Reader Comments

Post a new comment on this article

Maternal mortality in Mozambique: Methodologic issues

Posted by plosmedicine on 31 Mar 2009 at 00:20 GMT

Author: Adamson Muula
Position: No occupation was given
Institution: Department of Community Health, University of Malawi, College of Medicine, Blantyre, Malawi
Submitted Date: February 22, 2008
Published Date: February 22, 2008
This comment was originally posted as a “Reader Response” on the publication date indicated above. All Reader Responses are now available as comments.

Menéndez et al [1] must be commended for their work in identifying the causes of maternal deaths using autopsy findings in Mozambique. This should be the beginning of further studies in Mozambique and elsewhere, especially in the developing world, for us to have better understanding on maternal mortality. Despite the studies strength, there are several significant limitations in the methodology that deserve highlighting and possible clarification from the authors.

Firstly, the authors report that the study was conducted at Maputo Central Hospital between October 2002 and December 2004 but do not report how the women were recruited. Were these consecutive women attending prenatal care at the hospital or another criterion was used? This is important to know, because if the study recruited consecutive women attending care, Table 1 betrays any such criterion. In Table 1 in the category “Place and type of delivery” there were 9 deaths for women with out-f hospital delivery and 21 deaths among women of unknown place of delivery. Were these women those who had prenatal care at Maputo Hospital but delivered elsewhere? Or were these deaths among women who did not attend Maputo Hospital but either died or were brought in dead at the hospital? If these are deaths occurring elsewhere, then including them in the calculation of maternal mortality ratio (MMR) for deliveries occurring at the hospital may be misleading. This is complicated by the fact the live births were obtained from “All births occurring at the hospital during the study period” and yet the group of women from which deaths were reported did not have to deliver at the hospital.

The authors’ definition of a maternal death was “all women dying during pregnancy or within 42 d of completion of a pregnancy, irrespective of the cause of death, and for whom the family had given verbal (oral) informed consent.” The authors’ clarity in presenting the operational definition they used must be recognised and commended. However, it has also observed that such definition renders these findings difficult to compare with other studies as other authors would add to the definition an exclusion of deaths occurring for reasons not related to pregnancy or its management. In Menéndez et al’s definition, a traumatic (injury) death may have been included in the calculation of MMR if this had occurred among the women studied while other authors would have specifically excluded such deaths. Luckily though, such deaths did were not reported by Menéndez. The causes of deaths identified as “non-obstetric” may be removed from the calculation of MMR if researchers believe that these deaths may have occurred anyway irrespective of the pregnancy status (positive) of the woman.

I found it an important omission in the methods and discussion sections the fact that not all tests were carried out in all decedents. For example, malaria examination was carried out in less than half (61/139) of the decedents and yet the authors did not highlight the possible biases such experience may have had on the results. These unknown may have depressed the proportion of malaria deaths if there were some women not examined for malaria but had severe malaria.

The authors have reported that “Severe malaria was more frequent in women living in Maputo centre or periphery (2/16 [12.5%] and 9/63 [14.3%], respectively) than in women living in rural areas (1/28 [3.6%]) (p = 0.18). Evidence of active or past malaria was identified in 11/16 (68.7%) women living in Maputo centre, 44/63 women living in Maputo periphery (69.8%), and 23/28 (82.1%) women living in rural areas (p = 0.44).” There are at least two problems with the authors report. Firstly, the authors have used percentages for very small observations e.g. 1/28. To the naïve reader, such percentages may relay a lot of information when they should not. Secondly the authors report of a difference in frequency of severe malaria by location and yet their p value reported is non-significant (i.e. 0.18) at alpha 0.05.

The observations outlined above may need clarification if future studies are to be compared to Menendez findings.

Menéndez C, Romagosa C, Ismail MR, Carrilho C, Saute F, et al. (2008) An Autopsy Study of Maternal Mortality in Mozambique: The Contribution of Infectious Diseases. PLoS Med 5(2): e44 doi:10.1371/journal.pmed.0050044

No competing interests declared.