The role of digital examination in the setting of a sonographically short cervix

Universal transvaginal cervical length screening has been associated with a reduction in the frequency of preterm birth. However, there is no clinically set standard to guide the performance of a digital cervical examination in the setting of a sonographically short cervix.

To investigate the prevalence of cervical dilation at various midtrimester transvaginal cervical length thresholds. Furthermore, we sought to identify sonographic or clinical characteristics associated with cervical dilation that may inform who would benefit from a digital cervical examination in the setting of a sonographically short cervix.

This retrospective cohort study was conducted at an academic institution and included women with a singleton gestation and an ultrasonographically detected short cervix (defined as a transvaginally obtained cervical length ≤25 mm) who had a documented digital cervical examination within 1 week of the ultrasonography. Bivariable analyses were used to determine the relationship between cervical length and the presence of cervical dilation. Multivariable logistic regression and receiver operating characteristic curve were used to evaluate the relationship between clinical and sonographic risk factors and cervical dilation.

Of the 256 women who met eligibility criteria and had a sonographically detected short cervix, 103 (40.2%) were found to be dilated on digital cervical examination. The prevalence of cervical dilation increased as sonographic cervical length decreased; cervical dilation was identified in 15%, 39%, 53%, 64%, and 69% of women with a cervical length between 20.0 and 25.0 mm, 15.0 and 19.9 mm, 10.0 and 14.9 mm, 5.0 and 9.9 mm, and 0.0 and 4.9 mm, respectively. Maternal race or ethnicity (examined as a social construct), insurance status, nulliparity, previous cervical excisional procedure, funneling on ultrasonography, and sonographic cervical length were each associated with cervical dilation. However, including all of these variables into a regression yielded a model with only moderate predictive ability to identify cervical dilation, with receiver operating area under the curve of 0.77 (95% confidence interval, 0.71-0.83).

Consideration should be given to performing a digital cervical examination in the setting of a sonographically short cervix (especially <20 mm) to detect cervical dilation.