Maternal morbidity and mortality associated with interpregnancy intervalReferee Review
This study examines the relationship between interpregnancy interval and maternal morbidity and mortality. It concludes that short (< 6 months) and long (>59 months) interpregnancy intervals are associated with an increased risk of adverse maternal outcomes.
Originality:
- Many studies exist describing the association between the interpregnancy interval and fetal and neonatal outcomes. Data on maternal outcomes are more limited and consists of four published studies. These have provided conflicting data.
- What distinguishes this study from previous work on the subject is the study’s large sample size (456 889 women) and detailed database, including data on multiple pregnancy-related disorders. This allows for a greater ability to find associations and correct for confounders than any of the four previous studies. The focus of most of the previous work has been on single outcomes such as maternal death or anaemia.
Clinical importance: Relevance to general readers
- Most recent work on the subject of the effects of various interpregnancy intervals has been published in obstetric or epidemiology journals, except for one study published in the New England Journal Medicine last year. This study is probably better suited for one of these journals. Nevertheless, the findings provide further support for recommendations emphasising the importance of birth spacing and methods to control this, and would probably interest the general reader.
- The subject matter is important and the paper contributes to existing knowledge. While the study does not offer any new insights on how interpregnancy intervals may be "optimised", it does provide a new angle to the messages supporting its importance, ie. that there are maternal consequences in addition to infant ill-effects.
Suitability for BMJ:I believe that the paper is scientifically sound and worth publishing. I have mixed feelings, however, about whether the BMJ is the most appropriate forum, but would support its publication.
Scientific reliability:
Research question
This is specified clearly.
Overall design:
The study retrospectively analyses data from a large database that has been in existence for over 12 years. The authors acknowledge some of the limitations of the database in the paper. Overall, the study design is adequate.
Participants studied:
While not clearly stated, about 88% of parous women delivering singleton infants whose data were recorded in the database were included in the study. The rest were excluded because of incomplete data. No further information is provided about these exclusions, e.g. were they more common in any particular age category or setting? This needs to be addressed.
Methods:
The study methodology is adequately described. Specific deficiencies are:
- Calculation of interpregnancy interval
Interpregnancy interval was defined as the time elapsed between the women’s last delivery and the last menstrual period (LMP) for the index pregnancy. The authors recognise the inaccuracy of using the last menstrual period to calculate this interval and state that the date of the LMP was "possibly amended by means of ultrasonography." However, they do not describe how often a correction was made using this device.Was there any measure of how certain women were of the LMP date they provided?
Further, it is unclear if the date of the last delivery was obtained from information of the event on the database (i.e. index pregnancy linked to previous birth) or from a history obtained from the pregnant woman. If the latter method was used, the record sheet appears to collect this piece of information as "month/year of end of previous pregnancy" and it would be difficult to calculate the interpregnancy interval in weeks using this data. (The authors claim to have computed the interval in weeks and converted it to months).
Finally, it would be helpful to state how two or more births to the same women, during the study period, was handled, i.e. were they regarded as individual events? (As an aside, it would be interesting to know how interpregnancy intervals differed for each birth in these women, and if any correlations could be made between adverse outcomes and length of individual pregnancies in these women).
- Repeated reference is made to pre-pregnancy weight, e.g. to calculate the body mass index. Surely, this is more correctly termed the "first prenatal visit weight" as no measurements were made on these women between pregnancies? Was this weight adjusted for the timing of the first prenatal visit? (I would think an essential step but very difficult to do
- Definitions used
While the ICD-10 was used to define adverse maternal disease events, this does not seem to have been applied in defining events such as fetal death where terms such as abortion and stillbirth are used. These terms do not appear in the ICD-10 classification system and the authors’ definition of terms such as stillbirths being fetal deaths over 20 weeks gestation need to be referenced or justified (since they are being used to compare outcomes in the different age groups). Were standard definitions of anaemia and uniform definitions of gestational diabetes applied at each of the participating centres?
Some definitions are unusual and need further explanation, e.g. marital status categorised into "did or did not live with infant’s father."
Statistical analysis:
This appears to have been done appropriately.
Results:
The results are well presented, answer the research question and appear credible. Minor corrections required are listed in the comments to the authors.
Interpretation and conclusion:
- Comparison with other studies
There is an attempt to compare the data to the scanty published literature. The criticism of a recent study from Bangladesh that failed to show an effect of interpregnancy interval on maternal mortality appears reasonable, but the identified deficiencies do not adequately explain the different findings. (Unfortunately, this reference was unavailable in our library).
- Explanation of findings
While the authors admit that the reasons for an association between a short interpregnancy interval and adverse maternal outcomes are unclear, they believe that "most of our findings may be explained by the maternal depletion hypothesis." While this is a popular explanation for the association, the study provides no data to support this particular theory, which suggests that short interpregnancy intervals do not allow mothers to recover their nutritional resources essential for a successful pregnancy. Indeed, body mass index, one proxy of nutritional status, was no different in the various interpregnancy categories. Unfortunately, the study does not attempt to offer alternative explanations for the noted associations, or at least, comment on their possible importance in this setting – other nutritional deficiencies, postpartum stress levels, socio-economic factors (e.g. household income), occupation, community variables (e.g. crime, housing) and other behavioural or psychological determinants.
- Limitations of study
a) The following limitations are explained:i) The study is hospital-based and not population-basedThe authors discount this being a confounder by claiming that the similarity of results between countries eliminates this possibility. I view this factor as being more important in limiting the generalisabilty of the findings. While enrolling women from different settings assists in reducing bias, that they were all hospitalised women makes it possible that they were more likely to have more serious pregnancy complications. Were women who qualified for entry into the database, i.e. attending particular clinics or hospitals where data was being collected, similar to the general maternity population in these regions?ii) Generalisabilty of findingsThe authors correctly state that their findings may not be generalisable to populations other than those coming from developing countries. However, it is unclear how generalisable the findings are even to developing countries. While the database includes women from a host of Latin American and Caribbean countries, there is no indication as to the socio-economic status and demographic features of the sample. Were, for example, most of these women poor and uneducated, or vice versa. A general statement summarising the salient characteristics of the population would assist in contextualising the findings.b) A limitation not identified by the authors, but important in establishing the comparability of the various interpregnancy groups would be the unavailability of data comparing each group for:- racial/ ethnic differences- breastfeeding duration
- contraceptive use
These have all been shown to affect interpregnancy intervals and may ostensibly affect the outcomes measured in this study
Implications of study findings:
The public health implications of the findings need to be more clearly described.
References:
The references provided are up-to date- and relevant.
Abstract:
The abstract accurately reflects what the paper says. Minor corrections are detailed in the note to the authors
Does the paper read well
The data, results and discussion are presented clearly and logically. However, the paper is fraught with grammatical errors. These should be easily correctable by a copy editor.