Need 640,940 person-days Recall period about 380 days 640,940 / 380 = 1687 persons 1687 persons / 6 persons per household = 281 households 281 households / 0.75 response = 375 households


Methods - Sample size :Methods - Sample size 534 children 411 HHs 548 HHs 411 women 1.3 children / HH 75% response 1 woman / HH 30 clusters of 18 HHs 2466 HH members (937,080 person-days) 6 persons / HH 380-day recall period


Methods - Cluster sampling :Methods - Cluster sampling First stage: Random selection of village (probability proportional to size) Second stage (if necessary): Random selection of mosque within village (probability proportional to size) Third stage: Random selection of 18 households within village or mosque


Methods - Cluster sampling :Methods - Cluster sampling


Methods - Cluster sampling :Methods - Cluster sampling


Methods - Cluster sampling :Methods - Cluster sampling Second stage (in villages with >200 households) List names of mosques and number of households belonging to each Create cumulative list of households Select one mosque probability proportional to size by using random number table


Methods - Cluster sampling :Methods - Cluster sampling


Methods - Cluster sampling :Methods - Cluster sampling Choose 3-digit random number between 1 and 219


Methods - Cluster sampling :Methods - Cluster sampling


Methods - Cluster sampling :Methods - Cluster sampling Third stage – selecting households Asked village leaders to help create list of households Confirmed that all current residents listed Numbered households Chose households with random number table


Methods – Follow-up :Methods – Follow-up If no one home, ask neighbors If family expected to return, check house 2 more times Do not replace if family not found or if family permanently gone


Methods – Interview :Methods – Interview Explain survey, obtain verbal consent Interview adult household member Displacement Water source Receipt of relief food List all household members at Eid Qurban last year (4-6 March 2001 in the Gregorian calendar) Collect on each person Age & sex Current status (alive here, alive elsewhere, dead, missing) Recall period 376 – 392 days


Methods – Micronutrient examination (I) :Methods – Micronutrient examination (I) Women of reproductive age Signs of micronutrient deficiency Night blindness Examine for goiter


Methods – Micronutrient examination (II) :Methods – Micronutrient examination (II) Children <5 years of age Vitamin A Night blindness Bitots spots Vitamin C Gum bleeding Perifollicular hemorrhage Painful, swollen joints Bruises on legs Vitamin D Row of ricketts Non-painful, swollen joints Bowed legs Spinal deformity Anemia Palmar pallor


Methods – Anthropometric measurements :Methods – Anthropometric measurements Children <5 years of age Weight Height/length Edema Women of reproductive age Weight Height Pregnancy status Mid-upper arm circumference (MUAC)


Methods – Anthropometric measurements (II) :Methods – Anthropometric measurements (II) Measuring weight Measuring height


Methods – Anthropometric measurements (III) :Methods – Anthropometric measurements (III) Measuring length


Methods – Physical examination :Methods – Physical examination


Data collection form – HH census :Data collection form – HH census


Data collection form – Verbal autopsy :Data collection form – Verbal autopsy


Problems (I) :Problems (I) 4 separate units of analysis Household variables All household members from census Women of child-bearing age Children <5 years of age Required 4 datasets Complex relational data entry Complex relational data analysis


Problems (II) :Problems (II) Second and third stage sampling was somewhat complex Survey teams had no sampling or survey experience Learned to do sampling in 4-day training Poor security in 2 selected villages Non-randomly selected 2 nearby villages


Conclusion :Conclusion Feasible to carry out simple random sampling in villages Many feared that anything but EPI method is impossible These fears unfounded Naïve but intelligent personnel can be trained in sampling methods relatively quickly Sampling and listing in each village took 0.5 – 2 hours Computer problems with relational data entry