Editorial Commentary: Unappreciated but not unimportant: health disparities in the risk for cervical insufficiency

Although cervical insufficiency has been recognized within the medical literature for over 150 years as a pregnancy complication, a widely agreed-upon definition for the underlying diagnosis has yet to be achieved (Romero et al., 2006). Some authorities suggest that the definition should be applied to cases of cervical dilatation in the absence of significant uterine activity (ACOG, 2003). Others suggest that the definition should include a functional component of repeat pregnancy loss (Iams, 2009). To complicate matters further, the advent of ultrasonic cervical length measurement has added an additional criterion to the definition of cervical insufficiency (Andersen et al., 1990; Iams et al., 1996; Heath et al., 1998; Taipale and Hiilesmaa, 1998; Hassan et al., 2000). However, a variety of clinical conditions including preterm labor, uterine over distension, intrauterine and intra-cervical inflammation can each also meet many of the criteria suggested for the diagnosis of cervical insufficiency. For many practitioners, cervical insufficiency may well be a diagnosis of exclusion; that which is applied when other more certain diagnoses have been eliminated. This lack of a clear clinical definition likely spills over into perinatal epidemiological research. If it is difficult to agree upon a working clinical definition of cervical insufficiency, it is also difficult to extract a coherent definition from statistical registries and vital events data. This difficulty may account for the relatively few studies that have attempted to examine cervical insufficiency at the population level. However, just because it is difficult does not mean that we should not try. In this issue, Anum et al. should be congratulated for their efforts in taking on this difficult task (Anum et al., 2010). Using vital events data on 2 77 890 singleton deliveries recorded with the 1989 Revision of the US Standard Certificate of Live Birth, the authors offer a unique, large-scale, population-level examination of cervical insufficiency. Furthermore, this work is the first to document an increased incidence of this specific pregnancy complication in the African-American community. The risk of having a pregnancy complicated by cervical insufficiency is over 2-fold higher among African- American mothers than among white mothers. This is an unprecedented observation that agrees, at least on a wholly subjective level, with clinical experience. Given both the almost 2-fold higher rate of preterm births among African-Americans (Behrman and Butler, 2007) and the contribution of cervical insufficiency to spontaneous preterm birth (McElrath et al., 2008), the present findings offer new insights into one of the potentially important but thus far unappreciated constituent components of racial disparity in perinatal outcomes (Bryant et al., 2010). However, given the vagaries of the clinical definition of cervical insufficiency noted above, criticisms are inevitable. Let us consider these potential limitations and how the authors' findings should be contextualized. This analysis relies on a positive response to the ‘cervical incompetency’ question on the 1989 birth certificate revision. Many authors have questioned the validity and reliability of birth certificate data (Buescher et al., 1993; Piper et al., 1993; Schoendorf et al., 1993; Gaudino et al., 1997; Dobie et al., 1998; DiGiuseppe et al., 2002; Ananth, 2005). These questions and concerns may account for some of the relative underutilization of birth certificate data in perinatal research despite its availability and large sample size. Some concerns undoubtedly stem from the inter-institutional differences in birth certificate data collection procedures and protocols (Smulian et al., 2001), where the same pregnancy condition may be coded differently in separate hospitals. As there is no widely agreed-upon definition of cervical insufficiency, inter-institutional differences in data collection procedures would tend to underreport the true incidence of the condition. Further, Lydon-Rochelle et al. documented significant underreporting of all pregnancy complications on birth certificates (Lydon-Rochelle et al., 2005). Using birth certificate, billing and medical record data from Washington State, they compared the accuracy of information available on the birth certificates for 4541 deliveries against information abstracted from the medical record. While they did not comment specifically on complications associated with preterm delivery, they did observe that relative to the medical record, most diagnoses and procedures were underreported on the birth certificate. Combining this underreporting with the lack of a widely agreed-upon definition of cervical insufficiency, the true incidence of ‘cervical incompetency’ is again likely to be underreported in birth certificate data. However, one must then ask if these potential sources of underreporting would call into question the validity of the findings of Anum et al.? If one questions the validity of a data source with regard to a newly documented finding, one might seek reassurance by checking if other related but known associations apply. In this regard the authors offer us much to work with: they note that, after controlling for covariates, women reporting a single prior pregnancy termination have an odds ratio of 2.49 for a current complication of cervical insufficiency compared with women who do not report a prior termination. The association between cervical dilatation and both cervical insufficiency and preterm delivery has been previously observed (Johnstone et al., 1976; Henriet and Kaminski, 2001). Assuming that the ‘prior pregnancy terminations’ noted in the author's data include a large percentage of operative interventions, the association in the authors' data is reassuring. Further, the authors note a ‘dose’ effect with regard to the number of prior reported terminations. Women reporting two, three and four plus prior terminations have associated odds ratios of 4.66, 8.07 and 12.36, respectively. This observation is also consistent with prior findings regarding the number of elective pregnancy terminations and the risk of an adverse perinatal outcome (Mandelson et al., 1992). Anum et al. also observe that, compared with women receiving adequate levels of prenatal care, those who receive beyond adequate levels of care have an almost 3-fold increase in the risk of cervical insufficiency. Again, this observation is sensible as women with high-risk pregnancy complications are likely to either present for care more frequently or be followed more closely by their providers. In these respects, the authors' data seem to ‘check-out’ with regard to known associations between cervical insufficiency and maternal prenatal condition. Therefore, despite the potential for underreporting, these checks offer further reassurance that the racial disparity in the incidence of cervical insufficiency observed by the authors is likely to be valid. Although the underreporting of cervical insufficiency is a possibility, unless one assumes a race-specific bias in the reporting of cervical insufficiency, then not only does the observation of a disparity remain legitimate, but further, the magnitude of the estimate also remains appropriate. The authors also offer evidence to support this assertion: they note that the rates of missing data for cervical insufficiency are similar for whites, blacks and Asians. While more African-American women report significantly higher numbers of two or more prior pregnancy terminations, this difference suggests a potential explanation for the racial disparity in cervical insufficiency rather than an argument against the validity of the initial observation. If underreporting is not systematic with regard to maternal race, then the magnitude of the racial disparity effect estimated in the analysis is the same. Therefore, the use of birth certificate data may potentially lead to an underestimate of the incidence of cervical insufficiency, both overall and with regard to maternal race. However, the documentation of a racial disparity in the risk of cervical insufficiency remains valid. We have to acknowledge that the use of birth certificate data imposes potential limitations on our understanding of the epidemiology of cervical insufficiency. Yet clever researchers with appropriately framed questions and analyses will nonetheless be able to extract insightful and appropriate information. Hopefully the future will bring additional contributions to expand upon this beginning as the contribution of cervical insufficiency to adverse perinatal outcome has been thus far unappreciated but it is far from unimportant.