Strengths and Limitations
There are several strengths of our study. We evaluated a large cohort of
patients using CGA assessment as an adjunct screening tool for the
general population undergoing universal CL screening. Novel to our study
is quantification of the visualizable size of the CGA and evaluation of
these metrics for PTB prediction. Further strengthening our analysis,
all cervical images were obtained at a single institution using a
standardized protocol for CL imaging and subsequently assessed for CGA
visualization by a reviewer blinded to birth outcome.
Our study also has several limitations. The single-institution,
retrospective design is subject to the limitations of preexisting
medical record data and may impact the generalizability of our findings.
The power of our analysis may also be limited by examination of a rare
outcome and rare finding. Further, inclusion of patients with
progesterone exposure may confound PTB rates, though progesterone use is
not limited to cases of prior PTB or short cervix – for instance, in
conceptions via assisted reproductive technologies 37.
Progesterone use was controlled for in our regression models, and most
progesterone users (81%) were receiving the intramuscular formulation,
which has uncertain efficacy for affecting PTB risk41.
Further, our study population was relatively low-risk. There were low
rates of clinical comorbidities, and our predetermined exclusion
criteria, several of which are known risk factors for PTB, likely
contributed to the low-risk nature of our cohort. Our PTB rate was 2.6%
and reflects only spontaneous PTBs, but the overall rate rose to 7.1%
when calculated for the entire screened population, reflecting a
selection bias for low-risk pregnancies. Further, while the overall
national PTB rate is 10%, the national rate of spontaneous PTB is lower
(4.5%) and more comparable to our cohort 1, 42. Thus,
it is not surprising our cohort had a low rate of absent CGA (2.3%).
Given the retrospective nature of our study, it was not possible to
ensure all cervical images were optimized. Several patients were
excluded for unavailable or suboptimal imaging and this could have
inadvertently excluded patients with underlying CGA absence21. However, our CGA detection rate is comparable to
that of prior studies with a similar population 11, 14,
43.