Examining Changes in Clinical Management and Postpartum Readmissions for Hypertensive Disorders of Pregnancy over Time

Abstract

Objective:

In response to 2013 guidelines for hypertensive disorders of pregnancy (HDP), our study examined changes in antenatal management and postpartum readmission (PPR) over time.

Study Design:

This is a retrospective cohort study of individuals diagnosed antenatally with HDP who delivered at a tertiary care center from 2012 to 2017.

Main Outcome Measures:

The primary outcome was postpartum readmission for HDP in 2012–2013 vs. 2014–2017. Secondary outcomes included intravenous magnesium administration and prescription for oral (PO) antihypertensive medication during delivery admission. Multivariable logistic regression models assessed differences in outcomes over time, adjusted for age, race, and payer status, for HDP with and without severe features, defined by ACOG criteria.

Results:

Of 5,300 eligible individuals, 73.5% had HDP without severe features and 26.5% had severe features. The PPR frequency in this cohort was 1.1% (N=59). There was no difference in PPR for individuals with HDP without severe features (aOR 0.73; 95% CI 0.28–1.88) or with severe features (aOR 1.30; 95% CI 0.50–3.39) by epoch. Magnesium administration for HDP with severe features remained below 80% over time. Magnesium administration for HDP without severe features and discharge prescriptions for PO medications for HDP with severe features were lower after 2013. Neither magnesium administration nor discharge prescriptions were associated with decreased odds of PPR.

Conclusion:

Although there was no difference in PPR for HDP after 2013, there were changes in antenatal management of HDP, including decreased magnesium administration for individuals with HDP without severe features and PO medication for individuals with severe features.

Introduction

Postpartum hospital readmissions are increasing over time, though they remain lower than medical or surgical readmissions (1). Readmissions highlight a population that is potentially at greater risk for severe maternal morbidity during the postpartum period (2–4). They create considerable costs for the healthcare system (4, 5) and may disrupt the maternal-neonatal dyad during a critical period for bonding and initiation of breastfeeding. Understanding the factors driving postpartum readmissions may help to optimize postpartum care (6).

Hypertensive disorders of pregnancy (HDP) occur in approximately 10 percent of pregnancies. Existing studies estimate that between 9.3 percent and 21.5 percent of postpartum readmissions occur due to HDP (1, 7). Between 2013 and 2014, postpartum readmissions for HDP cost the healthcare system approximately $731 million and required 404,000 additional days in inpatient hospital care (8). Aside from postpartum readmissions for hypertension, individuals with HDP have demonstrated increased risk for postpartum readmissions for all causes (1, 9).

Previous studies have extensively examined risk factors for postpartum readmission for hypertension such as clinical factors or, rarely, social determinants of health. New guidelines were published in 2013 which changed the definitions of HDP as well as recommendations regarding its clinical management. Limited work has examined whether decisions made during the delivery admission – for example, administration of oral diuretics (10) – influence postpartum readmission for hypertension among people with HDP, but studies have not accounted for changes in clinical guidelines regarding HDP.

Thus, we aimed to investigate trends in postpartum readmissions and the clinical management of HDP over time among individuals with HDP with and without severe features, as defined by American College of Obstetricians and Gynecologists (ACOG) criteria. Examining the impact of clinical decisions during delivery admission for individuals with HDP on postpartum readmissions may provide future directions for research and quality improvement efforts focused on the postpartum care paradigm for this population.

Materials and Methods

This is a retrospective cohort study of all pregnant people with HDP who delivered at a single high-volume tertiary care center between September 2012 and December 2017. Individuals with chronic hypertension, hemolysis, elevated liver enzymes and low platelet count syndrome, pre-existing kidney, liver, rheumatologic, or hematologic disorders, or multifetal pregnancies were excluded. For this analysis, participants were classified as having severe HDP if they had systolic blood pressures ≥160 or diastolic blood pressure ≥110 that required administration of IV medications or immediate-release nifedipine, thrombocytopenia (platelet count <100,000/microliter), impaired liver function (elevation of transaminases to twice normal concentration), renal insufficiency (serum creatinine >1.1 mg/dL or twice baseline concentration), pulmonary edema, or cerebral or visual symptoms. Individuals without these criteria were classified as HDP without severe features.

The primary outcome for this analysis was postpartum readmission for HDP, comparing individuals who delivered in 2012–2013 to individuals who delivered in 2014–2017. Secondary outcomes included administration of IV magnesium for seizure prophylaxis or discharge prescription for oral (PO) antihypertensive medication during delivery admission.

The tertiary care center at which the study was performed has approximately 12,000 deliveries per year. With approximately 10 percent of patients experiencing HDP, we conservatively estimated that our study sample include 5,000 patients over a 5-year period. With an estimated sample size of 5,000, a 3:1 ratio between group sizes, assuming a baseline postpartum readmission rate of 1.0% in our control group (2012–2013), we would be adequately powered to detect a 1.18% difference in readmissions between epochs (alpha = 0.05, 80% power).

All data were either validated or obtained via abstraction from the electronic medical record. Individuals were initially identified using an electronic data warehouse using ICD codes and diagnoses of HDP were validated utilizing chart review. Individuals were categorized in a HDP group based on their final (and worst) diagnosis during their admission; for example, if an individual initially presented with HDP without severe features and subsequently met criteria for severe features, they were categorized as having severe HDP. Individuals with HDP without severe features were distinguished from individuals with severe HDP through review of lab results, which are routinely obtained for any individual with hypertension on labor and delivery at this institution, as well as blood pressures. Individuals with chronic hypertension were identified through the diagnosis included in the admission documentation as well as review of blood pressures prior to 20 weeks of gestation.

Bivariable comparisons of cohort characteristics and outcomes were performed by epoch. Multivariable logistic regression models were used to examine differences in outcomes between epochs, adjusted for age, self-identified race, and payer status. These were estimated separately for HDP with and without severe features. In the multivariable models, we adjusted for self-reported race and ethnicity (non-Hispanic White, non-Hispanic Black, Hispanic, and other), age (years), and insurance status (commercial insurance vs. other types). These covariates were chosen a priori based on established relationships with perinatal outcomes. Racial and ethnic identity was included in multivariable models as a social determinant of health that has been previously associated with health care and health outcomes related to HDP. We collected data on multiple other inter-admission and readmission characteristics that may impact postpartum readmission rate; anticipating a relatively small number of postpartum readmissions, we planned a priori to report descriptive data in an exploratory, hypothesis-generating approach.

All analyses were performed using Stata v 15 (Stata Corp, College Station, TX). This study was approved by the Northwestern University Institutional Review Board.

Results

Between September 2012 and December 2017, 5,300 people with HDP met eligibility criteria. The majority of individuals identified as non-Hispanic White (Table 1, 53.2%) or non-Hispanic Black (13.7%). Most individuals had commercial insurance (81.1%) and were nulliparous (74.1%). One thousand four hundred and four (26.5%) individuals had HDP with severe features whereas 3,896 (73.5%) had HDP without severe features (Table 1). There were no statistically significant differences in age, race, or payer status across epochs; the proportion of individuals who were nulliparous increased slightly in the 2014–17 epoch.

Table 1:

Demographics and Delivery Admission Characteristics of Study Population by Epoch^a

Characteristics	2012–2013	2014–2017	P valueb
	(n=1,085)	(n=4,215)
Age (median, IQR)	32 (29–35)	32 (29–35)	0.08
Race (n,%)			0.76
Declined	85 (7.8)	326 (7.7)
Non-Hispanic White	592 (54.6)	2,230 (52.9)
Non-Hispanic Black	144 (13.3)	580 (13.8)
Other	264 (24.3)	1,079 (25.6)
Latina ethnicityc	149 (13.7)	687 (16.3)	0.06
Nulliparous (n,%)c	814 (77.4)	2,946 (73.2)	0.01
Payer Status (n,%)c			0.19
Commercial	864 (79.7)	3,431 (81.5)
Public	220 (20.3)	781 (18.5)
BMI (median, IQR)	31.5 (28.3–35.6)	31.8 (28.5–36.0)	0.11
Cesarean (n, %)	302 (27.8)	1,152 (27.3)	0.16
HDP type			<0.001
No severe features	851 (78.4)	3,045 (72.2)
Severe features	234 (21.6)	1,170 (27.8)

^aIQR, interquartile range; BMI, body mass index; HDP, hypertensive disorders of pregnancy

^bPearson’s chi squared test for categorical variables, Mann Whitney Wilcoxon test for continuous variables

^cMissing data: Latina ethnicity, n=2; parity, n=224; payer status, n=4

The primary outcome, postpartum readmission (PPR) rate, did not differ by epoch (Table 2). In total, there were 59 PPR (1.1%), most of which were for patients with HDP with severe features during delivery admission. When looking at a visual representation of unadjusted PPR across years, there was a large decrease in PPR for patients with HDP without severe features and a large increase in PPR for patients with severe HDP between 2014 and 2015 (Figure).

Table 2:

Differences in Readmission and Delivery Admission Characteristics by Time Period

Outcomes	2012–2013 (n, %)	2014–2017 (n, %)	aOR (95% CI)a,b
Postpartum Readmission
Overall	11/1,085 (1.0)	48/4,215 (1.1)	1.10 (0.57–2.13)
HDP with severe features	5/234 (2.1)	32/1,170 (2.7)	1.30 (0.50–3.39)
HDP without severe features	6/851 (0.7)	16/3,045 (0.5)	0.73 (0.28–1.88)
IV Magnesium During Delivery Admissionc
Overall	299/1,085 (27.6)	1,052/4,214 (25.0)	0.87 (0.75–1.01)
HDP with severe features	186/234 (79.5)	916/1,170 (78.3)	0.92 (0.65–1.31)
HDP without severe features	113/851 (13.3)	136/3,044 (4.5)	0.31 (0.23–0.40)
PO Medication at Discharge from Delivery Admissionc
Overall	106/1,085 (9.7)	391/4,212 (9.3)	0.92 (0.74–1.16)
HDP with severe features	87/234 (37.2)	332/1,170 (28.4)	0.67 (0.50–0.90)
HDP without severe features	19/851 (2.2)	59/3,042 (1.9)	0.86 (0.51–1.45)

^aAdjusted odds ratios from multivariable logistic regression models adjusted for age, race and payer status.

^bAbbreviations: aOR, adjusted odds ratio; CI = confidence interval; HDP = hypertensive disease of pregnancy; IV = intravenous; PO = “per os” or oral administration

^cMissing values: IV magnesium, n=1, PO medications, n=3

Figure 1,

Unadjusted Postpartum Readmission Frequency for Hypertensive Disorders of Pregnancy over Time (%). This is a visual representation of the unadjusted postpartum readmissions rates for hypertension among patients with an antenatal diagnosis of hypertension, organized by delivery year. HDP stands for hypertensive disorders of pregnancy.

Secondary outcomes differed by HDP group (Table 2). Administration of IV magnesium for patients with HDP without severe features was lower in 2014–2017 compared to 2012–2013 (aOR 0.31, 95% CI 0.23–0.40). Discharge prescriptions for PO antihypertensive medications for patients with severe HDP were also lower in 2014–17 compared to 2012–13 (Table 2). The frequency of magnesium administration for patients with severe HDP remained below 80% throughout the study period. Individuals who received IV magnesium or PO medication at discharge had comparable odds of postpartum readmission to individuals who did not receive these interventions (data not shown).

Additional inter-admission and readmission characteristics were described for the 59 patients with postpartum readmissions. The mean length of stay for delivery admission was four days for all patients. Fifty percent of patients with HDP without severe features were scheduled for outpatient follow-up 7–10 days after discharge from their delivery admission; thirteen percent of patients with HDP with severe features were scheduled for outpatient follow-up 72 hours after admission. The recommendation for home blood pressure monitoring was made for 39 percent of patients. Only 10 percent of patients either initiated or up-titrated oral medications in the outpatient setting in between their delivery admission and their readmission. The median day of readmission was postpartum day 5 for all patients, and median length of stay for readmissions was 3 days. Twenty seven percent of patients received an initial course of magnesium during their readmission, while 63 percent of patients had previously received magnesium during their delivery admission and did not receive repeat magnesium administration. Thirty four percent of patients who were readmitted either initiated or up-titrated oral medications during their readmission.

Discussion

In this analysis, we have shown that postpartum readmission frequency among individuals with HDP remained unchanged after new guidelines were published regarding the management of HDP in 2013. We also identified quality indicators for antepartum and postpartum management of HDP that changed over time. Specifically, magnesium administration during the delivery admission for individuals with severe HDP was slightly less than 80% throughout the study period, and the use of magnesium for individuals with HDP without severe features declined in the second epoch. Additionally, patients with severe HDP had lower odds of receiving a discharge prescription for PO antihypertensive medication during their delivery admission in 2014–2017 compared to 2012–2013.

The incidence of HDP and postpartum readmissions identified by our study is similar to what has been previously described in the literature. Our study found no difference in readmissions for HDP over time when comparing deliveries between 2012–2013 to deliveries between 2014–2017. A large study identified an increasing rate of all-cause postpartum readmission between 2004–2011 but also found that other factors besides the year of delivery contributed significantly to the difference in readmissions (11). To our knowledge, no study has examined temporal trends in postpartum readmissions for HDP diagnosed during delivery admission specifically or described patterns in clinical management during those readmissions.

Several studies have attempted to identify risk factors for postpartum readmission for hypertension, both among individuals with and without HDP. Antecedent HDP, advanced maternal age, multiparity, elevated body mass index, and longer length of labor have all been shown to increase risk of postpartum readmission for hypertension (12–15). Our study controlled for several of these factors in assessing whether there was a difference in postpartum readmissions for hypertension over time. A clinically important next step would be to develop an accurate predictive model for postpartum readmission that would aid clinicians in appropriately modify clinical care to reduce the risk of morbidity and mortality (15).

Some studies have also sought to assess if clinical decisions in the intrapartum and immediate postpartum period can impact risk of postpartum readmission for hypertension. A nested case-control study of less than 500 deliveries showed that PO antihypertensive prescription during delivery admission as well as increased induction time may reduce risk of postpartum readmission among individuals with HDP (16). Our study did not support this finding regarding oral medications, potentially due to the difference in study design where case control studies have more power to examine rare outcomes such as postpartum readmissions. There are mixed findings in the literature as to whether mode of delivery is associated with risk of postpartum readmission for hypertension (16, 17), though studies have shown people with HDP have a higher risk of cesarean delivery (8) and that length-of-stay greater than five days among people with HDP who underwent cesarean is associated with a lower risk of postpartum readmission (18). Multiple studies have shown that tele-health monitoring of blood pressures following delivery can reduce risk of postpartum readmission (19–21). In addition, studies on the use of diuretics for people with HDP have found a positive impact on blood pressures postpartum but no decrease in risk of postpartum readmission (10, 22, 23). Finally, recent work utilizing machine learning to create models that predict the risk of postpartum readmission for patients with HDP may help clinicians and hospital systems to strategically stratify care for patients in the postpartum period (15).

Postpartum readmissions continue to create substantial costs for the healthcare system. While HDP is an independent risk factor for postpartum readmission, few studies specifically examine postpartum readmissions for hypertension among people with antecedent HDP. Evolving guidelines for the clinical management of HDP introduce new questions regarding the impact of clinical decision-making in the intrapartum and immediate postpartum period upon readmission incidence. In our study, fewer than 80 percent of individuals with HDP with severe features received IV magnesium during their delivery admission; however, individuals who received IV magnesium or PO medication at discharge did not have lower odds of postpartum readmission. Although our data on readmission and inter-admission characteristics are exploratory, relatively few patients had a plan for timely postpartum follow-up or initiated/increased medications between their delivery admission and readmission, highlighting potential future directions for research on the relationship between quality of care for HDP and risk of postpartum readmission.

In recent years, multiple systems-level interventions have been designed to improve quality of and access to care for HDP. For example, work by the Illinois Perinatal Quality Collaborative’s Maternal Hypertension Initiative has demonstrated the power of quality improvement initiatives to improve treatment for severe perinatal hypertension (24). Remote blood pressure monitoring programs have reduced racial disparities in blood pressure assessment and demonstrated high levels of patient satisfaction (21, 25). Our findings demonstrate additional areas for future research and potential intervention during delivery admission as well as postpartum. The dissemination and implementation of effective practices and ongoing work to standardize the approach to management of HDP will be an essential aspect of improving care, which may impact postpartum readmission rates as well as maternal morbidity and mortality (26).

This is one of few studies that specifically examines temporal trends in postpartum readmissions for hypertension. Our study has notable strengths, namely a relatively larger sample size than previous studies, the large number of obstetric providers at this site, and the provision of granular data regarding clinical management during inter-admission and readmission which may foster directions for future research and intervention. A limitation of our study is that despite the very high birthing volume at this institution, we were underpowered to detect a difference in postpartum readmission risk over time. Additionally, it is important to note that our study may not have completely accounted for potential lag in the implementation of changed guidelines into clinical practice; while our study compared the 2012–2013 vs 2014–2017 epochs based on the timing of publication of new guidelines for antenatal management of HDP in 2013, the unadjusted frequencies of postpartum readmissions increased between 2014 and 2015. Our study also only examined patients with HDP diagnosed during delivery admission and did not include patients with HDP diagnosed postpartum. HDP can include a wide range of phenotypes, and our study did not specify the criteria by which individual patients fit the diagnosis of HDP with severe features (e.g. severe hypertension, laboratory abnormalities, symptom criteria). Finally, our study presented data from a single tertiary center and therefore may not be generalizable.

Conclusions

In this diverse cohort of individuals with antecedent HDP, there was no difference in postpartum readmissions for hypertension between 2012–2013 and 2014–2017 despite changes in guidelines regarding clinical management of hypertension in pregnancy. Administration of IV magnesium and receipt of a discharge prescription for hypertensive medication during delivery admission did not reduce the risk of postpartum readmission. Exploratory data on inter-admission and readmission characteristics elucidate trends in management during PPR for HDP that could inform patient counseling and highlight potential future directions for research examining the association between quality of care and risk of postpartum readmission. Understanding changes in clinical management of HDP over time can provide insight into directions for future research seeking to alter the relationship between antenatal management of HDP and postpartum readmission.

Highlights:

• There was no difference in PPR for HDP between 2012–13 and 2014–17

• IV magnesium administration for HDP with severe features remained below 80% across epochs

• Discharge prescriptions for antihypertensive drugs decreased for HDP with severe features

Abbreviations:

HDP

hypertensive disorders of pregnancy

PPR

postpartum readmissions