A multistate prediction model for hypertension and cardiovascular disease after pregnancy using routine care data in the Netherlands

AIMS: Adverse pregnancy outcomes (APOs) are associated with increased cardiovascular risk, but risk stratification has largely focused on midlife populations. We developed a multistate prediction model to enable early postpartum risk stratification for hypertension and cardiovascular disease (CVD). METHODS: In this nationwide, registry-based retrospective study, we predicted hypertension (intermediate outcome) and CVD (primary outcome: hospitalization or death from myocardial infarction or stroke), while accounting for non-CVD mortality (competing risk). Candidate predictors included hypertensive disorders of pregnancy (HDP: pregnancy-induced hypertension [PIH] and (pre)eclampsia), preterm birth, small-for-gestational-age, recurrent miscarriages, diabetes, country of origin, and socioeconomic status (SES; income and education). Follow-up started 6 weeks after childbirth, with a maximum of 17 years. A stratified, time-dependent Cox model was used. RESULTS: Among 737,257 primiparous women, 1.7% developed hypertension and 0.3% developed CVD. HDP showed the strongest associations with both hypertension and CVD, attenuating with age. In Dutch women with PIH, hypertension risk was higher with low vs. high SES (15% vs. 5%). Surinamese women had higher hypertension risks than Dutch women. PIH combined with preterm birth and low SES resulted in the highest risks. Predictive performance was modest to good (AUCs 0.65-0.89), with alignment between predicted and observed risks. CONCLUSIONS: This study showed future cardiovascular risk can be stratified at 6 weeks postpartum. Women with HDP, low SES, Surinamese origin, and multiple risk factors were identified as the highest-risk groups. These findings support early identification of women at cardiovascular risk, although validation is required before clinical implementation.