Rheumatoid arthritis and cardiovascular complications during delivery: a United States inpatient analysis

Abstract

Background and Aims

Persons with rheumatoid arthritis (RA) have an increased risk of obstetric-associated complications, as well as long-term cardiovascular (CV) risk. Hence, the aim was to evaluate the association of RA with acute CV complications during delivery admissions.

Methods

Data from the National Inpatient Sample (2004–2019) were queried utilizing ICD-9 or ICD-10 codes to identify delivery hospitalizations and a diagnosis of RA.

Results

A total of 12 789 722 delivery hospitalizations were identified, of which 0.1% were among persons with RA (n = 11 979). Individuals with RA, vs. those without, were older (median 31 vs. 28 years, P < .01) and had a higher prevalence of chronic hypertension, chronic diabetes, gestational diabetes mellitus, obesity, and dyslipidaemia (P < .01). After adjustment for age, race/ethnicity, comorbidities, insurance, and income, RA remained an independent risk factor for peripartum CV complications including preeclampsia [adjusted odds ratio (aOR) 1.37 (95% confidence interval 1.27–1.47)], peripartum cardiomyopathy [aOR 2.10 (1.11–3.99)], and arrhythmias [aOR 2.00 (1.68–2.38)] compared with no RA. Likewise, the risk of acute kidney injury and venous thromboembolism was higher with RA. An overall increasing trend of obesity, gestational diabetes mellitus, and acute CV complications was also observed among individuals with RA from 2004–2019. For resource utilization, length of stay and cost of hospitalization were higher for deliveries among persons with RA.

Conclusions

Pregnant persons with RA had higher risk of preeclampsia, peripartum cardiomyopathy, arrhythmias, acute kidney injury, and venous thromboembolism during delivery hospitalizations. Furthermore, cardiometabolic risk factors among pregnant individuals with RA rose over this 15-year period.

Structured Graphical Abstract

Acute peripartum cardiovascular complications with rheumatoid arthritis. RA, rheumatoid arthritis; PCOS, polycystic ovary syndrome; GDM, gestational diabetes mellitus; RA, rheumatoid arthritis; NIS, National Inpatient Sample

Rheumatoid arthritis, Pregnancy, Gestational diabetes mellitus, Preeclampsia, Cardiovascular disease

See the editorial comment for this article ‘Rheumatoid arthritis and peripartum cardiovascular complications: focusing on non-traditional cardiovascular risk factors to improve maternal outcomes’, by S. Goldstein and K. Lindley, https://doi.org/10.1093/eurheartj/ehae164.

Introduction

Inflammation is a major driver of cardiovascular (CV) disease pathogenesis and progression.^1,2 Rheumatoid arthritis (RA) is a chronic autoimmune disease with a multitude of articular and extra-articular manifestations.^3,4 Extra-articular manifestations of RA, such as CV and kidney complications, lead to increased morbidity and premature mortality.⁴ Cardiovascular disease, including myocardial infarction and heart failure, is the most common cause of death among individuals with RA.^4–7 It is well established that individuals with RA have an approximately two-fold higher risk of major CV events compared with the general population and up to 50% increased CV mortality risk in the long-term.^4,8 Furthermore RA is known to be associated with worse obstetrics outcomes such as pre-term birth, antepartum haemorrhage, and caesarean delivery.⁹ However, acute CV complications in pregnant individuals with RA at the time of delivery have not been well evaluated. Given the rising maternal mortality rates in the USA due to peripartum CV complications,¹⁰ it is crucial to better understand the impact of non-traditional CV risk factors like RA on pregnancy-related cardiac complications, to help tailor preventive efforts in reducing maternal mortality and morbidity.

This study aimed to evaluate the trends, outcomes, and risk factors of CV complications associated with RA during delivery hospitalizations using a US nationwide real-world population database.

Methods

Study data

This study used data from the National Inpatient Sample (NIS) database from 2004 to 2019. The NIS is managed by the Agency for Healthcare Research and Quality (AHRQ) through a Federal-State-Industry partnership called the Healthcare Cost and Utilization Project (HCUP).¹¹ The NIS contains administrative claims data from more than 7 million inpatient hospitalizations annually in 47 participating states plus the District of Columbia, representing more than 97% of the US population. The NIS provides sample weights in order to use the data to calculate national estimates. Since NIS data are compiled annually, the data can be used for the analysis of disease trends over time using trend weights compiled by the HCUP. For the cost of care, charge to cost ratio supplied by HCUP derived from the Centers for Medicare and Medicaid Services (CMS) was applied to total hospital charges. The data on race/ethnicity are collected by the HUCP participating organizations.¹² The NIS furnishes data regarding income quartile classification, which denotes the estimated median household income of patients residing within specific zip codes. Demographic data based on zip codes are sourced from Claritas and undergoes annual updates.¹³ In accordance with median household income, the study population was categorized into four groups: 0–25th percentile, 26–50th percentile, 51–75th percentile, and individuals in the 75–100th percentile. Supplementary data online, Table S1 presents the data on median household income quartiles as provided by the NIS spanning from 2004 to 2019. This study was deemed exempt from Institutional Review Board approval and informed consent because NIS data are de-identified and publicly available.

Study design and data selection

This study analysed NIS data using International Classification of Diseases, 9th Revision, Clinical Modification (ICD-9-CM) and International Classification of Diseases, 10th Revision, Clinical Modification (ICD-10-CM) claims codes. First, delivery hospitalizations for adult individuals (age ≥ 18) were identified using ICD-9-CM and ICD-10-CM codes (see Supplementary data online, Table S2).^14–18 Among the selected cases, ICD-9-CM codes 7140, 7141, and 7142 and ICD-10-CM codes MO5–MO5.9, MO6.0, MO6.8, and MO6.9 were used to identify delivery hospitalizations in individuals with RA. All diagnosis fields were queried to select and categorize the study population. Individuals with inflammatory arthropathies other than RA were excluded. The study design checklist for studies published using the NIS is provided in Supplementary data online, Figure S1.¹⁹ The study overview and detailed methods flowsheet are presented in Structured Graphical Abstract and Figure 1, respectively.

Figure 1

Study flow chart. NIS, National Inpatient Sample; RA, rheumatoid arthritis; PCOS, polycystic ovary syndrome; GDM, gestational diabetes mellitus; APC, annual percentage change; CI, confidence interval. Estimated are based on the non-weighted sample

Study endpoint

The primary study endpoints were preeclampsia/eclampsia, peripartum cardiomyopathy (PPCM), and acute heart failure. Secondary endpoints included ischaemic and haemorrhagic stroke, pulmonary oedema, cardiac arrhythmias, acute kidney injury (AKI), venous thromboembolism (VTE), length of stay (LOS), and cost of hospitalization. Acute peripartum CV complications were defined as a composite of preeclampsia, PPCM, acute heart failure, stroke, pulmonary oedema, cardiac arrhythmias, AKI, and VTE. Associated procedures and complications were identified using ICD-9-CM and ICD-10-CM codes. Due to low number of cases of eclampsia and the overlapping ICD code of HELLP (haemolysis, elevated liver enzymes, and low platelets) syndrome, which may be classified as either preeclampsia or eclampsia, the choice was made to categorize them as preeclampsia (see Supplementary data online, Table S2).^16,20

Statistical analysis

The primary analysis was conducted on a non-weighted sample that accounted for ∼20% of the discharges from US hospitals. A supplementary analysis was also carried out after incorporating sample discharge weights provided by HCUP. The reason for including a supplementary analysis with sample weights was to obtain a representative sample of the US population. Weighting ensures that the sample is nationally representative and minimizes the risk of obtaining biased or skewed results. A weighted sample improves the precision and accuracy of the estimates by adjusting for various factors such as the sampling design, non-response bias, and hospital and patient characteristics, thus enhancing the generalizability of the results to the entire population.

Descriptive statistics were presented as frequencies with percentages for categorical variables and as medians with interquartile ranges (IQRs) for continuous variables. Baseline characteristics were compared using a Pearson χ² test or Fisher’s exact test as appropriate for categorical variables and the Mann–Whitney U test for continuous variables.

Unadjusted odds ratios (ORs) were derived using Cochran–Mantel–Haenszel test. A multivariable logistic regression model was fitted to test the association of RA with in-hospital outcomes, adjusted for age, race/ethnicity, hospital region, pre-pregnancy comorbidities [chronic hypertension, chronic diabetes, dyslipidaemia, heart failure, chronic kidney disease, coronary artery disease (CAD), polycystic ovary syndrome (PCOS), and obesity], smoking, multiple gestation, gestational diabetes mellitus (GDM), caesarean delivery, median household income, and primary insurance. To examine potential interactions for the association between RA and acute CV complications, we incorporated two-way interaction terms (e.g. RA × GDM, RA × PCOS, RA × chronic diabetes) into the model. We controlled for various confounding factors, including age, race/ethnicity, hospital region, pre-pregnancy comorbidities (chronic hypertension, chronic diabetes, dyslipidaemia, heart failure, chronic kidney disease, CAD, PCOS, and obesity), smoking, multiple gestation, GDM, caesarean delivery,²¹ median household income, and primary insurance.

Given the known association between RA and preeclampsia/eclampsia,²² we utilized the aforementioned multivariable logistic regression model to examine whether the association between RA and acute CV complications persisted after accounting for the presence of preeclampsia/eclampsia. In this additional analysis, we assessed the association of RA with several acute CV complications, including PPCM, acute heart failure, AKI, stroke, pulmonary oedema, cardiac arrhythmias, and VTE, after additional adjustment for preeclampsia/eclampsia.

To check for multicollinearity in the model, diagnostic testing was conducted, with a tolerance of 0.2 or less and a variance inflation factor (VIF) of 5 or more indicating the potential presence of multicollinearity. The previously mentioned multivariable logistic regression model was used to determine whether RA was linked to acute CV complications after the exclusion of known traditional cardiometabolic risk factors. Hence, a supplementary analysis was performed, excluding hospitalizations with PCOS, GDM, pre-existing CAD, chronic hypertension, chronic diabetes, chronic heart failure, and chronic kidney disease, to re-evaluate the relationship between RA and peripartum CV complications.

An additional supplementary analysis aimed at investigating potential disparities in in-hospital outcomes based on race/ethnicity and median household income was conducted. To achieve this, a multivariable logistic regression model was constructed to examine the relationship between race/ethnicity and median household income with and in-hospital outcomes while accounting for various covariates. These covariates included age, hospital region, pre-pregnancy comorbidities (chronic hypertension, chronic diabetes, dyslipidaemia, heart failure, chronic kidney disease, CAD, PCOS, and obesity), smoking, multiple gestation, GDM, and caesarean delivery.

To examine the prevalence trend of RA, cardiometabolic risk factors during pregnancy, and acute CV complications among patients with RA over time, we calculated the annual prevalence change per 100 000 delivery admissions and annual percentage change (APC) with a 95% confidence interval (CI) using non-weighted data. A log-linear regression model was fitted with the outcome of interest as the dependent variable and the year as the independent variable. This approach enabled the measurement of any significant trend, including the degree and direction of the trend that occurred in the prevalence of RA, cardiometabolic risk factors, and acute CV complications in RA patients over the study period.

In order to investigate the association between age and acute CV complications among patients with RA, a cubic spline model was fitted. The analysis was performed using the ‘Splines’ package in R.²³ We used the ‘ns’ function from the ‘Splines’ package to model age as a non-linear variable with three knots. We fitted a linear regression model to the data using the ‘lm’ function from the ‘stats’ package.²⁴

All covariates were selected based on a prior literature review. The missing values present in the dataset are reported in Table 1. The missing values were predominantly present in the race/ethnicity (13.6%) variable which was re-coded with the ‘Other’ category. Given the overall low number of missing data (<1.6%) in other variables, listwise deletion was used and did not include missing data in the logistic regression analysis.

Table 1

Characteristics of delivery hospitalizations with and without rheumatoid arthritis

Variable n (%)	Without RA (12 777 743)	With RA (11 979)
Demographics
Age, yr (median, IQR)	28 (24–32)	31 (27–35)
Race/ethnicity
White	5 827 257 (45.6)	6782 (56.6)
Black	1 565 175 (12.2)	1073 (9.0)
Hispanic	2 446 023 (19.1)	1798 (15.0)
Asian or Pacific Islander	613 410 (4.8)	401 (3.3)
Native American	85 078 (0.7)	152 (1.3)
Other	2 240 800 (17.5)	1773 (14.8)
Hospital regions
Northeast	2 080 315 (16.3)	2127 (17.8)
Midwest	2 676 156 (20.9)	2696 (22.5)
South	4 885 287 (38.2)	4029 (33.6)
West	3 135 985 (24.5)	3127 (26.1)
Pre-existing comorbidities
PCOS	38 534 (0.3)	105 (0.9)
GDM	490 061 (3.8)	708 (5.9)
Chronic diabetes	120 048 (0.9)	231 (1.9)
Dyslipidaemia	18 962 (0.1)	80 (0.7)
Chronic hypertension	83 533 (0.7)	175 (1.5)
Heart failure	7876 (0.1)	21 (0.2)
Chronic kidney disease	1754 (0.0)	<11 (<0.01)a
Coronary artery disease	1491 (0.0)	<11 (0.1)a
Obesity	511 145 (4.0)	810 (6.8)
Smoking	243 743 (1.9)	332 (2.8)
Obstetric characteristics
Multiple gestation	245 186 (1.9)	356 (3.0)
Caesarean delivery	4 033 350 (31.6)	4806 (40.1)
Pre-term birth	942 496 (7.4)	1283 (10.7)
Still birth	87 002 (0.7)	86 (0.7)
Socioeconomic characteristics
Median household income
0–25th percentile	3 434 032 (27.3)	2541 (21.5)
26–50th percentile	3 151 915 (25.1)	2819 (23.9)
51–75th percentile	3 101 625 (24.7)	3102 (26.3)
76–100th percentile	2 882 804 (22.9)	3339 (28.3)
Missing	207 367 (1.6)	178 (1.5)
Primary insurance
Medicare	84 914 (0.7)	491 (4.1)
Medicaid	5 365 292 (42.1)	3578 (29.9)
Private insurance	6 552 271 (51.4)	7341 (61.4)
Self-pay	385 831 (3.0)	173 (1.4)
No charge	19 894 (0.2)	<11 (<0.01)a
Other	349 521 (2.7)	370 (3.1)
Missing	20 020 (0.2)	18 (0.2)

Results presented as n (%).

GDM, gestational diabetes mellitus; HCUP, Healthcare Cost and Utilization Project; IQR, interquartile range; PCOS, polycystic ovary syndrome; RA, rheumatoid arthritis.

^aCounts < 11 are not reported as per HCUP guidelines.

All statistical analyses were performed using Statistical Package for Social Science (SPSS) version 27 (IBM Corp) and R version 3.6 (R Foundation for Statistical Computing, Vienna, Austria).

Results

Hospitalization characteristics of the study population

A total of 12 790 852 unweighted hospitalizations for deliveries were identified in the USA from 2004 to 2019. After excluding other inflammatory arthropathies (n = 1130), a total of 12 789 722 unweighted hospitalizations for delivery were included in this analysis (Figure 1). Of the included individuals, 0.1% had a diagnosis of RA (n = 11 979). Individuals with RA had a higher median (IQR) age of 31 (27–35) years compared with 28 (24–32) years for individuals without RA (P < .01). Chronic hypertension (1.5% vs. 0.7%), chronic diabetes (1.9% vs. 0.9%), obesity (6.8% vs. 4.0%), PCOS (0.9% vs. 0.3%), and dyslipidaemia (0.7% vs. 0.1%) were more frequent in the RA group when compared to individuals without RA (P < .01 for all). The detailed hospitalization characteristics are given in Table 1.

On a weighted sample of data, 63 230 342 delivery admissions were identified from 2004 to 2019. Of the total hospitalizations, 106 437 had an associated diagnosis of RA. Results of the weighted analysis mirrored the results of the unweighted sample by showing a higher age at delivery for RA patients as well as a higher prevalence of hypertension, diabetes, PCOS, and dyslipidaemia (see Supplementary data online, Figure S2 and Supplementary data online, Table S3).

Trends for prevalence of rheumatoid arthritis, preeclampsia, gestational diabetes mellitus, and obesity

During the study duration, the prevalence of RA increased from 60 in 2004 to 139 cases in 2019 per 100 000 deliveries (Figure 2). The prevalence of PCOS increased from 0.2% in 2004 to 1.6% in 2019, obesity increased from 0.2% in 2004 to 11.8% in 2019, and GDM prevalence increased from 3.6% in 2014 to 8.4% in 2019 among individuals with RA (Figure 3). Similarly, the occurrence of acute CV complications during pregnancy in individuals with RA was observed to rise from 7.2% in 2004 to 11.2% in 2019. This indicates a notable increase in the APC of 6.3%, with a 95% CI ranging from 4.9% to 7.5% (Figure 4).

Figure 2

Trends of prevalence of rheumatoid arthritis of during delivery hospitalizations (2004–2019). The graph illustrates a rise in the prevalence of rheumatoid arthritis during delivery hospitalizations in the USA between the years of 2004 and 2019. Estimates are based on the non-weighted sample

Figure 3

Trends of prevalence of polycystic ovary syndrome, gestational diabetes, and obesity among individuals with rheumatoid arthritis during delivery hospitalizations. The graph illustrates a rise in the prevalence of cardiometabolic risk factors during delivery hospitalizations among rheumatoid arthritis patients in the USA between the years 2004 and 2019. Effect sizes are presented as an annual percentage rise in prevalence from 2004 to 2019 with 95% confidence intervals. US, United States; PCOS, polycystic ovary syndrome; GDM, gestational diabetes mellitus; RA, rheumatoid arthritis; APC, annual percentage change; CI, confidence interval

Figure 4

Trend of prevalence of acute peripartum cardiovascular complications among patients with rheumatoid arthritis. The graph illustrates a rise in the prevalence of acute cardiovascular complications during delivery hospitalizations among patients with rheumatoid arthritis in the USA between the years 2004 and 2019. Effect size is presented as an annual percentage rise in prevalence from 2004 to 2019 with 95% confidence intervals. Estimates are based on the non-weighted sample

Cardiovascular complications associated with rheumatoid arthritis

Individuals with RA had a higher incidence of CV complications compared with patients without RA during delivery hospitalizations (Table 2). Patients with RA had higher rates of the development of preeclampsia (7171 vs. 4591 per 100 000 deliveries, P < .01). Similarly, RA was associated with higher rates of PPCM (109 vs. 33, P < .01). Other CV complications, including cardiac arrhythmias, AKI and VTE, were also more common with deliveries in individuals with RA. In an additional analysis, when patients with RA were categorized based on their preeclampsia status, an elevated occurrence of acute peripartum CV complications among individuals with preeclampsia and RA was observed, as opposed to those with RA without preeclampsia (see Supplementary data online, Table S4). The weighted sample of data also mirrored the results of the unweighted analysis and showed a higher prevalence of acute CV complications among patients with RA (see Supplementary data online, Table S5).

Table 2

Complication rates and hospital resource utilization in patients with and without rheumatoid arthritis

Variables	Without RA (12 777 743)	With RA (11 979)	P-value
Complication rates (per 100 000 delivery hospitalizations)
Preeclampsia	4591	7171	<.01
Peripartum cardiomyopathy	33	109	<.01
Heart failure	44	142	<.01
Acute kidney injury	52	117	.02
Stroke	32	42	.57
Pulmonary oedema	39	75	.04
Cardiac arrhythmias	520	1110	<.01
Venous thromboembolism	37	92	<.01
Resource utilization
Length of stay, mean (IQR), days	2 (2–3)	3 (2–3)	<.01
Cost of hospitalization, mean (IQR) $	3718 (2602–5396)	4377 (3026–6457)	<.01

IQR, interquartile range; RA, rheumatoid arthritis.

Odds ratios for in-hospital complications

After adjustment for age, race/ethnicity, comorbidities, insurance, and income, RA remained an independent risk factor of many CV complications during delivery hospitalization (Figure 5). Rheumatoid arthritis was associated with a higher risk of preeclampsia [adjusted odds ratio (aOR) 1.37, 95% CI 1.27–1.47, P < .01]. Similarly, deliveries among individuals with a history of RA were associated with PPCM (aOR 2.10, 95% CI 1.11–3.99, P < .01), AKI (aOR 1.79, 95% CI 1.05–3.01, P = .03), cardiac arrhythmias (aOR 2.00, 95% CI 1.68–2.38, P < .01), and VTE (aOR 1.90, 95% CI 1.05–3.43, P = .03), compared to delivery hospitalizations for individuals without RA.

Figure 5

Unadjusted and adjusted odds ratio for in-hospital complications for persons with rheumatic arthritis (RA) compared to those without RA during delivery admission. Estimated are based on the non-weighted sample

The analysis revealed a lack of significant interactions between RA and cardiometabolic risk factors during pregnancy, except hypertension, which demonstrated a significant interaction (see Supplementary data online, Table S6). In our stratified analysis based on chronic hypertension status, we observed a notable and significant increase in the rates of acute peripartum CV complications among patients with RA in comparison to those individuals without a history of chronic hypertension (OR 10.27, 95% CI 7.40–14.25, P < .01) (see Supplementary data online, Figure S4). A significant racial/ethnic and socioeconomic disparity was observed as Black individuals with RA had higher odds of developing acute CV complications. Conversely, individuals in the highest quartile of income had lower odds of developing acute CV complications with RA at the time of delivery admission (see Supplementary data online, Table S7).

All variables included in the model had a tolerance of more than 0.2 and VIF of <5 (e.g. hypertension tolerance 0.98, VIF, 1.02, chronic diabetes tolerance 0.99, VIF, 1.02, GDM tolerance 0.97, VIF, 1.03, PCOS tolerance 0.99, VIF, 1.01), indicating the absence of multicollinearity. An additional analysis was conducted, which involved adjusting for preeclampsia/eclampsia and produced results that were consistent with our primary analysis, revealing a significant association between RA and acute peripartum CV even after accounting for preeclampsia (see Supplementary data online, Table S8). Furthermore, a sensitivity analysis was performed by excluding patients with PCOS, GDM, chronic hypertension, chronic diabetes, chronic heart failure, dyslipidaemia, chronic kidney disease, and CAD. This analysis also revealed a significant association between RA and acute peripartum CV complications, thus strengthening the evidence for an independent association between RA and these complications (see Supplementary data online, Tables S9–S11).

After applying sample weighting, the multivariable logistic regression analysis showed results consistent with the unweighted sample, with few exceptions. Notably, the larger sample size led to significantly higher odds of acute heart failure among patients with RA. Additionally, the confidence intervals were narrower due to the increased sample size (see Supplementary data online, Figure S3 and Supplementary data online, Table S12).

Relationship between age and acute peripartum cardiovascular complications among patients with rheumatoid arthritis

A linear regression model examined the relationship between age and acute CV complications, using a cubic spline function with three knots to capture potential non-linear effects. The effect size for the highest knot of the cubic spline model indicated that for every one-year increase in age above 35, there was a percentage increase of ∼13.9% in the odds of experiencing an acute CV complication among patients with RA (P < .01) (Figure 6).

Figure 6

Relationship between age and acute peripartum cardiovascular complications among patients with rheumatoid arthritis. One-year increase in age above 35, there was a percentage increase of ∼13.9% in the odds of experiencing an acute cardiovascular complication among patients with rheumatoid arthritis (P < .01). Results are based on the non-weighted sample

Resource utilization

In terms of resource utilization, length of hospital stay was higher for deliveries among individuals with RA vs. individuals without RA (3 days vs. 2 days, P < .01). Similarly, deliveries for individuals with RA had a higher cost of hospitalization ($4377 vs. $3718, P < .01) (Table 2). Similar results were obtained for weighted sample, where RA delivery admissions were associated with significant increased hospital LOS and increase in cost of hospitalization (see Supplementary data online, Table S5).

Discussion

This large contemporary, retrospective cohort study, including 12 million delivery hospitalizations in the USA with over 11 000 among individuals with RA, revealed the following principal findings: (i) a diagnosis of RA is independently associated with higher CV complications during delivery hospitalizations including the development of preeclampsia, PPCM, cardiac arrhythmias, AKI, and VTE; (ii) RA during delivery admissions is associated with increased cost and length of delivery hospitalizations; and (iii) the prevalence of PCOS, obesity, GDM, and acute CV complications among individuals with RA during delivery hospitalizations is increasing in the USA over a period of 15 years.

Association of rheumatoid arthritis with cardiovascular disease

Chronic inflammation is known to be a driver of atherothrombosis and CV risk.^1,2 As such, current US prevention guidelines consider chronic inflammatory conditions such as systemic lupus erythematosus (SLE), RA, psoriasis, and human immunodeficiency virus to be ‘risk-enhancing’ factors that warrant intensified preventive efforts to reduce the risk of long-term CV complications.²⁵ Persons with RA are at heightened risk for CV disease,^26–28 including a 50% greater risk for CV mortality compared to persons without RA.²⁹ Despite this, current risk estimation tools do not adequately capture CV risk in persons with RA, and even RA-specific CV risk calculators have not performed better than risk calculators developed in the general population.³⁰

Among pregnant persons, having a chronic inflammatory condition is also known to be associated with an increased risk of developing pregnancy-associated complications such as preeclampsia.^31,32 Previous studies have investigated CV risks associated with SLE in pregnancy.^33,34 This current study, now focusing on RA specifically, represents a significant addition to the existing literature by showing that RA is associated with acute peripartum CV events, an association that is independent of traditional risk factors, GDM,³⁵ as well as preeclampsia.²⁰

It can be postulated that the association of RA with acute CV complications at pregnancy delivery is multifactorial.^7,8,36,37 This study provides insights into the cardiometabolic health of pregnant persons by showing that individuals with RA were older and had a higher prevalence of chronic hypertension, obesity, and obesity-related comorbidities including pre-existing diabetes and GDM at pregnancy delivery. After adjusting for these cardiometabolic comorbidities in these analyses, RA remained associated with elevated peripartum CV risks. Furthermore, it is imperative to underscore that on supplementary analysis even following the exclusion of patients presenting with cardiometabolic risk factors such as chronic hypertension, GDM, pre-existing diabetes, chronic kidney disease, and CAD, a significant association between RA and acute CV complications persisted. These findings strongly suggest the presence of a potential independent association of RA, necessitating further in-depth investigations. They also indicate that RA itself may serve as a risk-enhancing factor, rendering pregnancies at higher risk for CV complications. Chronic hypertension is common in the RA population,³⁸ and is a risk factor for superimposed preeclampsia. However, these analyses further adjusted for preeclampsia and found that the RA was still independently associated with CV complications. Furthermore, this analysis showed that chronic hypertension in individuals with RA is associated with a 10-fold higher risk of acute peripartum CV complications compared to individuals who do not have chronic hypertension at the time of delivery. Though this study does not have data on disease activity, it was postulated that the underlying inflammatory state might have a predominant role to play as the association of RA with peripartum CV complications remained a significant event after the adjustment of traditional cardiometabolic risk factors.

The increased prevalence of cardiometabolic risk factors could be due to the use of steroids in these patients which has shown to be associated with a two-fold higher risk of CV events.^6,39 When pregnancy is planned or identified, certain disease-modifying antirheumatic drugs (DMARDs) such as methotrexate and leflunomide are discontinued while other medications such as hydroxychloroquine, sulfasalazine, azathioprine, or anti-tumour necrosis factor alpha agents are safe to continue as indicated. It should be noted that prior studies evaluating the association of immunosuppressive medications with CV complications (in the non-pregnant population) have revealed conflicting results, with some suggesting that reducing RA inflammatory activity is beneficial while others show that they lead to long-term higher CV complications.^8,40 We were not able to account for past or current medication use in our analysis due to limitations of the dataset.

Worsening trends in cardiometabolic health among pregnant individuals in the USA

Prior studies have described concerning population-level trends of worsening cardiometabolic health of pregnant individuals in the USA, and this is also highlighted by this current analysis. This study shows an exponential increase in the prevalence of PCOS, GDM, obesity, and acute peripartum CV complications in patients with RA during 15 years from a nationally representative cohort. This study supports the findings of prior studies that have reported a significant decline CV health among reproductive-age individuals.^41,42 Furthermore, presence of cardiometabolic risk factors has now been shown to be associated with acute CV peripartum complications as well.⁴² Similarly, due to better management strategies such as the development of biologic disease-modifying drugs and improvement of the overall health of persons with RA of reproductive age, the trend of increased prevalence of RA among pregnant individuals has also been observed.⁴³ Hence, this study provides the most recent available data on population-level trends of RA that have not been described before and warrant an urgent intervention to prevent CV disease among pregnant individuals.

These current findings regarding the trend of RA prevalence are consistent with prior research. For example, a study conducted by Shi et al., which examined the global prevalence, incidence, and years lived with disability related to RA, indicated that the global burden of RA has been on the rise over the past three decades and is projected to continue increasing.⁴⁴ Likewise, results from a population-level study conducted by Myasoedova et al.⁴⁵ demonstrated that the incidence of RA among women increased by 2.5% from 1995 to 2007 (with a 95% CI of 0.3% to 4.7% and a P-value of .02). However, this upward trend in RA incidence was not observed among men. Breaking down the incidence rates of RA among women of reproductive age further, this current study observed that among individuals aged 18–34, the incidence rate was 13.8 per 100 000 population. For those aged 35–44, the incidence rate rose to 55 per 100 000, and for individuals aged 45–54, the incidence rate reached 62.4 per 100 000 population. These findings provide important insights into the changing landscape of RA incidence in reproductive-aged individuals.

Varied responses of autoimmune inflammatory conditions to hormonal changes during pregnancy and adverse cardiovascular outcomes

Autoimmune inflammatory conditions may exhibit varied responses to the hormonal changes that occur during pregnancy. For instance, the response to sex hormones differs between RA and SLE. While oestrogen and pregnancy can mitigate the development of RA, SLE may flare during pregnancy and in response to oestrogen.⁴⁶ In a study by Singh and colleagues among pregnant individuals, it was observed that there were increased risks of several adverse outcomes, with the relative risks often being greatest for SLE. For example, individuals with RA or SLE were more likely to require rehospitalization, with the greatest risk occurring at <6 months post-partum (4% in RA and 6% in SLE).⁴⁷ Similarly, SLE was associated with 4.4 times higher odds of developing PPCM compared with a two times higher risk of PPCM with RA in our study.⁴⁸ These findings suggest that while autoimmune inflammatory states may elevate CV risk, they may not share the same level of risk, with some autoimmune inflammatory conditions causing more severe disease. These findings underscore the need for further translational studies to understand why the risk of adverse CV complications differs between different autoimmune inflammatory states and what mitigation measures may be taken.

Implication of study findings for CV prevention in persons with rheumatoid arthritis

Considering the present study findings, individuals with RA should be counselled on the possible risk of developing acute CV complications during pregnancy and peripartum. In addition to managing the traditional risk factors of hypertension, diabetes, obesity, and dyslipidaemia, the underlying inflammatory state due to RA also needs to be controlled. Furthermore, patients need to be counselled on elevated CV risk with advanced age, as risk of CV complications with RA increase exponentially after the age of 35 years. Hence, urgent steps are needed for pre-pregnancy screening to identify cardiometabolic risk factors and focus on efforts to prevent and aggressively treat these risk factors once identified. The guidelines from the European Society of Cardiology (ESC) regarding the management of CV disease during pregnancy suggest that expert centres with a multidisciplinary team should handle pre-pregnancy counselling and management throughout pregnancy and the delivery period.⁴⁹ In terms of preconception evaluation, responsibility could be shared between the general obstetrician and the rheumatologist. The obstetrician should take a thorough cardiac history and refer patients who require further evaluation to a rheumatologist for risk stratification and management of their RA during pregnancy. The rheumatologist can then collaborate with the obstetrician and maternal–foetal medicine specialist to provide appropriate care during pregnancy, post-partum, and lactation. Rheumatologists can provide continued management of the underlying inflammatory state and monitor for any RA flare-ups during pregnancy. Cardiac assessment may also be necessary depending on individual risk factors, and in some cases, involve the engagement of a cardiologist as part of a multidisciplinary Cardio-Obstetrics team. Furthermore, peripartum RA care can also help address individuals’ hesitancy to continue DMARDs during pregnancy and lactation and help control RA disease activity. After delivery, efforts for long-term CV prevention need to be implemented as well given the known future risk of CV disease. There is a growing need for specialists in Cardio-Rheumatology (as well as Cardio-Obstetrics), and there has been an emergence of such training programmes to address these gaps.^50,51

Racial/ethnic and socioeconomic disparities

This present study’s findings align with previous research, highlighting significant disparities in both racial/ethnic and socioeconomic factors.^20,52 In this analysis, it was observed that Black individuals faced a significantly higher risk of developing acute CV complications with RA. Conversely, individuals from lower socioeconomic backgrounds, particularly those with the lowest income levels, exhibited higher rates of acute CV complications. Prior studies have also reported an interaction between racial/ethnic background and socioeconomic status.^52,53 These findings emphasize the urgent need for measures to address disparities in healthcare access and pre-pregnancy counselling in order to reduce the escalating maternal mortality rate. It is highly likely that these disparities reflect underlying biases, structural racism, and limited access to care for socially disadvantaged groups who are both undertreated and under screened. Consequently, these findings underscore the immediate necessity for comprehensive initiatives to meet the unmet needs of at-risk populations in the USA, with the aim of reversing the disturbing trends in increasing maternal mortality.

A multidisciplinary team-based approach is crucial for these high-risk patients, and collaboration among obstetricians, rheumatologists, maternal–foetal medicine specialists, and cardiologists can ensure appropriate care and monitoring to prevent acute CV complications and reduce maternal morbidity and mortality. Practical guidelines and algorithms can be developed to facilitate the implementation of this approach in clinical practice. In light of these study findings, these measures are urgently needed.

Resource utilization

This study also reported an increase in the LOS and consequently the cost of hospitalization at the time of delivery in individuals with a diagnosis of RA. It can be postulated that increased hospital resource utilization is also a surrogate for the greater rates of adverse CV events during hospitalization. Moreover, RA complications are a source of significant economic burden on the healthcare system. For example, one study report estimated that the average annual direct and indirect cost for RA was 3.6 million dollars.⁵⁴ The present reported cost analysis underscores the impact of the cumulative cost of RA hospitalization during delivery and may be helpful information for policymakers.

Study strengths and limitations

This study has many strengths as it analysed a large multiethnic nationally representative sample of the US delivery population which allowed sufficient statistical power to examine CV complications associated with deliveries among individuals with RA over a 15-year period. However, these study findings should be considered in the context of several important limitations. The NIS is an administrative claim-based database that uses ICD codes for diagnosis; although diagnosis codes less prone to error were used, coding errors cannot be excluded. There may be under coding and underreporting of RA, which may preclude accurate capture of the true disease prevalence in this population. Similarly, a stratified analysis based on seropositivity for RA could not be performed due to lack of specific codes. Also, important variables such as gestational age at delivery, previous history of preeclampsia/eclampsia, number of prenatal care check-ups, or pre-pregnancy body mass index were not able to be included in the regression models due to the lack of specific ICD codes for these diagnoses. This study was also not able to capture medications such as pre-pregnancy use of DMARDs, RA disease activity during pregnancy, or RA severity pre-pregnancy.

There was also a change in the methodology of NIS to improve national estimates in 2012 and a change in coding practices from ICD-9 to ICD-10 in quarter four of 2015. This may have led to different estimates of disease prevalence in 2012 or 2015, although the trends observed were present across the full study period. Trends in the prevalence of RA, PCOS, obesity, GDM, and acute CV complications over time may be due to better capturing of these diagnoses by ICD coding. Nevertheless, the true prevalence of cardiometabolic risk factors may be underestimated given the reliance on ICD coding for diagnosis. This study could not evaluate and report temporal trends for all cardiometabolic risk factors of pregnancy such as chronic hypertension, chronic diabetes and dyslipidaemias among RA individuals due to cell counts being <11 which are not reportable per HCUP guidelines. In the reporting of cardiac arrhythmia rates, a notable constraint was encountered. Specifically, the ability to present rates for distinct cardiac arrhythmia types, such as supraventricular tachycardia or ventricular tachycardia, was hindered due to a predominant coding pattern of non-specific arrhythmias. This limitation restricted the capacity to undertake a comprehensive investigation of the specific types of arrhythmias prevalent among individuals with RA during pregnancy.

Another noteworthy limitation of this study pertains to inherent limitations of NIS, which records information solely on inpatient discharges, with each admission treated as an independent event. Consequently, this study was unable to definitively identify individuals contributing to the dataset with multiple pregnancies. The lack of individual-level data precluded the possibility of conducting a sensitivity analysis, particularly regarding the assessment of first deliveries or primiparous individuals. Additionally, it is important to note that this study lacked specific data regarding gestational age. Moreover, it is crucial to acknowledge that the timing of recorded outcomes in this study was not subject to centralized adjudication but, instead, relied upon the ICD coding system. This limitation hampers the ability to interpret these results comprehensively, as preventive measures against maternal morbidities are often tailored based on obstetric history for whom data are not available in this dataset. NIS samples are not designed to follow individuals longitudinally, so long-term outcomes could not be assessed from the present dataset. Thus, only information at the time of hospital delivery was available for analysis and that has important implications; for instance, PPCM is most likely diagnosed 1–4 weeks post-partum which was not captured in the current analysis.

This study showed an independent association of RA with acute CV complications after adjustment of preeclampsia/eclampsia. It is possible that preeclampsia might be a mediating variable in the causation of CV complications. In the supplementary analysis, a noteworthy finding is the impact of preeclampsia, which plays a substantial role in driving the occurrence of acute CV complications among individuals with RA. Further meticulously designed prospective studies are required to examine the potential mediating role of preeclampsia in the development of acute CV complications in individuals with RA. This study also did not conduct adjustments for multiple comparisons in our analysis. Consequently, there is a potential risk that approximately one out of every 20 significant associations we observed could be a false positive result. Additionally, like any observational study, association does not mean causation and conclusions should be drawn cautiously.

Conclusions

In conclusion, this study reported higher CV complication rates including preeclampsia, PPCM, AKI, VTE, and cardiac arrhythmia among pregnant individuals with RA, compared to those without RA, during delivery hospitalizations in the USA over 15 years. Moreover, there was a trend of increasing prevalence of RA, obesity, and related comorbidities including GDM and PCOS in the USA. Furthermore, acute CV complication among RA patients was also shown to have a significant uptrend during the study period. A multidisciplinary approach to the management of RA pregnancies in liaison with a rheumatologist, cardiologist, and high-risk obstetrician/maternal foetal medicine specialist might be helpful to improve outcomes. Furthermore, focused studies are needed to best strategize for the prevention and management of acute and long-term pregnancy-associated CV complications among individuals with RA.

Supplementary data

Supplementary data are available at European Heart Journal online.

Declarations

Disclosure of Interest

E.D.M. reports advisory board participation for Amgen, AstraZeneca, Boehringer Ingelheim, Edwards Life Science, Esperion, Medtronic, Merck, New Amsterdam, Novartis, Novo Nordisk, and Pfizer. No other authors report any conflicts of interest.

Data Availability

The data presented in this article originate from publicly available source, the Healthcare Cost and Utilization Project (HCUP).

Funding

E.D.M. is supported by the Amato Fund for Women’s Cardiovascular Health research at Johns Hopkins University and by an American Heart Association grant 946222. A.S.M. is supported by the National Institutes of Health grant KL2TR003099.

Ethical Approval

Ethical Approval was not required because NIS data are de-identified and publicly available.

Pre-registered Clinical Trial Number

None supplied.