Relative validity of interviewer-administered 24-hour recalls collected by telephone and in-person compared with weighed food records among rural Sri Lankan adults

Background Quantitative dietary assessment is hindered by cost, time, and logistical burdens. Collecting 24-hour recall (24HR) dietary data by phone may be particularly advantageous in rural areas.

Objectives To assess the criterion and relative validity and costs of phone compared with in-person 24HR compared with observed weighed food records (WFR), a reference measure for true intake.

Methods We used a randomized crossover design to collect phone and in-person 24HRs from 103 rural Sri Lankan adults, each matched to a WFR collected on the reference day. Agreement was assessed using equivalence testing with 15% error bounds and concordance correlation coefficients. Memory and portion size estimation errors were examined using descriptive statistics. Activity-based costing was used to estimate 24HR costs using expenditures and micro-costing.

Results Compared with WFR, respondents underreported energy intake by 263 kcal (12.6%) in phone 24HR and 229 kcal (10.9%) in in-person 24HR, primarily due to portion size misestimation. Mean intakes of protein and 6 micronutrients were equivalent between phone 24HR and WFR. Although no nutrients were equivalent for in-person recalls, the 2 recall methods had relative equivalence for energy, carbohydrates, fat, zinc, and niacin. The average concordance correlation coefficient was slightly higher for phone recalls, whereas memory error was lower for in-person recalls. Costs per respondent were $254 for in-person 24HR and $186 for phone 24HR in a most likely implementation scenario.

Conclusions Among rural Sri Lankan adults, phone-based 24HR were comparable to in-person 24HR in estimating nutrient intakes, although both methods underestimated true intakes. Phone-based recalls were less expensive and performed at least as accurately as those collected in-person.