Objectives: This study investigates the effect of the recall period on the accuracy of recall of the out-of-pocket payments (OOPs) and timing of inpatient transactions in Vietnam using household surveys and hospital records.

Methods: 1229 households were identified using hospitalization records and interviewed with either a 12-month or a 6-month recall period. The Bland-Altman method was used to assess the agreement between household and provider OOPs for the whole sample, and by amount of provider OOPs (low- ≤ USD 8.8; high - > USD 8.8). Multinomial regression was performed to assess the effect of the recall period on whether a transaction was recalled correctly in expenditure and time.

Results: Households reported higher OOPs than hospital overall. No evidence of an effect of recall period on OOPs in the whole sample, but there was an interaction between recall period and the level of provider OOPs. For the low OOPs, there was substantial variability in the ratio of household to provider OOPs. For the high OOPs, the mean ratio for the 12-month period was 1.13 (95% limits of agreement 0.03, 37.5) and for the 6-month was 1.7 (0.09, 33.7); p=0.009. About 60% of transactions were reported in the correct recall period. Overall, and for the low OOPs, there was no evidence of an effect by recall period, whereas for the high OOPs, transactions of the 6-month period were 5.9 (95% CI 2.7, 13.2) times more likely to be forward telescoped and less likely to be forgotten (RR 0.5, 0.3-0.9).

Conclusions: The study provides evidence of the drivers of higher mean OOPs in the 6-month compared to the 12-month recall period. The 12-month suffered less from over-reported amounts or forward telescoping, but had a higher risk of forgetting. Impact of recall period on how accurately a transaction was reported differed by OOP amounts.