Purpose

Previous research has suggested gastroschisis, a congenital malformation, may be linked to environmental or infectious factors and cases can occur in clusters. The objective of this study was to identify geographic areas of elevated gastroschisis risk.

Methods

Cases of gastroschisis were identified from birth defect registries in Massachusetts and Texas. Random samples of live births were selected as controls. Generalized additive models were used to create a continuous map surface of odds ratios (ORs) by smoothing over latitude and longitude. Maternal age, race/ethnicity, education, cigarette smoking, and insurance status (MA only) were assessed for confounding. We used permutation tests to identify statistically significant areas of increased risk.

Results

An area of increased risk was identified in North Central Massachusetts but was not significant after adjustment (P value = .07; OR = 2.0). In Texas, two statistically significant areas of increased risk were identified after adjustment (P value = .02; OR = 1.3 and 1.2). Texas had sufficient data to assess the combination of space and time, which identified an increased risk in 2003 and 2004.

Conclusions

This study suggests there were areas of elevated gastroschisis risk in Massachusetts and Texas, which cannot be explained by the risk factors we assessed. Additional exploration of underlying artifactual, environmental, infectious, or behavioral factors may further our understanding of gastroschisis.

Background

Gastroschisis is a birth defect where loops of bowel are protruding from the abdominal wall at birth. Previous research has suggested that gastroschisis cases can occur in clusters. The objective of this study was to identify if there were areas of elevated gastroschisis risk using data from the National Birth Defects Prevention Study (NBDPS), 1997 through 2007.

Methods

We obtained data on cases (n = 371) through population-based birth defects surveillance systems in Arkansas, California, and Utah; controls (n = 2359) were selected from the same geographic areas as cases. Mothers were interviewed on demographic information and exposures during pregnancy, including residential history. We used first trimester maternal addresses and generalized additive models to create a continuous map surface of odds ratios (OR) by smoothing over latitude and longitude. Permutation tests were used to assess whether location of maternal residence was important and identify locations with statistically significant ORs.

Results

In Arkansas, adjusted ORs in the southwest corner were 2.0 and the global deviance was not statistically significant (p-value: 0.57). Adjusted ORs for California indicated areas of increased risk with ORs 1.3 (p-value: 0.34). In Utah, the adjusted ORs were elevated (OR: 2.4) in the south-eastern corner of the study area (p-value: 0.34).

Conclusion

The results of this study, while not statistically significant, suggest there were spatial variations in gastroschisis births. We cannot rule out that these variations were due to edge effects or residual confounding.