Delker Erin, Baer Rebecca J, Chambers Christina D, Bandoli Gretchen
Department of Pediatrics, University of California San Diego, San Diego, California, USA.
California Preterm Birth Initiative, University of California San Francisco, San Francisco, California, USA.
Pharmacoepidemiol Drug Saf. 2024 Dec;33(12):e70059. doi: 10.1002/pds.70059.
Administrative data sources are used to describe the epidemiology of chronic hypertension in pregnancy and its consequences. Differences in identification across sources may affect research estimates. We compared identification of chronic hypertension in birth certificate records, hospital discharge records, and Medi-Cal claims in the same individuals.
We used data from 820 140 2016-2020 California Medi-Cal covered births. We identified chronic hypertension on birth certificates using the prepregnancy hypertension check box and in hospital discharge records and Medi-Cal claims using ICD codes. We compared the prevalence of chronic hypertension and identified predictors of agreement. We also compared absolute and relative estimates of racial/ethnic disparities in chronic hypertension and associations with neonatal outcomes.
The prevalence of chronic hypertension was 0.7% in birth records, 2.1% in hospital discharge records, and 3.9% in Medi-Cal claims. There was low to fair agreement between birth certificate records and hospitalization records (kappa = 0.36) and Medi-Cal claims (kappa = 0.25). Characteristics associated with the worst agreement were eligibility for Women Infant and Children benefits, US-born, and normal body mass index. Estimates of the relative risk for racial/ethnic disparities and associations with preterm birth and SGA age at delivery were similar across sources. Estimates of risk differences were larger in hospitalization and Medi-Cal claims data.
Reliance on birth certificate data may contribute to underestimated prevalence estimates and biased causal estimates. Epidemiologic research and public health surveillance of chronic hypertension and its consequences should incorporate data from multiple data sources to improve validity of estimates.
利用行政数据源描述妊娠期慢性高血压的流行病学特征及其后果。不同数据源在识别方面的差异可能会影响研究估计值。我们比较了同一人群出生证明记录、医院出院记录和医疗补助申请中慢性高血压的识别情况。
我们使用了2016年至2020年加利福尼亚州820140例有医疗补助覆盖的出生数据。我们通过孕前高血压复选框在出生证明上识别慢性高血压，并通过国际疾病分类代码在医院出院记录和医疗补助申请中识别慢性高血压。我们比较了慢性高血压的患病率，并确定了一致性的预测因素。我们还比较了慢性高血压种族/族裔差异的绝对和相对估计值以及与新生儿结局的关联。
出生记录中慢性高血压的患病率为0.7%，医院出院记录中为2.1%，医疗补助申请中为3.9%。出生证明记录与住院记录（kappa = 0.36）以及医疗补助申请（kappa = 0.25）之间的一致性较低至中等。与一致性最差相关的特征是符合妇女婴儿和儿童福利资格、在美国出生以及正常体重指数。不同数据源对种族/族裔差异以及与早产和小于胎龄儿出生时年龄关联的相对风险估计相似。住院和医疗补助申请数据中的风险差异估计值更大。
依赖出生证明数据可能导致患病率估计值被低估以及因果估计出现偏差。慢性高血压及其后果的流行病学研究和公共卫生监测应纳入多个数据源的数据，以提高估计值的有效性。
