Objective
The objective of this study was to review the medical literature that has reported the risk for intrauterine fetal death (IUFD) in pregnancies with gastroschisis.
Study Design
We systematically searched the literature to identify all published studies of IUFD and gastroschisis through June 2011 that were archived in MEDLINE, PubMed, or referenced in published manuscripts. The MESH terms gastroschisis or abdominal wall defect were used.
Results
Fifty-four articles were included in the metaanalysis. There were 3276 pregnancies in the study and a pooled prevalence of IUFD of 4.48 per 100. Those articles that included gestational age of IUFD had a pooled prevalence of IUFD of 1.28 per 100 births at ≥36 weeks’ gestation. The prevalence did not appear to increase at >35 weeks’ gestation.
Conclusion
The overall incidence of IUFD in gastroschisis is much lower than previously reported. The largest risk of IUFD occurs before routine and elective early delivery would be acceptable. Risk for IUFD should not be the primary indication for routine elective preterm delivery in pregnancies that are affected by gastroschisis.
For Editors’ Commentary, see Contents
Gastroschisis is an abdominal wall defect of unclear cause and increasing incidence worldwide; current estimates are near 5 per 10,000 births. There have been great improvements in survival in this patient population because >95% of infants survive from birth to initial hospital discharge. However, there remain many questions about perinatal management and, in particular, about the optimal gestational age at delivery. Intrauterine fetal death (IUFD) is more common in pregnancies that are affected by congenital anomalies. Among all major congenital anomalies, 2% of pregnancies result in stillbirth, which is much higher than the 0.6% baseline rate in the general population. This higher risk of stillbirth results in a higher frequency and level of antenatal monitoring and, in some cases, elective delivery at <39 weeks’ gestation. Decisions regarding obstetric management must be based on accurate knowledge of the risk for fetal death.
The mean age of spontaneous labor in pregnancies that are affected by gastroschisis is between 36 and 37 weeks’ gestation, yet the average age of delivery is approximately 1 week earlier. This discrepancy leads to the conclusion that infants with gastroschisis deliver early either for fetal/maternal indications or electively. Although some clinicians advocate for early delivery to improve postnatal clinical outcomes (such as earlier initiation of enteral feeds and shorter hospitalization time), the literature does not document a consistent benefit. Therefore, the primary rationale for elective delivery before the onset of labor may be the prevention of IUFD.
The reported incidence of IUFD in pregnancies that are affected by gastroschisis is as high as 12.5%. Although the cause for the increased risk of IUFD is unknown, hypotheses include umbilical cord compression after acute intestinal dilation, oligohydramnios, cardiovascular compromise that is related to high protein loss through the defect and subsequent hypovolemia, and cytokine-mediated inflammation. Additionally, there is increased risk for volvulus and vascular compromise that could lead to fetal death. Studies that have documented high rates of IUFD are limited by small numbers, and many were conducted at a time when prenatal diagnosis of gastroschisis was uncommon. These studies found that most IUFDs occurred late in the third trimester. Obstetricians developed the practice of early elective delivery based on these studies. Additional studies that have suggested lower rates of IUFD are also limited by sample sizes and evaluations of single institutions or populations. Our own experience suggests a much lower rate of IUFD than 10-12%. The limitations of individual studies compromise ascertainment of the true incidence of IUFDs with gastroschisis.
We present a metaanalysis to generate a more accurate representation of the prevalence of IUFD among infants with prenatal diagnosis of gastroschisis. We hypothesize that the prevalence of IUFD is less than previously reported and that the risk of IUFD does not vary with gestational age.
Materials and Methods
We conducted a metaanalysis of the published, English-language literature that is related to gastroschisis.
Literature search
A systematic search was done independently by 2 of the authors (A.S., K.S.) who reviewed the literature to identify all published studies through June 2011 that were archived in MEDLINE and PubMed or were referenced in published articles. The MESH terms gastroschisis or abdominal wall defect were used. Abstracts were reviewed initially and excluded based on predetermined criteria that included non-English language, nonhuman subjects, or no relation to gastroschisis. The remaining articles were selected for full text review, which led to further exclusion of articles that did not report the number of IUFDs, case reports, studies with small sample sizes (n <10), and datasets that did not represent the total population (eg, case series of live births with gastroschisis or if the total number of pregnancies with gastroschisis was not disclosed). When there were multiple studies that used the same dataset, we included only 1 article and prioritized the article that reported the gestational age of IUFD. If both articles reported gestational age at IUFD, the article with the larger number of infants was included. The included articles were divided into those with a stated gestational age at IUFD and those without.
Data extraction
Data regarding all reported pregnancies, including termination of pregnancy, were extracted independently from all included studies by 2 authors (A.S., K.S.). Extracted data included gestational age at delivery, gestational age at IUFD, country of origin, year the study was published, presence of comorbidities in addition to gastroschisis, and obstetric delivery plan. IUFD was defined as an unplanned fetal death or stillbirth at any gestational age. The mean or median gestational age at delivery was extracted for each study. Early delivery plan was defined as systematic elective delivery at any predefined gestational age, compared with awaiting the onset of spontaneous labor or delivery because of maternal or fetal indications.
Quality assessment
A scoring system that was based on a previous metaanalysis was used to create a grading scale for the articles. Studies were independently graded (A.S., K.S.) with the use of a standardized evaluation form that had been developed for the purpose of this metaanalysis. Each study was assigned a grade of 1-5 according to the quality of reporting of 5 factors. Variables were chosen to represent the factors that we believed to be essential for contributing valid data (population-based data, prospective data collection) or essential for understanding results (identified obstetric delivery plan, reported gestational age at birth, and reported gestational age at the time of IUFD). Differences between reviewers’ grades were resolved by consensus among all 3 authors. The quality markers that were chosen for this study were identified before the start of data abstraction. The rate of IUFD was compared among studies on the basis of the assigned quality assessment scores.
Statistical analysis
The rate of IUFD was calculated for each study with the number of IUFDs reported in the numerator and the number of live births plus IUFDs in the denominator. Pregnancies that were terminated electively were not included in the numerator or denominator, because these pregnancies were considered not at risk for an IUFD. A random-effects model was used to aggregate individual effect sizes to create a pooled prevalence of IUFD. Random-effects models are based on the assumption that the studies that were selected for analysis are a sample of all potential studies by incorporating between-study variability in the overall pooled estimation. Pooled prevalence estimates of IUFD with 95% confidence intervals were reported from these models with the use of the Der Simonian-Laird random-effects method. All rates were calculated as deaths per 100 total births, with total births being the summation of live births and fetal deaths. Subgroup analyses were performed for the prevalence of IUFD with the following stratifications: gestational age, early delivery plan, study site (within US vs international), study grading, and years in which the study occurred.
Homogeneity across studies was tested with the I 2 index, which provides a measure (or percentage) of the variation in prevalence attributable to between-study heterogeneity. An I 2 value of >75% is interpreted as high heterogeneity. Post-hoc sensitivity analyses were conducted to investigate the potential sources of heterogeneity from specific studies that may have biased the analyses. Studies that potentially influenced heterogeneity were removed from analyses, and the results were compared with the original findings. A forest plot was created to illustrate the prevalence of each study, with 95% confidence intervals, that contributed to the analysis along with the pooled prevalence estimate. Finally, all studies that reported mean/median gestational ages of the live births of gastroschisis were divided into 3 time periods: before 1990, 1990-1999, and after 2000. The mean/median birth rates of gastroschisis were described for each time period to determine trends in timing of delivery.
Results
Study and patient characteristics
Our search produced 1123 results. Review of these abstracts resulted in 100 articles for further review ( Figure 1 ). Six articles were excluded because of use of overlapping datasets with articles that were included in the analysis. Three articles were removed for having <10 subjects. One article was removed because it did not appear to study consecutive cases of pregnancies that were affected by gastroschisis and thus did not reflect the total population at risk, and 3 additional articles were excluded for being case reports. We included 54 eligible studies in the final statistical analysis ( Figure 2 ). Thirty-five studies reported information regarding gestational ages at the time of the IUFD and/or a mean or median gestational age at delivery. Nineteen articles included the total number of IUFDs but did not provide the gestational age of each IUFD.
Final eligible studies included 3276 total pregnancies that were affected by gastroschisis (IUFD plus live births) and 177 IUFDs. Of the 54 studies that were included, 12 studies (22%) reported no IUFDs ( Table 1 ). Sixteen studies (30%) reported a planned elective delivery before onset of labor. Seventeen studies (32%) took place in the United States. The median gestational age of 48 IUFDs for which information was available was 33 weeks (range, 18–41 weeks). For those studies that reported a mean or median gestational age, the average reported mean or median gestational age at delivery for included studies was 35.7 weeks (median, 36 weeks; range of medians, 34–37 weeks). Only 4 studies (7.4%) had a quality grade of 5 (the highest quality). Twenty-two studies (40.7%) had a grade of 4; 14 studies (25.9%) had a grade of 3; 11 studies (20.4%) had a grade of 2, and only 3 studies had the lowest grade of 1 ( Table 2 ).
| Study | Origin | Study design | Study years | Total, n | Spontaneous abortion, n | Termination of pregnancy, n | Intrauterine fetal death, n/N (%) | Delivery plan | Gestational age at birth, wk + d | Gestational age of intrauterine fetal death, wk | Quality assessment score |
|---|---|---|---|---|---|---|---|---|---|---|---|
| Abuhamad et al, 1997 | VA | PR | NR | 17 | NR | 0 | 1/17 (6) | No standard delivery plan | 35.8 a ± 2.74 | 28 | 4 |
| Adair et al, 1996 | NC | RR | 1985-1994 | 29 | NR | 0 | 4/29 (13.8) | No standard delivery plan | NR | 28-41 | 2 |
| Adra et al, 1996 | FL | RR | 1986-1994 | 47 | NR | 3 | 2/44 (4.5) | No standard delivery plan | 36.0 a ± 2.4 | 28, 36 | 4 |
| Ajayi et al, 2011 | OH | RR | 2000-2008 | 74 | NR | 0 | 0/74 | Elective delivery at 36-37 wks’ gestation | 35.2 b (26.3–38.1) | N/A | 3 |
| Alfaraj et al, 2011 | Canada | RR | 2001-2010 | 98 | NR | 0 | 1/98 (1.0) | NR | NR | NR | 2 |
| Alsulyman et al, 1996 | LA | RR | 1988-1995 | 23 | NR | 1 | 0/22 | NR | 34.3 a ± 6.3 | N/A | 4 |
| Axt et al, 1999 | Germany | RR | 1989-1997 | 18 | NR | 3 | 0/15 | NR | 36.1 a ± 3.1 | N/A | 4 |
| Badillo et al, 2008 | PA | RR | 2000-2007 | 64 | NR | 1 | 2/63 (3.2) | NR | NR | NR | 2 |
| Barisic et al, 2001 | 11 European nations | RR | 1996-1998 | 106 | NR | 31 | 13/75 (17.3) | NR | 36.3 a ± 2.2 | NR | 3 |
| Bond et al, 1988 | CA | RR | 1982-1986 | 15 | NR | 3 | 0/11 | NR | NR | N/A | 2 |
| Boyd et al, 1998 | UK | RR | 1985-1995 | 41 | NR | 7 | 0/34 | NR | 37 c | N/A | 3 |
| Brantberg et al, 2004 | Norway | PR | 1988-2002 | 64 | NR | 3 | 1/61 (1.6) | C/S at 37-39 wks’ gestation | 36 + 1 (28–39) b | 35 + 5 | 5 |
| Bugge and Holm, 2002 | Denmark | RR | 1970-1989 | 166 | NR | NR | 9/166 (3.3) | NR | NR | NR | 2 |
| Burge and Ade-Ajayi, 1997 | UK | RR | 1982-1995 | 57 | NR | 0 | 3/54 (5.6) | Spontaneous labor | 36 a | 32, 36, 39 | 4 |
| Calzolari et al, 1995 | Italy | PR | 1980-1990 | 274 | NR | NR | 40/274 (14.6) | NR | NR | 20-27 (n = 18) NR (n = 22) | 3 |
| Chen et al, 1996 | China | RR | 1/1987-9/1994 | 15 | NR | 2 | 2/13 (15.4) | NR | NR | NR | 2 |
| Chescheir et al, 1991 | NC | RR | 1986-1990 | 19 | NR | 0 | 1/19 (5.3) | NR | NR | 28 | 3 |
| Cohen-Overbeek et al, 2008 | The Netherlands | RR | 1/1991-6/2003 | 33 | NR | 2 | 3/31 (9.7) | Induction at 37 wks’ gestation | NR | 19, 33, 36 | 3 |
| Crawford et al, 1992 | UK | RR | 1986-1991 | 26 | NR | 2 | 3/24 (12.5) | No standard delivery plan | NR | 34, 35, 37 | 4 |
| Dillon and Renwick, 1995 | UK | PR | 1988-1992 | 56 | 3 | 2 | 3/51 (5.9) | No standard delivery plan | NR | 32, 32, 37 | 4 |
| Durfee et al, 2002 | MA | RR | 4/1990-12/2000 | 26 | NR | 2 | 0/24 | NR | NR | N/A | 2 |
| Eurenius and Axelsson, 1994 | Sweden | RR | 1983-1990 | 24 | NR | 4 | 1/20 (5.0) | NR | NR | NR | 2 |
| Feldkamp et al, 2008 | UT | PR | 1/1997-12/2005 | 189 | NR | 3 | 11/186 (5.9) | NR | NR | NR | 2 |
| Fillingham and Rankin, 2008 | UK | RR | 1/97-12/06 | 143 | NR | 3 | 2/140 (1.4) | NR | NR | NR | 2 |
| Fitzsimmons et al, 1988 | WA | RR | 1/1980-12/1986 | 15 | NR | 0 | 1/15 (6.7) | C/S at 36 wks’ gestation | 35.9 b (31–37) | 41 | 3 |
| Forrester and Merz, 1999 | HI | RR | 1986-1997 | 74 | NR | 6 | 7/68 (10.3) | NR | NR | NR | 1 |
| Fratelli et al, 2007 | UK | RR | 1/1997-4/2006 | 40 | NR | 2 | 2/38 (5.3) | Induction at 38-39 wks’ gestation | 37 + 1 (36 + 0 to 38 + 1) c | 18, 22 | 4 |
| Garcia et al, 2010 | Brazil | RR | 1/1997-8/2009 | 94 | NR | NR | 5/94 (5.3) | Elective C/S at 37 wks’ gestation | 36.5 ± 1.4 a | 32, 34, 35, 36, 37 | 4 |
| Garne et al, 2007 | Denmark and UK | RR | 1997-2002 | 216 | NR | 39 | 9/177 (5.1) | NR | 36 c | NR | 3 |
| Goldkrand et al, 2004 | GA | RR | 1/1994-9/2002 | 34 | NR | NR | 2/34 (5.8) | Planned delivery at >37 wks’ gestation | NR | 32.7, 36 | 3 |
| Heinig et al, 2008 | Germany | RR | 10/2001-9/2005 | 14 | NR | NR | 2/14 (14.3) | C/S at 37-39 wks’ gestation | 33 + 6 to 36 + 6 | 33 + 6, 35 + 3 | 4 |
| Hidaka et al, 2009 | Japan | RR | 1990-2006 | 11 | NR | NR | 1/11 (9.1) | C/S at 37-38 wks’ gestation | NR | 35 | 4 |
| Horton et al, 2010 | NC | RR | 1/2000- 1/2907 | 71 | NR | NR | 2/71 (2.8) | Spontaneous labor | 35 + 4 ± 2.4 a | 27, 33 | 4 |
| Huang et al, 2002 | RI | RR | 1991-2001 | 60 | NR | NR | 3/60 (5.0) | NR | NR | NR | 1 |
| Japaraj et al, 2003 | Australia | RR | 1/1993-5/2001 | 45 | NR | NR | 0/45 | NR | 35.6 b (24–39) | N/A | 3 |
| Kamata et al, 1996 | Japan | RR | 1982-1994 | 12 | NR | NR | 1/12 (8.3) | Spontaneous labor | 35.4 a ± 3.7 | 31 | 4 |
| Lafferty et al, 1989 | UK | RR | 1981-1986 | 27 | NR | 4 | 1/23 (4.3) | Spontaneous labor | 37.2 b (33.5–40.0) | 37 | 3 |
| Lausman et al, 2007 | Canada | RR | 1/1980- 12/2001 | 158 | 1 | 3 | 2/154 (1.3) | Eighty-six women had spontaneous labor; 66 women had planned delivery | 36.6 ± 2 a | 24, 35 | 4 |
| Logghe et al, 2005 | UK | RCT | 5/1995- 9/1999 | 42 | NR | NR | 1/42 (2.4) | Two groups of 21 women randomly assigned to induction at 36 wks’ gestation or spontaneous delivery | Induction: 35.8 ± 0.7 a ; spontaneous: 36.7 ± 1.5 a | 31 | 5 |
| Mears et al, 2010 | UK | RR | 2004-2008 | 60 | NR | 0 | 3/60 (5.0) | Induction at 37 wks’ gestation | 36 a | NR | 3 |
| Moir et al, 2004 | MN | PR | NR | 27 | NR | 0 | 0/27 | Deliver at >29 wks’ gestation and 3/4 criteria: (1) maximum bowel diameter >10 mm, (2) wall thickness >2 mm, (3) lack of peristalsis, (4) intestinal matting | Delivery plan: 34.2 ± 2.4 a ; controlled trial: 37.7 ± 1.8 a | N/A | 5 |
| Morrow et al, 1993 | Scotland | RR | 1983-1989 | 47 | 6 | 11 | 2/30 (6.7) | NR | 36 b (31–38) | >28 (n = 2) | 2 |
| Nicholas et al, 2009 | WA | RR | 1991-2006 | 80 | NR | 4 | 2/76 (2.6) | Spontaneous labor | NR | NR | 2 |
| Rankin et al, 1999 | UK | RR | 1986-1996 | 126 | NR | 12 | 4/108 (3.7) | NR | NR | NR | 3 |
| Reid et al, 2003 | Australia | PR | 1980-2001 | 122 | NR | NR | 12/122 (9.8) | Elective delivery at 38 wks’ gestation | 37 c (24–41) | 34 c (24–39) | 5 |
| Reigstad et al, 2011 | Norway | RR | 1993-2008 | 36 | NR | 0 | 6/36 (17) | Two groups: (1) spontaneous labor (n = 10); (2) elective C/S at 36-37 wks’ gestation (n = 20) | Group 1: 36.5 (34–40) c ; group 2: 35.0 (34–37) c | <20 (n = 3) 28, 29, 39 | 4 |
| Rinehart et al, 1999 | MS | RR | 9/1992-6/1998 | 33 | NR | 1 | 0/32 | NR | Outside center: 35.3 ± 2.2 a ; tertiary center: 35.6 ± 1.4 a | N/A | 3 |
| Salomon et al, 2004 | France | PR | 3/1998-7/2001 | 31 | NR | 1/31 (3.2) | NR | Low risk (n = 20 women): 35.5 (32–38) c ; high risk (n = 11 women: 34.5 (32–36) c | NR | 3 | |
| Santiago-Munoz et al, 2007 | US | RR | 1/1998-6/2006 | 66 | NR | 0 | 3/66 (4.5) | Spontaneous labor | 37.1 ± 1.9 a | 33, 38, 40 | 4 |
| Serra et al, 2008 | Germany | RR | 1999-2004 | 23 | NR | 0/23 | Two groups: (1) C/S at 34 wks’ gestation; (2) spontaneous labor | Group 1: 243 (226–264) days c ; group 2: 257 (235–282) days c | N/A | 4 | |
| Sipes et al, 1990 | IA, WI | RR | 12/1979-1/1989 | 33 | NR | 1 | 0/32 | Spontaneous labor | 36.3 ± 2.4 a | N/A | 3 |
| Skarsgard et al, 2008 | Canada | RR | 2005-2006 | 114 | NR | 3 (4 lost to follow up) | 1/107 (0.9) | NR | 35.9 ± 2.3 a | NR | 2 |
| Towers and Carr, 2008 | US | RR | 1/1986-12/2003 | 85 | NR | 0 | 2/84 (2.4) | Spontaneous labor | NR | 29 + 4, 31 + 3 | 4 |
| Vegunta et al, 2005 | IL | RR | 6/1998-8/2002 | 30 | NR | 0 | 0/30 | C/S 36-38 wks’ gestation | 35.7 (28.4–38.6) c | N/A | 3 |
a Data were reported as mean ± SD
b Data were reported as mean (range)
| Grade | Studies, n | Pregnancies, n | Pooled prevalence per 100 births | 95% CI |
|---|---|---|---|---|
| 1, 2 | 14 | 1262 | 4.49 | 2.81–7.08 |
| 3 | 14 | 603 | 2.50 | 1.29–4.78 |
| 4, 5 | 26 | 1411 | 5.65 | 4.01–7.89 |
Metaanalysis
The pooled prevalence of IUFD for all studies was 4.48 per 100 gastroschisis pregnancies (live births + IUFD; 95% confidence interval [CI], 3.48–5.76). There was no significant difference in IUFD rate between centers with and without an early delivery plan in place (prevalence, 4.09; 95% CI, 2.39–6.91 vs 4.64; 95% CI, 3.47–6.17 per 100 births, respectively; P = .7). The mean gestational age at delivery for those studies that reported an early delivery plan was not different from those studies in which there was no delivery plan (prevalence, 35.5 ± 0.83 (SD) vs 35.8 ± 0.85; P = .22). There was also no difference in IUFD rate between studies conducted in the United States vs outside the United States (prevalence, 3.65; 95% CI, 2.26–5.84 vs 4.89; 95% CI, 3.63–6.56; P = .30). Twenty-two of the 54 publications (40.7%) reported on study populations or study sites within the United States while 3 were from Canada. Three studies were from Asia (2 Japan, 1 China), and 1 study was from South America (Brazil). The remaining studies were from the United Kingdom, Europe, or Australia. The 3 studies from Asia had relatively high rates of IUFD that ranged from 8.33–15.38 per 100 births. These studies also had study years beginning in the 1980s; therefore, rates may reflect the practices of that region and time. The prevalence of IUFD across the publication years was quite variable, with a range from 0 (2000 and 2004) to 13.8 per 100 births (2010). The highest rates occurred in 1986 (13.3 per 100 births), 1990 (13.4 per 100 births), and 2010. No significant trends were seen among other years ( P = .8, with the use of a simple regression model).
Thirty-five articles described gestational age at the time of IUFD, which totaled 37% (n = 66) of all identified IUFDs. Fourteen of 66 IUFDs (21%) occurred at ≥36 weeks’ gestation. Figure 3 shows the prevalence of IUFD at each gestational age and cumulative prevalence of IUFD across each gestational age. Nineteen percent of IUFDs occurred at ≤30 weeks’ gestation. The pooled prevalence of IUFD that occurred at ≥36 weeks’ gestation was 1.28 per 100 births (95% CI, 0.72–2.26). The weekly prevalence of IUFD did not appear to increase at >35 weeks’ gestation. The difference between this graph of crude prevalence estimates and the pooled prevalence estimates for gestational age of ≥36 weeks is that the metaanalysis included all IUFDs that occurred at ≥36 weeks, although the graph only includes IUFDs up to 38 weeks’ gestation. In addition, the metaanalysis weights the prevalence calculations based on sample size of the studies, which can also add to the differences in the 2 point estimates.
Thirty-seven studies reported a mean/median gestational age for live births. The proportion of deliveries that occurred at 37 weeks’ gestation was no different between the 90s and 2000s (2 decades, 14% and 15.8%, respectively). The proportions that occurred at 36 weeks’ gestation were 50% and 37%, respectively. The proportions that occurred at 35 weeks were 21.4% and 42%, respectively. This temporal pattern did not vary significantly when alternative grouping strategies were used.
Twenty-eight studies reported at least 1 termination of pregnancy. There was no difference in the prevalence of IUFD among studies that reported a termination of pregnancy and studies that had none or were not reported ( P = .63). The pooled prevalence of IUFD among studies with elective termination was 4.21 (95% CI, 2.93–6.03), compared with 4.77 (95% CI, 3.32–6.81) among studies without a case of elective termination.
The 12 studies in which there were no reported fetal death included a total of 369 cases. These studies, which represent 11% of the total number of cases that were included in the metaanalysis, appeared to have a slightly lower gestational age at delivery (35.1 ± 0.99 weeks), compared with studies that reported at least 1 IUFD (35.9 ± 0.70 weeks; P = .01). We were unable to identify a pattern of differences in management or study methods between those studies with and those without reported IUFDs in terms of quality score, geographic location, or reporting of an obstetric delivery plan.
Sensitivity analysis
A pooled prevalence was calculated that included only those studies (n = 35) that reported a gestational age at the time of the IUFD. Among the 35 articles, there were 1483 infants (births + IUFDs) and 60 IUFDs. The pooled prevalence among these studies was 3.80 (95% CI, 2.68–5.35), which is consistent with findings of all included studies. When we restricted the analysis only to the 26 studies that had a quality grade of 4 or 5 (n = 1411 IUFD + live births), the pooled prevalence was 5.6 per 100 births (95% CI, 4.01–7.89). Among the 14 studies with lowest quality grades of 1 or 2, the pooled prevalence rate was 4.49 (95% CI, 2.81–7.08; Table 2 ). None of the 5 quality assessment variables were related independently to risk for IUFD.
Heterogeneity
We detected moderate heterogeneity across studies (I 2 = 63%), which indicated moderate between-studies variability. This likely is due to differences in the definition of IUFD (gestational age cutoff for spontaneous abortion vs IUFD) and practice variability in the management of gastroschisis pregnancies. Because 12 studies reported no cases of IUFD, these 12 studies were not included in the initial assessment of heterogeneity (they provided no estimates of variances for prevalence rates). To estimate their contribution if they had reported IUFDs, we assumed 1 case occurred in each study and then recalculated the I 2 , which decreased slightly from 63% to 56%. We investigated each study’s individual contribution to the heterogeneity by removing each study individually from the analysis and recalculating the pooled results, which included the assessment of the 12 studies that had the largest sample sizes (n > 100). Results indicated that the article by Calzolari et al (prevalence, 14.6 per 100 births; 95% CI, 10.78–19.16) contributed the most to the I 2 (pooled prevalence without the study of Calzolari et al, 4.47; 95% CI, 3.46–5.76; I 2 = 55%]. Skarsgard et al also contributed significantly to the heterogeneity (prevalence, 0.94 per 100 births; 95% CI, 0.05–4.56; pooled prevalence without Skarsgard, 4.40; 95% CI, 3.40–5.67; I 2 = 43%]. The removal of both of these studies decreased the I 2 to 46%.
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
