Validation of algorithms to identify adverse perinatal outcomes in the Medicaid Analytic Extract database
Validation of algorithms to identify adverse perinatal outcomes in the Medicaid Analytic Extract database
复制标题
DOI：
10.1002/pds.4967
复制
发表时间：
2020-03-02
影响因子：
2.6
通讯作者：
Bateman, Brian T.
中科院分区：
文献类型：
--
作者：
He, Mengdong;Huybrechts, Krista F.;Bateman, Brian T.
医疗补助分析提取（MAX）是一个来自公共保险个体的医疗保健利用数据库，已用于妊娠期药物安全性的研究。用于定义许多重要的孕产妇和新生儿结局的基于索赔的算法尚未得到验证。目的利用医院病历验证基于索赔的算法识别MAX患者的妊娠结局。方法将2000年至2010年在单一大型医疗保健系统内分娩的母亲的医疗记录与她们在MAX中的索赔联系起来。定义了基于索赔的胎盘早剥、子痫前期、产后出血、小于胎龄和非心脏先天性畸形的算法。针对使用这些算法确定的每种结果，随机抽取50例病例，并由两名医生独立审查其医疗记录，以确认感兴趣的诊断的存在；分歧由第三位医师审稿人解决。以医疗记录为金标准计算基于索赔的算法的阳性预测值（ppv）和95%置信区间（ci）。结果相关队列包括10899例活产妊娠。胎盘早剥的PPV为92% (95% CI, 82%-97%)，先兆子痫的PPV为82% (95% CI, 70%-91%)，产后出血的PPV为74% (95% CI, 61%-85%)，胎龄小的PPV为92% (95% CI, 82%-97%)，非心源性先天性畸形的PPV为86% （95% CI, 74%-94%）。结论：考虑围产期结局，ppv在74%至92%之间。这些ppv可以为偏差分析提供信息，纠正结果的错误分类。
Background The Medicaid Analytic eXtract (MAX) is a health care utilization database from publicly insured individuals that has been used for studies of drug safety in pregnancy. Claims-based algorithms for defining many important maternal and neonatal outcomes have not been validated.Objective To validate claims-based algorithms for identifying selected pregnancy outcomes in MAX using hospital medical records.Methods The medical records of mothers who delivered between 2000 and 2010 within a single large healthcare system were linked to their claims in MAX. Claims-based algorithms for placental abruption, preeclampsia, postpartum hemorrhage, small for gestational age, and noncardiac congenital malformation were defined. Fifty randomly sampled cases for each outcome identified using these algorithms were selected, and their medical records were independently reviewed by two physicians to confirm the presence of the diagnosis of interest; disagreements were resolved by a third physician reviewer. Positive predictive values (PPVs) and 95% confidence intervals (CIs) of the claims-based algorithms were calculated using medical records as the gold standard.Results The linked cohort included 10,899 live-birth pregnancies. The PPV was 92% (95% CI, 82%-97%) for placental abruption, 82% (95% CI, 70%-91%) for preeclampsia, 74% (95% CI, 61%-85%) for postpartum hemorrhage, 92% (95% CI, 82%-97%) for small for gestational age, and 86% (95% CI, 74%-94%) for noncardiac congenital malformation.Conclusions Across the perinatal outcomes considered, PPVs ranged between 74% and 92%. These PPVs can inform bias analyses that correct for outcome misclassification.