The authoring team identified two numerical discrepancies in the adjusted odds ratios reported in the abstract, Table 8, and conclusions of the original article (Table 1), detected while reanalyzing the study dataset in the context of a follow-up study (Table 2).

The discrepancies affect two variables in the multivariable logistic regression model; the two remaining main findings of the article — delay in access to care exceeding two hours, and lack of recognition of warning signs by the birth companion — are confirmed in the same direction.

For the variable history of anemia, the raw distribution observed in the analytical dataset (13/49 cases, 26.5% vs. 58/97 controls, 59.8%) is arithmetically compatible only with an odds ratio in a protective direction. The corrected value of 0.16 indicates that women with a documented history of anemia had a lower probability of maternal death compared to those without such a record. This finding is interpretable as survivorship bias in a study design using near-miss controls: women who survived severe obstetric complications — the controls — have more documented medical history precisely because prior contact with the health system contributed to their survival, a pattern described in the international literature [1,2].

For the variable no partner, the corrected value of 0.47 does not reach statistical significance (p = 0.219), and therefore no association between this variable and maternal mortality can be established in the multivariable model.

The corrected findings do not alter the main conclusions of the article regarding delayed access to medical care as a risk factor for maternal mortality.