PREreview of The long-term impact and effectiveness of rotavirus vaccination in Malawi: an interrupted time-series and case-control analysis

Brief summary

This preprint evaluated vaccine impact and effectiveness after introduction from January 2013 to December 2019 using data from children under 5 years hospitalised with acute gastroenteritis at Queen Elizabeth Central Hospital in Blantyre, Malawi.The researchers also tested whether switching from one type of oral polio vaccine (tOPV) to another (bOPV) changed how well the rotavirus vaccine worked. The main findings were: the vaccine gave 67% protection to infants under 1 year, protection dropped to 29% in older children; the polio vaccine switch did not affect rotavirus vaccine performance. There is one major limitation that the preprint only includes children from one urban hospital, so results may not apply to rural areas.

Strengths of the paper

1. Long follow-up period

The study evaluates rotavirus vaccine performance more than 7 years after introduction, providing evidence on long-term effectiveness. The analysis covers data from July 1997 to December 2019, which allows observation of trends well. This longer window may help capture changes in vaccine protection over time, such as waning immunity or shifts in disease burden to different groups, which shorter follow-up studies could miss.

2. Combining interrupted time-series and test-negative case control study designs

Interrupted time-series used to measure population-level impact and test-negative case-control used to measure individual-level effectiveness allow the findings to compare results across community and individuals. This combination helps assess both direct vaccine protection and broader herd effects. Using two designs together may also provide a way to check whether findings from one approach are consistent with the other, which could strengthen confidence in the conclusions.

3. Adjustment for confounders in the analysis

The authors used generalized linear regression models to adjust for seasonality trends. They also adjusted for changes in healthcare seeking and surveillance intensity by including test-negative diarrhoeal cases as a proxy for background admission trends. This approach helps reduce bias from external factors that are not directly related to vaccination.

Major concerns

1. Gap in the monitoring data (from December 2009 to November 2011).

This missing data may affect the accuracy of the interrupted time series model's estimation of what would have happened without vaccination, especially when it is unknown whether the epidemiology of rotavirus has changed during this gap period. The gap appears between two surveillance periods and is noted but not further assessed in the analysis. Readers might wonder whether rotavirus transmission patterns shifted during those two years.

2. Imprecise estimates for older children

In children aged ≥1 to <5 years, the vaccine effectiveness estimate was 29% with a 95% confidence interval ranging from −136% to 74%, indicating very low precision. This makes it difficult to draw reliable conclusions about protection in this age group. The authors mention this limitation but do not explore whether alternative analytical approaches could improve precision.

Competing interests

The authors declare that they have no competing interests.

Use of Artificial Intelligence (AI)

The authors declare that they did not use generative AI to come up with new ideas for their review.