Erratum: Accelerated weight gain, prematurity, and the risk of childhood obesity: A meta-analysis and systematic review (PLoS ONE (2020) 15: 5 (e0232238) DOI: 10.1371/journal.pone.0232238)

Abstract

After this article [1] was published, the authors identified errors in Fig 3. In Fig 3B, the x-axis labels “Favors preterm SGA” and “Favors preterm AGA” are swapped. The left label should be “Favors preterm AGA” and the right label should be “Favors preterm SGA”. Also in Fig 3B, there is an error in the reported aOR (95% CI) for Gaskin 2010. This error also resulted in incorrect % weight values, the I-squared statistic and p value, and aOR (95% CI) overall. The authors provide a corrected Fig 3 here. In the second sentence of the Research question 2 subsection of the Results, the correct sentence is: “The result of meta-analysis revealed no significant difference on childhood obesity between SGA and AGA infants (adjusted OR = 1.03; 95% CI [0.69, 1.53]; p = 0.107; Fig 3B)” In addition, after publication of this article [1], concerns were raised about the use of the term “impact”, as this incorrectly implies a causal relationship. Therefore, “impact of. . .on” is corrected throughout the article to “association between. . .and”. The authors apologize for the errors in the published article, which do not affect the Conclusions. Some concerns were also raised about possible biases in the statistical analyses. Concerns were raised that confounding effects may have introduced bias and may not have been adequately discussed. The authors provide additional information to Table 1 here about the covariates/confounders that were adjusted and reported in each original study selected. Concerns were also raised about the lack of sensitivity analyses included in the study. The authors provide an additional S1 Table to summarize the sensitivity analysis using the “leave-one-out” approach, below. The authors provide the following clarifications discussing the results of the sensitivity analysis, which does not affect the original conclusions. It was noted that in the meta-analysis assessing the association between preterm status and childhood obesity (Fig 3A), one study (Mardones et al. 2008) [2] contributed 97% of the data. The authors clarify here that during the sensitivity analysis, as shown in S1 Table, the pooled estimate shifted to non-significant (OR = 1.213, 95%CI: 0.884, 1.663, p = 0.232) upon excluding Mardones et al. (S1 Table). Accordingly, readers are advised to exercise caution when interpreting the conclusion of this meta-analysis on the difference between preterm vs term infants in association with childhood obesity, as the pooled estimate is substantially influenced by the study of Mardones et al. It was also noted that in the meta-analysis assessing the association between preterm SGA (vs. AGA) and childhood obesity (Fig 3B), one study (Ramírez-Vélez et al.) [3] that controls for later weight contributed the most data to the pooled estimate. The authors clarify here that during the sensitivity analysis, as shown in S1 Table, the pooled estimate remained similar to the original estimate (OR = 0.932, 95%CI: 0.416; 2.089, p = 0.865) upon excluding Ramírez-Vélez et al., suggesting the meta-analysis outcome is robust (S1 Table).