Baer Rebecca J, McLemore Monica R, Adler Nancy, Oltman Scott P, Chambers Brittany D, Kuppermann Miriam, Pantell Matthew S, Rogers Elizabeth E, Ryckman Kelli K, Sirota Marina, Rand Larry, Jelliffe-Pawlowski Laura L
Department of Pediatrics, University of California San Diego, La Jolla, CA, United States; California Preterm Birth Initiative, University of California San Francisco, San Francisco, CA, United States.
California Preterm Birth Initiative, University of California San Francisco, San Francisco, CA, United States; Department of Family Health Care Nursing, University of California San Francisco School of Nursing, San Francisco, CA, United States.
Eur J Obstet Gynecol Reprod Biol. 2018 Dec;231:235-240. doi: 10.1016/j.ejogrb.2018.11.004. Epub 2018 Nov 5.
Objective To develop a pre-pregnancy or first-trimester risk score to identify women at high risk of preterm birth. Study design In this retrospective cohort analysis, the sample was drawn from California singleton livebirths from 2007 to 2012 with linked birth certificate and hospital discharge records. The dataset was divided into a training (2/3 of sample) and a testing (1/3 of sample) set for discovery and validation. Predictive models for preterm birth using pre-pregnancy or first-trimester maternal factors were developed using backward stepwise logistic regression on a training dataset. A risk score for preterm birth was created for each pregnancy using beta-coefficients for each maternal factor remaining in the final multivariable model. Risk score utility was replicated in a testing dataset and by race/ethnicity and payer for prenatal care. Results The sample included 2,339,696 pregnancies divided into training and testing datasets. Twenty-three maternal risk factors were identified including several that were associated with a two or more increased odds of preterm birth (preexisting diabetes, preexisting hypertension, sickle cell anemia, and previous preterm birth). Approximately 40% of women with a risk score ≥ 3.0 in the training and testing samples delivered preterm (40.6% and 40.8%, respectively) compared to 3.1-3.3% of women with a risk score of 0.0 [odds ratio (OR) 13.0, 95% confidence interval (CI) 10.7-15.8, training; OR 12.2, 95% CI 9.4-15.9, testing). Additionally, over 18% of women with a risk score ≥ 3.0 had an adverse outcome other than preterm birth. Conclusion Maternal factors that are identifiable prior to pregnancy or during the first-trimester can be used create a cumulative risk score to identify women at the lowest and highest risk for preterm birth regardless of race/ethnicity or socioeconomic status. Further, we found that this cumulative risk score could also identify women at risk for other adverse outcomes who did not have a preterm birth. The risk score is not an effective screening test, but does identify women at very high risk of a preterm birth.
目的 制定一种孕前或孕早期风险评分，以识别早产高危女性。研究设计 在这项回顾性队列分析中，样本取自2007年至2012年加利福尼亚州的单胎活产，关联了出生证明和医院出院记录。数据集被分为一个训练集（样本的2/3）和一个测试集（样本的1/3）用于发现和验证。使用训练数据集上的向后逐步逻辑回归，开发了使用孕前或孕早期母亲因素的早产预测模型。使用最终多变量模型中每个母亲因素的β系数，为每次妊娠创建早产风险评分。风险评分效用在测试数据集中以及按种族/族裔和产前护理支付方进行了重复验证。结果 样本包括2339696次妊娠，分为训练和测试数据集。确定了23个母亲风险因素，包括几个与早产几率增加两倍或更多相关的因素（既往糖尿病、既往高血压、镰状细胞贫血和既往早产）。在训练和测试样本中，风险评分≥3.0的女性中约40%早产（分别为40.6%和40.8%），而风险评分为0.0的女性为3.1 - 3.3%[比值比（OR）13.0，95%置信区间（CI）10.7 - 15.8，训练集；OR 12.2，95% CI 9.4 - 15.9，测试集]。此外，风险评分≥3.0的女性中超过18%有除早产以外的不良结局。结论 孕前或孕早期可识别的母亲因素可用于创建累积风险评分，以识别无论种族/族裔或社会经济地位如何的早产风险最低和最高的女性。此外，我们发现这种累积风险评分还可以识别未早产但有其他不良结局风险的女性。该风险评分不是一种有效的筛查测试，但确实能识别早产风险非常高的女性。
