Long-term cardiovascular risks in patients with hypertensive disorders of pregnancy treated with nifedipine or labetalol

Abstract

Background

Hypertensive disorders of pregnancy (HDP) are established precursors to future cardiovascular disease. Current guidelines endorse labetalol, nifedipine, and methyldopa as first-line antihypertensive therapies, but data on long-term maternal cardiovascular risk are limited.

Objectives

We evaluated five-year postpartum outcomes of HDP patients who were exposed to nifedipine or labetalol during pregnancy.

Methods

We conducted a retrospective cohort study using the TriNetX US Collaborative Network. Patients were identified using ICD-10-CM codes for HDP—excluding those with pre-existing hypertension-and categorized by treatment with labetalol or nifedipine. Patients exposed to amlodipine or methyldopa were excluded. After 1:1 propensity score matching, we compared cardiovascular and renal outcomes over five years. Results are reported as adjusted odds ratios (aOR) with 95% confidence intervals (CI) and as hazard ratios (HR) from Kaplan-Meier survival analysis.

Results

After matching, 11,764 patients remained in each cohort (23,528 total). Nifedipine exposure was associated with lower odds of chronic hypertension (CH, aOR 0.835, 95% CI [0.778, 0.896], p < 0.001), heart failure (HF, aOR 0.699, 95% CI [0.527, 0.926], p = 0.012), HF with reduced ejection fraction (HFrEF, aOR 0.608, 95% CI [0.418, 0.886], p = 0.009), and chronic kidney disease (CKD, aOR 0.653, 95% CI [0.492, 0.867], p = 0.003) compared to labetalol. Time-to-event analysis confirmed these findings.

Conclusion

Among HDP patients, nifedipine exposure during the index pregnancy was associated with lower five-year risks of CH, HF, HFrEF, and CKD compared with labetalol. These findings suggest that antihypertensive choice in pregnancy may differentially affect long-term maternal cardiovascular health.

Graphical abstract

1. Introduction

Hypertensive disorders of pregnancy (HDP) affect between 13 and 15% of pregnancies in the United States [1]. HDP are a leading cause of maternal mortality and morbidity due to complications such as preeclampsia, eclampsia, stroke, or placental abruption [2]. The fetus can also suffer adverse effects, ranging from low birth weight to premature delivery or even stillbirth [2]. Recent work has validated a correlation between HDP and neurodevelopmental outcomes [3], though causative data are still lacking. HDP are also well-established precursors to future cardiovascular disease (CVD) [3,4], making effective management important for the long-term health of the pregnant patient. The American College of Obstetricians and Gynecologists (ACOG) and European Society of Cardiology (ESC) guidelines currently recommend the drugs labetalol, nifedipine, or methyldopa for blood pressure control in pregnancy [5,6]. Randomized trials and observational studies comparing labetalol and nifedipine in pregnancy show similar efficacy for blood pressure lowering and comparable perinatal safety [7,8], but they rarely extend follow-up beyond delivery or the early postpartum period [[9], [10], [11]].

HDP are delineated based upon time of diagnosis and are described as chronic or gestational hypertension that may progress into preeclampsia. Chronic hypertension is diagnosed in pregnant patients having blood pressures of 140/90 mmHg or above that predate the pregnancy [12,13], even though the American College of Cardiology and the American Heart Association (AHA) changed the threshold for hypertension in the general population to 130/80 mmHg or above in 2017 [11,14]. Gestational hypertension is likewise defined as a maternal blood pressure of 140/90 mm Hg or above but arises de novo after 20 weeks gestation in the absence of proteinuria and without biochemical or hematologic abnormalities [2,11,15,16]. Regardless of the timing of hypertension onset or changing definitions, an elevated blood pressure of 140/90 mmHg or above is associated with increased fetal and maternal risks [4,11,15]. If blood pressures rise with a systolic pressure of 160 mmHg or above or a diastolic pressure of 110 mmHg or above, immediate intervention may be needed to prevent neurologic or organ dysfunction [2,3,12,13].

Preeclampsia is a complication of gestational hypertension with blood pressures of 140/90 mmHg or above with either proteinuria (300 mg/24 h or more) or evidence of maternal end-organ dysfunction, such as acute kidney injury, liver dysfunction, or neurological complications [3,13]. Preeclampsia increases the risks of preterm delivery, reduced fetal growth, placental abruption, and perinatal mortality [14]. Preeclampsia can also progress rapidly without warning and may be fatal [2,8]. It is responsible for more than 70,000 maternal deaths and more than 500,000 fetal and neonatal deaths annually worldwide [2]. The only definitive treatment for preeclampsia is delivery of the placenta, so the patient, their family, and healthcare providers may be forced to consider the maternal health benefits of pregnancy termination or premature delivery compared with the maternal risks of organ failure or death in addition to the neonatal risks of an early-term birth and associated complications [17,18].

Per the most current guidelines, HDP are treated with labetalol, nifedipine, or methyldopa. Labetalol is an alpha and beta adrenergic antagonist, methyldopa is an alpha-2 adrenergic agonist, and nifedipine is a calcium channel blocker [11,19]. While nifedipine is the newest addition to the guidelines for treating HDP [9], all three medications are considered relatively safe and effective for use in pregnancy and while breastfeeding [3,10,12]. These medications are preferred as they have not been associated with teratogenicity or neurodevelopmental effects during randomized clinical trials [4,15,20], although ACOG now recommends avoiding use of methyldopa due to its association with depression [12]. However, these guidelines generally treat the available agents as therapeutically equivalent, and recommendations are based largely on short-term outcomes such as blood pressure control, preterm birth, fetal growth, and peripartum complications. The data remain limited for long-term outcomes of these drugs with regards to their preventive effects on maternal cardiovascular health, including hypertension and HF. Much of the long-term follow-up data surround methyldopa, with studies focusing on child neurodevelopment without mentioning maternal cardiovascular outcomes [[21], [22], [23], [24]]. Long-term data on labetalol and nifedipine are lacking. Consequently, it remains unknown whether selection of a specific antihypertensive agent during an HDP episode influences a pregnant person's long-term cardiovascular trajectory.

To address this knowledge gap, we utilized a large, multi-institution electronic health record (EHR) network to compare long-term maternal cardiovascular outcomes over five years for individuals with HDP treated with nifedipine or labetalol. By focusing on incident diagnoses after the index pregnancy, we aimed to capture new-onset CVD rather than recurrence of pre-existing conditions. We hypothesized that, despite comparable short-term obstetric outcomes, exposure to different antihypertensive classes during and around pregnancy would be associated with distinct long-term risks of chronic hypertension and CVD.

2. Methods

We conducted a retrospective review of pregnant adult patients in the United States using the TriNetX platform. TriNetX is a federated health research network providing access to de-identified electronic medical records from participating healthcare organizations [25]. We queried the US Collaborative Network, which at the time of analysis included approximately 66 healthcare organizations. We placed gender filters to include only female patients aged 18 years or older. Temporal constraints were set to include patients from November 13, 2000 to November 13, 2020, yielding 20 years of data with a minimum five-year follow-up window for each patient.

We then narrowed this population to parous patients who had encounters for delivery-related procedures using ICD-10-CM codes and Current Procedural Terminology (CPT) codes (Supplemental Table 1). ICD-10-CM codes O80-O82 (encounter for delivery) and Z33.2 (encounter for elective termination of pregnancy) identified delivery encounters. CPT codes included 1008991 (Cesarean delivery procedures), 1008996 (delivery procedures after previous Cesarean delivery), 1009003 (abortion procedures), 1014217 (vaginal delivery only, with or without episiotomy and/or forceps), 59,400 (routine obstetric care including antepartum care, vaginal delivery, with or without episiotomy and/or forceps, and postpartum care), and 59,414 (delivery of placenta, separate procedure).

We then filtered these cohorts to only include patients with HDP using ICD-10-CM codes: O11 (pre-existing hypertension with preeclampsia), O13 (gestational hypertension without significant proteinuria), O14 (preeclampsia), O15 (eclampsia), and O16 (unspecified maternal hypertension). We excluded O10 (pre-existing hypertension complicating pregnancy, childbirth, and the puerperium without superimposed preeclampsia) to isolate patients whose hypertensive pathology arose during or was directly related to pregnancy. We retained O11 because it captures pre-existing hypertension with superimposed preeclampsia, a clinically distinct entity with pregnancy-specific pathophysiology. Temporal filters ensured that HDP diagnoses occurred within 294 days (approximately 42 weeks, reflecting the maximum expected gestational duration) before or up to 42 days (6 weeks) after the instance of delivery-related procedures. This design captured HDP diagnoses from early pregnancy through 6 weeks postpartum.

Next, we divided patients into two mutually exclusive cohorts based on medication exposure, identifying drugs of interest using RxNorm codes. One cohort included patients exposed to nifedipine (RxNorm 7417), and the other included patients exposed to labetalol (RxNorm 6185). By design, each cohort received only one of these two agents during the study window; no patient appeared in both cohorts. Patients with any exposure to methyldopa (ATC C02AB) or amlodipine (RxNorm 17,767) were also excluded to minimize cross-class contamination and confounding from additional antihypertensive use.

From the TriNetX database, we collected patient age, sex, race, relevant medical diagnoses and procedures, medication exposures, and laboratory values (including urine albumin-to-creatinine ratio). We also collected data on nine cardiovascular and renal outcomes over the five-year study period: chronic hypertension (ICD-10-CM I10, I15), heart failure (I50), heart failure with reduced ejection fraction (HFrEF; I50.2), heart failure with preserved ejection fraction (HFpEF; I50.3), cardiomyopathy (I42), hypertensive heart disease (I11), stroke (I63), ischemic heart disease (I20-I25), and chronic kidney disease (CKD; N18). We defined the index event for each cohort as the combination of delivery encounter, HDP diagnosis, and relevant antihypertensive exposure. The follow-up window was set from 1 day to 1825 days (5 years) after the index event. Patients who had a given outcome diagnosed before the start of the time window were excluded from the analysis of that specific outcome to ensure capture of incident cases.

2.1. Statistical analysis

For statistical analysis, we compared patient demographics and medical data to identify significant differences between cohorts. Propensity score matching (PSM) was performed using a 1:1 greedy nearest-neighbor algorithm without replacement and a caliper width of 0.10 pooled standard deviations, as implemented within the TriNetX platform using logistic regression from the scikit-learn package (Python 3.7) [25]. Patients were matched on the following covariates: age at index, race, ethnicity, pregnancy-related diagnoses (ICD-10-CM O00-O9A), endocrine and metabolic diseases (E00-E89), circulatory system diseases (I00-I99), alcohol-related disorders (F10), nicotine dependence (F17), cardiovascular medication use (VA CV000), and urine albumin-to-creatinine ratio (Supplemental Table 2). Covariate balance was assessed using the standardized mean difference (SMD), with SMD <0.1 indicating acceptable balance.

After PSM, outcomes were evaluated using two complementary approaches. First, measures of association analysis calculated the fraction of patients with each outcome, and results are reported with adjusted odds ratios (aOR) and 95% confidence intervals (CI). Second, Kaplan-Meier survival analysis estimated the probability of each outcome at daily intervals over the five-year window. Log-rank tests compared survival distributions between cohorts, and hazard ratios (HR) with 95% CIs are reported. The proportional hazards assumption was assessed using a chi-squared test for each outcome. Statistical significance was defined as p < 0.05.

To quantify the robustness of observed associations to potential unmeasured confounding, we calculated E-values for all statistically significant outcomes [26]. The E-value represents the minimum strength of association that an unmeasured confounder would need to have with both the treatment and the outcome to explain away the observed association, conditional on measured covariates. E-values were calculated from HR using the method described by VanderWeele and Ding [27].

All data collection, processing, and transmission were done in compliance with applicable data protection laws. This retrospective study is exempt from informed consent. The data reviewed are a secondary analysis of existing data, do not involve intervention or interaction with human subjects, and are de-identified per the standard defined in Section 164.514(a) of the HIPAA Privacy Rule. This de-identification was attested through formal determination by a qualified expert as defined in Section 164.514(b) (1) of the HIPAA Privacy Rule, refreshed in December 2020.

3. Results

3.1. Study population and baseline characteristics

We identified 36,041 adult pregnant patients diagnosed with HDP from early pregnancy through 6 weeks postpartumanchored to delivery-related encounters. This included 11,878 nifedipine-exposed and 24,163 labetalol-exposed individuals. After 1:1 PSM, the final analytic sample comprised 23,528 patients, with 11,764 in each cohort (Fig. 1).

Fig. 1.

Flow chart showing cohort selection for this retrospective study comparing long-term outcomes in patients with hypertensive disorders of pregnancy treated with either nifedipine or labetalol. After filtering patients who met all selection criteria for study inclusion, propensity score matching was used to create two balanced cohorts used in the final comparison.

Before PSM, clinically meaningful differences were observed between exposure groups, including race/ethnicity distribution (Black or African American: 33.9% nifedipine vs. 23.1% labetalol, SMD 0.241; White: 49.3% vs. 60.6%, SMD 0.227), pregnancy-related diagnoses (98.1% vs. 94.8%, SMD 0.182), and cardiovascular medication use (72.5% vs. 67.3%, SMD 0.113). After PSM (Table 1), covariate balance was achieved across all measured confounders, with all standardized mean differences below 0.035.

Table 1.

Baseline characteristics before and after propensity score matching.

	Before Matching			After Matching
	Nifedipine (n = 11,878)	Labetalol (n = 24,163)	SMD	Nifedipine (n = 11,764)	Labetalol (n = 11,764)	SMD
	Mean ± SD	Mean ± SD		Mean ± SD	Mean ± SD
Age at Index (years)	29.8 ± 6.5	29.6 ± 6.6	0.027	29.8 ± 6.5	29.8 ± 6.5	0.002
Race/Ethnicity	n (%)			n (%)	n (%)
White	5841 (49.3)	14,565 (60.6)	0.227	5841 (49.7)	5902 (50.2)	0.010
Black or African American	4015 (33.9)	5556 (23.1)	0.241	3951 (33.6)	4009 (34.1)	0.010
Hispanic or Latino	1919 (16.2)	4056 (16.9)	0.018	1909 (16.2)	1860 (15.8)	0.011
Not Hispanic or Latino	7760 (65.6)	16,153 (67.2)	0.034	7731 (65.7)	7759 (66.0)	0.005
Unknown Ethnicity	2158 (18.2)	3835 (15.9)	0.061	2124 (18.1)	2145 (18.2)	0.005
Unknown Race	840 (7.1)	1837 (7.6)	0.021	840 (7.1)	807 (6.9)	0.011
Other Race	612 (5.2)	893 (3.7)	0.071	603 (5.1)	546 (4.6)	0.022
Asian	382 (3.2)	816 (3.4)	0.009	382 (3.2)	385 (3.3)	0.001
Native Hawaiian or Other Pacific Islander	91 (0.8)	216 (0.9)	0.014	91 (0.8)	78 (0.7)	0.013
American Indian or Alaska Native	56 (0.5)	161 (0.7)	0.026	56 (0.5)	37 (0.3)	0.026
Diagnoses
Pregnancy, childbirth, and the puerperium	11,614 (98.1)	22,783 (94.8)	0.182	11,541 (98.1)	11,547 (98.2)	0.004
Endocrine, nutritional, and metabolic diseases	5682 (48.0)	11,726 (48.8)	0.015	5661 (48.1)	5628 (47.8)	0.006
Diseases of the circulatory system	3327 (28.1)	7649 (31.8)	0.081	3325 (28.3)	3302 (28.1)	0.004
Alcohol-related disorders	208 (1.8)	300 (1.2)	0.042	199 (1.7)	180 (1.5)	0.013
Nicotine dependence	1305 (11.0)	2260 (9.4)	0.054	1274 (10.8)	1168 (9.9)	0.030
Medications
Cardiovascular medications	8577 (72.5)	16,176 (67.3)	0.113	8505 (72.3)	8516 (72.4)	0.002

Note: SD: Standard deviation; SMD: Standardized mean difference.

3.2. Five-year cardiovascular and renal outcomes

In matched analyses, nifedipine exposure during the index HDP pregnancy was associated with significantly lower odds of several incident cardiovascular and renal outcomes over five years compared with labetalol exposure (Table 2). Nifedipine was associated with lower odds of developing incident chronic hypertension (aOR 0.835, 95% CI [0.778, 0.896], p < 0.001), heart failure (aOR 0.699, 95% CI [0.527, 0.926], p = 0.012), HFrEF (aOR 0.608, 95% CI [0.418, 0.886], p = 0.009), and CKD (aOR 0.653, 95% CI [0.492, 0.867], p = 0.003). No statistically significant differences were observed for HFpEF (aOR 0.947, p = 0.815), cardiomyopathy (aOR 0.807, p = 0.154), hypertensive heart disease (aOR 0.859, p = 0.360), stroke (aOR 0.731, p = 0.127), or ischemic heart disease (aOR 0.893, p = 0.388).

Table 2.

Measures of association: adjusted odds ratios (aOR) for cardiovascular and renal outcomes when treated with nifedipine or labetalol during the five-year follow-up period.

Outcome	Nifedipine Events (N)	Labetalol Events (N)	aOR	95% CI	p-value
Chronic hypertension	1896	2120	0.835	(0.778, 0.896)	<0.001
Heart failure	83	118	0.699	(0.527, 0.926)	0.012
HFrEF	44	72	0.608	(0.418, 0.886)	0.009
HFpEF	36	38	0.947	(0.600, 1.495)	0.815
Cardiomyopathy	80	99	0.807	(0.600, 1.085)	0.154
Hypertensive heart disease	68	79	0.859	(0.621, 1.189)	0.360
Stroke	41	56	0.731	(0.488, 1.095)	0.127
Ischemic heart disease	111	124	0.893	(0.690, 1.155)	0.388
Chronic kidney disease	80	122	0.653	(0.492, 0.867)	0.003

Note: CI: confidence interval; CKD: chronic kidney disease; HFrEF: heart failure with reduced ejection fraction; HFpEF: heart failure with preserved ejection fraction. Patients with the outcome diagnosed before the time window were excluded from each respective analysis.

3.3. Time-to-event analysis

Kaplan-Meier survival analysis (Table 3) confirmed the associations between nifedipine exposure and lower odds of cardiovascular and renal outcomes. Nifedipine was associated with significantly longer time to develop chronic hypertension (HR 0.880, 95% CI [0.827, 0.936] log-rank p < 0.001), heart failure (HR 0.724, 95% CI [0.547, 0.959] p = 0.024), HFrEF (HR 0.628, 95% CI [0.431, 0.913] p = 0.014), and CKD (HR 0.681, 95% CI [0.514, 0.903], p = 0.007). The probability of five-year survival, free from chronic hypertension, was 73.85% in the nifedipine group, compared with 70.89% in the labetalol group.

Table 3.

Kaplan-Meier survival analysis and E-values for cardiovascular and renal outcomes over the five-year follow-up in patients treated with nifedipine or labetalol.

Outcome	HR	95% CI	Log-Rank p	Nifedipine 5-yr Survival, n (%)	Labetalol 5-yr Survival, n (%)	PH Test p	E-value (point, CI)
Chronic hypertension	0.880	(0.827, 0.936)	<0.001	1896 (73.85)	2120 (70.89)	0.895	1.53, 1.34
Heart failure	0.724	(0.547, 0.959)	0.024	83 (99.05)	118 (98.77)	0.015	2.11, 1.25
HFrEF	0.628	(0.431, 0.913)	0.014	44 (99.50)	72 (99.27)	0.022	2.56, 1.42
HFpEF	0.982	(0.623, 1.549)	0.938	36 (99.60)	38 (99.60)	0.892	--
Cardiomyopathy	0.833	(0.620, 1.119)	0.224	80 (99.12)	99 (98.98)	0.153	--
Hypertensive heart disease	0.894	(0.646, 1.236)	0.496	68 (99.21)	79 (99.16)	0.046	--
Stroke	0.759	(0.507, 1.136)	0.178	41 (99.54)	56 (99.41)	0.334	--
Ischemic heart disease	0.936	(0.725, 1.210)	0.615	111 (98.68)	124 (98.63)	0.042	--
Chronic kidney disease	0.681	(0.514, 0.903)	0.007	80 (99.08)	122 (98.67)	0.412	2.30, 1.45

Note: CI: confidence interval; HR: hazard ratio; PH: proportional hazards. E-values are reported for statistically significant outcomes. A higher E-value indicates greater robustness to potential unmeasured confounding.

The proportional hazards assumption was satisfied for chronic hypertension (p = 0.895) and CKD (p = 0.412), supporting the validity of the reported HR for these outcomes. For heart failure (p = 0.015) and HFrEF (p = 0.022), the assumption was violated, indicating that the protective effect of nifedipine may not be constant across the full five-year follow-up.

3.4. E-values for unmeasured confounding

E-values were calculated to assess vulnerability to unmeasured confounding (Table 3). For the outcomes with significant HR, E-values (point estimate, lower confidence interval) were the following: chronic hypertension (1.53, 1.34), heart failure (2.11, 1.25), HFrEF (2.56, 1.42), and CKD (2.30, 1.45). The E-value for HFrEF indicates that an unmeasured confounder would need to be associated with both treatment selection and the outcome by a risk ratio of at least 2.56 to fully explain the observed association, a magnitude that exceeds typical unmeasured confounders in observational studies of antihypertensive agents.

4. Discussion

4.1. Principal findings

We found that nifedipine exposure during the index pregnancy was associated with lower five-year odds of developing incident chronic hypertension and several major cardiovascular and renal outcomes compared with labetalol exposure, including heart failure, HFrEF, and CKD. Both measures of association and time-to-event analyses yielded consistent results, strengthening the observed associations. This large, multi-institution, retrospective study used a geographically diverse database of EHRs with a five-year postpartum follow-up window and PSM to reduce measured confounding. Our focus on incident diagnoses better captures new-onset disease after the index pregnancy and extends the comparative literature on antihypertensive choices in pregnancy beyond short-term perinatal safety and blood pressure control. These results represent real-world practice and suggest that the selection of medication to treat HDP may correlate with longer-term maternal cardiovascular and renal trajectories.

4.2. Context within current guidelines and prior evidence

Professional societies broadly recommend labetalol and nifedipine (with methyldopa historically included) as first-line treatment options for hypertension in pregnancy, largely emphasizing maternal/fetal safety and short-term obstetric outcomes. ACOG guidance on chronic hypertension in pregnancy and hypertensive disorders such as preeclampsia centers on diagnostic thresholds, acute treatment of severe-range blood pressures, and prevention of maternal complications rather than differential, long-term, maternal cardiovascular outcomes by drug class [12,28]. Similarly, the AHA scientific statement on hypertension in pregnancy recognizes HDP as a powerful marker of future CVD risk but does not specify an evidence-based preference between labetalol and nifedipine for long-term maternal prevention [4,16]. The 2025 ESC guidelines further reinforce HDP and adverse pregnancy outcomes as opportunities for long-term cardiovascular risk assessment and follow-up while recommending commonly used agents for acute and maintenance blood pressure control during pregnancy [13]. In aggregate, existing guidelines generally treat the main first-line agents as clinically interchangeable with respect to maternal long-term outcomes because comparative longitudinal outcome data have been limited.

Comparative effectiveness studies in pregnancy have primarily evaluated short-term maternal and perinatal endpoints. For example, recent comparative work in chronic hypertension during pregnancy suggests broadly similar effectiveness and safety profiles for labetalol versus nifedipine with respect to pregnancy outcomes, with limited extension into longer postpartum horizons [10]. Network meta-analyses of oral antihypertensives in pregnancy likewise focus on maternal/perinatal morbidity and do not directly address five-year incident maternal HF phenotypes or CKD as outcomes [9]. In contrast, our study evaluates clinically consequential cardiovascular and renal diagnoses over a five-year postpartum period, a timeframe aligned with emerging recommendations for intensified postpartum surveillance and risk factor modification after HDP [15].

4.3. Postpartum prescribing as a potential confounder

An important consideration when interpreting our findings is the role of postpartum antihypertensive management. Our study captures medication exposure during pregnancy but does not account for whether nifedipine or labetalol was continued, switched, or discontinued after delivery. Lovgren et al. demonstrated that nifedipine monotherapy at discharge was associated with a 65% reduction in the risk of postpartum readmission for hypertensive complications (aOR 0.27), while labetalol monotherapy was associated with a 66% increase in readmission risk (aOR 1.66) [26]. These data suggest that the two agents may have meaningfully different effects on blood pressure stability in the postpartum period, which could propagate into differential long-term risk. Additionally, Lihme et al. reported that up to 44% of women with HDP initiated antihypertensive therapy within 2 years postpartum, with nearly 25% starting more than 3 months after delivery and 11.6% starting more than 1 year later [29]. This finding underscores that postpartum prescribing is common, often delayed, and likely varies by initial in-hospital treatment. If women initially treated with labetalol are more likely to remain on or return to beta-blocker therapy postpartum (rather than being transitioned to a calcium channel blocker or other agent), the persistent pharmacologic exposure could contribute to the outcome differences we observed. Our study cannot disentangle the effects of the index pregnancy exposure from subsequent postpartum prescribing, and future studies with detailed medication reconciliation data are needed to clarify this relationship.

4.4. Potential explanations and biologic plausibility

Several non-mutually exclusive mechanisms could account for the associations we observed. First, nifedipine (a dihydropyridine calcium channel blocker) and labetalol (a combined alpha/beta adrenergic antagonist) have distinct hemodynamic profiles; differential effects on systemic vascular resistance, afterload, and vascular remodeling may influence longer-term blood pressure trajectories and downstream myocardial and renal stress [20]. Second, patient adherence to treatment and persistence may differ by regimen complexity and tolerability; nifedipine is frequently administered in once-daily extended-release formulations, whereas labetalol often requires multiple daily doses [3,12]. If postpartum continuation of therapy is common among those with persistent hypertension after HDP, differences in adherence could translate into differential cumulative blood pressure exposure over time. Third, clinician selection of the treatment may be influenced by clinical severity, comorbidities, or contraindications (e.g., asthma). This could generate residual confounding by indication even after PSM.

It is worth considering the broader cardiovascular context of pregnancy. Higher parity has been independently associated with deterioration in left ventricular diastolic function, with grand multiparous women demonstrating significantly higher rates of diastolic dysfunction than nulliparous women [30]. In addition, multiparity has been linked to increased aortic stiffness and epicardial fat thickness, reflecting cumulative pregnancy-related vascular changes [31]. These findings suggest that each pregnancy places additional hemodynamic and structural stress on the cardiovascular system, and that patients experiencing HDP may be particularly vulnerable to cumulative cardiac remodeling. While our study focused on a single index pregnancy and did not account for parity, the interaction between number of pregnancies and antihypertensive selection warrants further investigation.

4.5. Clinical and research implications

From a clinical standpoint, our findings support two practical considerations. First, postpartum cardiovascular prevention after HDP should be integrated into routine care pathways, consistent with cardio-obstetrics recommendations emphasizing early identification and modification of risk factors after adverse pregnancy outcomes [13,15,16]. Second, if replicated, the possibility that nifedipine exposure is associated with lower longer-term cardiovascular risk could inform shared decision-making when multiple first-line options are clinically appropriate. However, it is important to note that while the proportional hazards assumption was satisfied for chronic hypertension and CKD, it was not the case for HF. This suggests that the magnitude of nifedipine's benefit may vary over time, potentially reflecting early postpartum differences in blood pressure control, medication continuation, or vascular remodeling that attenuate or intensify as years pass.

Postpartum cardiovascular care may include use of personal smart devices and applications for these devices. Emerging evidence supports the use of digital health interventions and mobile health (mHealth) technologies for cardiovascular risk management in at-risk populations. Studies from the LIGHT trial have demonstrated that mobile applications paired with smart devices can improve heart rate variability parameters [32], peak oxygen consumption [33], and physical activity levels [34] in patients at elevated cardiovascular risk. While these studies were conducted in non-obstetric populations, the concept of leveraging wearable devices and mobile platforms for postpartum cardiovascular surveillance in HDP survivors represents a promising area for future research. Such tools could facilitate early detection of hypertension recurrence, encourage adherence to recommended lifestyle modifications, and enable remote monitoring in the critical postpartum period.

Prospective studies are needed to validate our findings and determine whether the observed differences reflect medication effects, prescribing patterns, postpartum continuation, or unmeasured confounding. These prospective studies should incorporate how effectively smart device technology supports long-term health improvements. Future work should prioritize: (1) prospective longitudinal cohorts with standardized blood pressure measurements and medication reconciliation; (2) pragmatic randomized comparisons with postpartum follow-up extending beyond one year; (3) mechanistic studies of vascular and myocardial remodeling after HDP by antihypertensive class; and (4) equity-focused analyses addressing differential access to postpartum care and cardiovascular risk surveillance.

4.6. Limitations

The limitations of our study are inherent to all retrospective EHR-based analyses. First, medication exposure is subject to misclassification because prescriptions do not confirm adherence, dosing, or duration, and medication use outside participating health systems may be incompletely captured. Second, as discussed above, postpartum antihypertensive management was not captured, and differential postpartum prescribing may confound the long-term associations observed. Third, HDP subtype severity, blood pressure values, body mass index, smoking status, socioeconomic factors, breastfeeding status, and postpartum medication changes are often unavailable or inconsistently recorded in EHR data and may contribute to residual confounding. Fourth, outcome ascertainment based on diagnostic codes may introduce misclassification, and differential healthcare utilization could bias detection of chronic hypertension or HF diagnoses. Fifth, we did not perform subgroup analyses stratified by HDP subtype, age, or race/ethnicity; such analyses require adequate sample sizes within each stratum and will be the focus of future work. Sixth, while the proportional hazards assumption was satisfied for chronic hypertension and CKD, it was violated for heart failure and HFrEF, suggesting that the hazard ratio for these outcomes may not remain constant over the full follow-up period. Despite these limitations, the large sample size, multi-institutional data source, use of both association and time-to-event analyses, and calculation of E-values to assess robustness to unmeasured confounding provide meaningful evidence for hypothesis generation.

5. Conclusion

Among individuals with HDP and no pre-existing hypertension, nifedipine exposure during the index pregnancy was associated with lower five-year odds of incident chronic hypertension, heart failure (including HFrEF), and CKD compared with labetalol exposure. These findings were confirmed by Kaplan-Meier survival analysis. Hypertensive heart disease, which showed significance in a broader analysis including patients with pre-existing hypertension complicating pregnancy (ICD-10-CM O10), did not remain significant after restricting to pregnancy-specific hypertensive pathology, suggesting that the earlier finding may have been partly driven by the inclusion of patients with baseline hypertensive cardiac remodeling. These results highlight the need to evaluate pregnancy antihypertensive selection through a life-course cardiovascular lens and support future prospective and mechanistic studies to determine whether specific agents can modify long-term maternal cardiovascular risk after HDP.

Data availability statement

The data used in this study were derived from the TriNetX US Collaborative Network, a federated, de-identified electronic health record platform operating under the HIPAA Privacy Rule de-identification standard (45 CFR 164.514). Patient-level data cannot be shared or transferred, as access is governed by data use agreements between TriNetX and participating healthcare organizations. The aggregate-level results supporting the findings of this study are reported within the manuscript and its supplementary materials. Researchers with institutional access to the TriNetX platform may independently query similar cohorts using the analytic approach described in the Methods section.

Funding

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

CRediT authorship contribution statement

Pushan Aggarwal: Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Writing – original draft. Ritika Tuli: Writing – original draft, Writing – review & editing. Prisha Dargan: Data curation. Mahathi Indaram: Writing – review & editing.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary data

The following is the Supplementary data to this article: