Preconception care for diabetic women for improving maternal and fetal outcomes: a systematic review and meta-analysis

Background Preexisting diabetes mellitus is associated with increased risk for maternal and fetal adverse outcomes. Despite improvement in the access and quality of antenatal care recent population based studies demonstrating increased congenital abnormalities and perinatal mortality in diabetic mothers as compared to the background population. This systematic review was carried out to evaluate the effectiveness and safety of preconception care in improving maternal and fetal outcomes for women with preexisting diabetes mellitus. Methods We searched the following databases, MEDLINE, EMBASE, WEB OF SCIENCE, Cochrane Library, including the CENTRAL register of controlled trials and CINHAL up to December 2009, without language restriction, for any preconception care aiming at health promotion, glycemic control and screening and treatment of diabetes complications in women of reproductive age group with type I or type II diabetes. Study design were trials (randomized and non-randomized), cohort and case-control studies. Of the 1612 title scanned 44 full papers were retrieved of those 24 were included in this review. Twelve cohort studies at low and medium risk of bias, with 2502 women, were included in the meta-analysis. Results Meta-analysis suggested that preconception care is effective in reducing congenital malformation, RR 0.25 (95% CI 0.15-0.42), NNT17 (95% CI 14-24), preterm delivery, RR 0.70 (95% CI 0.55-0.90), NNT = 8 (95% CI 5-23) and perinatal mortality RR 0.35 (95% CI 0.15-0.82), NNT = 32 (95% CI 19-109). Preconception care lowers HbA1c in the first trimester of pregnancy by an average of 2.43% (95% CI 2.27-2.58). Women who received preconception care booked earlier for antenatal care by an average of 1.32 weeks (95% CI 1.23-1.40). Conclusion Preconception care is effective in reducing diabetes related congenital malformations, preterm delivery and maternal hyperglycemia in the first trimester of pregnancy.


Background
Diabetes mellitus (DM) is a global public health problem with expected 300 million diabetics by the year 2030 worldwide [1]. In many areas around the globe including the West as well as many developing and Middle Eastern countries, diabetes has become a major health burden affecting young adults and women in their reproductive age [2,3].
Despite improved access and quality of antenatal care, women with pre-gestational diabetes and their fetuses are at increased risk of developing serious complications compared with the non-diabetic pregnant women, including spontaneous abortion, preterm labor, hypertensive disorders, and delivery by cesarean section [4,5]. In the recent report of The Confidential Inquiry into Maternal and Child Health (CEMACH) from England, Wales and Northern Ireland, the perinatal mortality in mothers with type 1 and type 2 DM is four times higher and the risk of congenital malformation in the babies of women with diabetes is nearly three times greater [4]. Similar reports from North America showed no significant improvement in fetal and neonatal outcomes of women with pre-gestational diabetes between 1988 and 2002 [6] despite the Saint Vincent Declaration in 1989 which sets a healthcare goal to improve the outcome of pregnancies in diabetic women [7]. Similar reports from the Middle East showed higher rate of perinatal mortality in diabetic as compared to non-diabetic women [8].
Many of the complications of DM during pregnancy can be prevented by optimizing maternal health in the preconception period. Glycemic control is one of the most important aspects of preconception care (PCC) [9]; however other aspects such as folic acid supplementation, smoking cessation, screening and treatment of diabetes complications and discontinuing teratogenic medication, are as important for improving maternal and fetal outcomes [10].
We carried out a systematic review to assess the effectiveness and safety of PCC in improving maternal and fetal outcomes for women with preexisting type 1 or type 2 DM.

Type of studies
We included in this review randomized trials (including cluster and quasi randomized studies) and cohort and case control studies, comparing the frequency of maternal and fetal adverse outcomes in diabetic women who received PCC with those who did not receive PCC.

Type of participants
Women of reproductive age with preexisting type 1 or type 2 diabetes mellitus who were not pregnant at the time of intervention.

Type of intervention
For the purpose of this review PCC is defined as the following either as sole intervention or in combination 1. Glycemic control by insulin and/or diet aiming at fasting blood glucose ≤5.7 mmol/l or/and postprandial blood glucose ≤7.8 mmol/l and/or glycosylated hemoglobin A (HbA1C) ≤7.0%) 2. Women counseling and/or education about diabetes complications during pregnancy, the importance of glycemic control and self monitoring of blood glucose level. 3. Preconception screening and treatment of complications of diabetes 4. The use of contraception until optimization of glycemic control is achieved 5. Intake of multivitamin or folic acid in the preconception period.
Type of outcome Maternal outcomes 1. HbA1C level in the first trimester.
2. Gestation age at the time of the first visit to antenatal care clinic (booking visit). 3. Pregnancy complications including spontaneous abortion, termination of pregnancy due to congenital malformations, polyhydramninos, pre-eclampsia, preterm delivery (before 37 completed weeks from the last menstrual period) and induction of labour due to complication of diabetes. 4. Delivery by cesarean section or instrumental delivery. 5. Maternal hypoglycemia in the first trimester or any other adverse effect reported by the authors.

Exclusion criteria
We excluded from this review reports which are not of comparative design and reports of conference proceedings or abstracts when there is no complete description of the trial or study.

Search strategy
The search strategy was developed in consultation of an information retrieval specialist. We searched the follow- We reviewed the reference list of all relevant studies for any potential study not retrieved by the search strategy. Unpublished reports were not actively sought and there was no language limitation.

Identification of included studies
All titles and abstracts retrieved by the electronic search were screened independently by three reviewers and the studies which clearly did not meet the inclusion criteria were excluded. Copies of the full text of potentially relevant studies and trials were obtained and their eligibility was assessed independently by two reviews. Differences between reviewers were resolved by discussion or by consulting a third reviewer.

Data extraction and studies assessment
Three authors extracted data from the included studies using a designed form. The accuracy of the extracted data was checked by two other reviewers.
The Newcastle Ottawa Scale (NOS) was used for the assessment of cohort, case control studies and non-randomized trials [11]. Risk of bias in each study, was assigned according to the number of items on the NOS judged to be inadequate. We considered low risk of bias when one item is inadequate, medium risk of bias when up to three items are inadequate and high risk of bias when more than three items are inadequate. Data analysis was carried out with the use of Review Manager Software 5.0 (Cochrane Collaboration, Oxford, United Kingdom).
Meta-analysis was performed for studies with similar design and type of intervention, which we assessed to be at medium or low risk of bias using the fixed effect model. Heterogeneity is considered high when I 2 >50% and explanation was attempted however subgroup analysis was not possible in most of the cases due to the small number of studies. Pooled data were presented as risk ratio (RR) with 95% confidence intervals (95% CI) for dichotomous outcomes and as the means difference with 95% confidence intervals for continuous outcomes.

Results
The search retrieved 1612 potentially relevant titles of which the full papers of 44 relevant reports were reviewed ( Figure 1). A total of 24 reports of 20 studies were included in this review [10,. (Three articles described the same cohort study with two interim [17,18] and one final report [19], one study reported the outcomes for the same cohort in two articles [10,29] and two articles report the outcomes of one cohort with one interim [31] and one final report [28]).
Twenty studies were excluded, 16 of them were excluded because they did not meet the inclusion criteria, 2 reports were of conference proceedings and in 2 studies data were not extractable (Additional file 1: Appendix 2).
Of the included studies, only one was a controlled trial, 11 studies were prospective cohort studies, 7 studies were retrospective cohort studies and one was a case control study (Tables 1-4).

Assessment of the methodological quality of the included studies
The cohort studies included in this review (Table1 & 2) had adequate description of participants including description of some confounding factors such as the frequency of renal and vascular complications of diabetes between the PCC group and the control group. However all studies did not address the effect of the presence of confounding factors on the outcomes except for 2 reports which used regression analysis to evaluate the effectiveness of the PCC [10,29].
In most of the cohort studies blinding of the control group was adequate because they were recruited after pregnancy when they attended for antenatal care, except for 2 studies [24,28], in which inadequate blinding of the control group cannot be excluded because they were informed about the importance of the PCC and were invited to attend. All participants received the same antenatal and post natal care except for one study [24] where participants were followed up in different health settings.
All cohort studies had adequate follow up for participants except for one study in which 52% of the PCC group were lost to follow up [26]. The assessors of the outcomes were not blinded to the participants' allocation except in one study [25].
Some of the studies at high risk of bias were initially designed to assess aspects of PCC other than its effectiveness in improving maternal and fetal outcomes, hence the poor methodological design when assessed with the NOS [16,22,30]. PCC in all the cohort studies included control and self monitoring of blood glucose except for one which was designed to examine the effectiveness of preconception counseling on fetal and neonatal outcomes [30]. In addition to glycemic control, 4 studies included screening and treatment of complications of diabetes in the PCC program [16,19,21,22]. Only one cohort study (two reports) had comprehensive PCC program including, control and self monitoring of blood glucose, folic acid supplementation, smoking cessation advice and discontinuation of teratogenic drugs [10,29].
One case-control study was included in this review [13] (Table 3). It examined the effectiveness of multivitamin supplementation in the preconception period in preventing diabetes related congenital abnormalities. The study is at medium risk of bias due to possibility of recall bias during the interview of the mothers and the possibility that interviewers were not blinded to the outcome.
One trial was included in this review [12] ( Table 4). The design of the trial was not clear as authors reported it as a randomized trial but the method for randomization was not described. There was no allocation concealment and lack of blinding introduced bias because both groups were aware of the importance of the glycemic control and the complications of diabetes during pregnancy.
Only 2 studies in this review evaluated maternal hypoglycemia, as an adverse effect of PCC [10,28].

Outcome of PCC
Similarity of participants, intervention, and outcomes in addition to the score of low or medium risk of bias, made meta-analysis possible for12 cohort studies [10,15,16,[19][20][21]23,25,26,28,32,33] with 2502 participants (Tables 1, 2 & 5 and Figures 2,3,4,5,6, and 7). Both dichotomous and continuous data were pooled but only when standard deviation and similar units were available for continuous data. Studies which were at high risk of bias or of a design other than cohort were excluded from the meta-analysis.
Meta-analysis of 5 trials show that PCC lowers HbA1C in the first trimester of pregnancy by an average of 2.43% (95% CI 2.27-2.58) and while there is high heterogeneity (I 2 = 97%) this variation is in the size of the effect rather than the direction ( Figure 5).
Women who received PCC booked earlier during pregnancy for antenatal care compared to women who did not, by an average of 1.32 week (95% CI 1.4-1.23) ( Figure 6)    The evidence did not support the effectiveness of the PCC in reducing, spontaneous abortion, preeclampsia, cesarean delivery, macrosomia, RDS, SGA and neonatal hypoglycemia (Table 5 and Additional file 1:Appendix 3).
The use of multivitamins in the preconception period as a sole intervention, was evaluated by one case control study [13] and was found not to be effective in reducing the rate of congenital malformations (Odd Ratio (OR) 0.15 95% CI 0.00-1.99).
Similarly one study, at high risk of bias, evaluated the effectiveness of preconception counseling, as a sole intervention, in improving fetal and neonatal outcomes, showed improvement in total mortality (still birth and neonatal death) and the rate of congenital malformation [30].
Hypoglycemia as adverse effect of PCC was evaluated by two studies [10,28]. Meta-analysis of the pooled data did not show difference between the PCC and the control group (Figure 7). Data were not available for the evaluation of the effects of PCC on polyhydramninos, termination of pregnancy for congenital malformations, induction of labor, birth trauma, shoulder dystocia and admission to NICU.

Discussion
Our systematic review of the effectiveness of PCC in the improvement of maternal and fetal outcomes, found sufficient evidence to support its implementation in practice.
The nature of the intervention lent strength to the observational studies by avoiding certain biases known to occur in such study designs. Lack of allocation concealment and blinding of participants were avoided by recruiting the intervention and the control groups at different times during the course of the study (preconception period and antenatal period). Due to the relatively short duration of the pregnancy, attrition bias was noted in only one study, [26] all other studies had complete follow up of both groups. However the problem of confounding factors such as smoking, maternal age, parity and vascular complications of diabetes, was noted by most of the studies but only one study used the appropriate statistical test to quantify the effect of the PCC apart from the confounders [10].
The homogeneity of the participants, the intervention and the outcomes gives confidence in the estimated effects of the PCC from the pooled data.
The effectiveness of PCC in reducing congenital malformations is impressive (Table 5 and Figure 2) and has practical implication considering the recent report of the CEMCH [4] which showed that congenital malformations rate in infants of diabetic mothers in England, Wales and Northern Ireland is more than twice the background population rate. This finding is also of a paramount  High (Unclear report of the outcome, the control group was aware of the importance of glycemic control and was repeatedly advised to change into intensive therapy when planning pregnancy. So intervention was not restricted to the preconception group. No specific target level of the blood sugar was stated for the preconception group) Key: HbA1C = Glycosylated Hemoglobin A, PCC = Preconception Care, NPCC = No Preconception Care.

Figure 2
Risk ratio for congenital malformations from 11 studies of women with preexisting diabetes mellitus who did or did not receive preconception care. PCC= the group who received preconception care; NPCC= the group who did not received preconception care; CI= Confidence intervals.  Figure 3 Risk ratio for preterm delivery from 4 studies of women with preexisting diabetes mellitus who did or did not receive preconception care. PCC= Preconception care; NPCC= No preconception care; CI= Confidence intervals.

Figure 4
Risk ratio for perinatal mortality from 5 studies of women with preexisting diabetes mellitus who did or did not receive preconception care. PCC= the group who received preconception care; NPCC= the group who did not received preconception care; CI= Confidence intervals. importance to many communities in the Middle East [35], North Africa [36] and some communities in Asia [37] where the burden of congenital malformation is very high due to many causes including maternal diabetes. The effect of PCC in reducing the rate of congenital malformations reflected positively on its effect in reducing the perinatal mortality among women who utilized the care (Figure 4). This effect addresses a major health problem of four folds increase in the perinatal mortality in mothers with preexisting diabetes when compared to the general population [38] The meta-analysis supported the effectiveness of the PCC in reducing the rate of preterm delivery (Table 5 and Figure 3). We believe that effect would have been larger if data were available for very preterm delivery ≤34 weeks of gestation when the effect of the preconception rather than the antenatal care is evaluated as demonstrated by one study [10].
Maternal hyperglycemia during the period of organogenesis is known to be associated with congenital malformations [39,40]. The analysis of the pooled data in this review suggested that PCC is effective in reducing the level of HbA1C during the first trimester of pregnancy and hence the risk of congenital malformations (Table 5 and Figure 5).
We were surprised that meta-analysis did not support the effectiveness of PCC in improving the rates of spontaneous abortion (Table 5 and Additional file 1: Appendix 3). We suggest that this result is due to late attendance of the control group for antenatal care by which time some events of spontaneous abortion might have been missed. This suggestion was further supported by meta-analysis of the gestation age at first visit for the PCC and the control groups, (Figure 6) which showed significant difference between the two groups.
In this review one case control study addressed the effectiveness of multivitamins supplementation in the preconception period, as an isolated intervention, in reducing the rate of congenital malformations [13]. The role of folic acid and multivitamins in the prevention of some congenital malformation is well documented [41]. However all other studies included in this review, except for one recent report [10], did not include multivitamin or folic acid in their program of PCC, which supports an expectation of larger effect of PCC in improving fetal and neonatal outcomes if folic acid or multivitamin supplementation becomes an integral part of that care.
Another isolated preconception intervention proved to be effective in improving fetal and neonatal outcomes, is women counseling, an intervention evaluated by only one study [30] Other outcomes which did not improve by PCC, such as pre-eclampsia, cesarean delivery and macrosomia (Table 5 and Additional file 1: Appendix 3), might be related to care during the latter part of pregnancy rather than the preconception period. However few studies were included in the meta-analysis for these outcomes and further larger studies, with more participants might prove the effectiveness of PCC in improving some or all of these outcomes. Only two studies evaluated maternal hypoglycemia as adverse effect of PCC [10,28] and the pooled data showed no difference between the two groups ( Figure 7). Marked heterogeneity might be due to the differences in the target blood glucose level between the two studies.
One study conducted an economic evaluation of PCC and found that it is associated with considerable saving and reduced resources utilization [22] and yet population based studies showed that only 34-38% of eligible women receive PCC [4,30].
We suggest that more research is needed in methods of encouraging diabetic women to utilize PCC.
Our review confirms previous findings by Ray et al [9]. The strength of our review comes from the comprehensive evaluation of the available evidence on the effectiveness and safety of PCC in improving maternal and fetal outcomes together with assessment of wide range of interventions which we considered as PCC and all the possible maternal, fetal or neonatal outcomes which are affected by maternal preexisting DM. However we are aware of the limitation of the observations studies as the sole source of evidence and the inherent bias associated with the design of the cohort studies included in the meta-analysis.
The review carries important implications for practice and research as it highlights the importance of the integration of PCC in the routine care of diabetic women during their reproductive age.

Conclusion
PCC for women with preexisting type 1 or type 2 DM is effective in improving rates of congenital malformation, perinatal mortality, preterm labour, level of maternal HbA1c in the first trimester of pregnancy and maternal early utilization of antenatal care.

Additional material
Additional files 1: Appendices Appendix 1: Search Strategy (the systematic review search strategy). Appendix 2: Table of excluded studies (description of studies excluded from the review and the reasons for exclusion). Appendix 3: maternal and fetal outcomes not improved by preconception care (Forest Plots of meta-analysis of outcomes which are not improved by preconception care).