Maternal and Fetal Outcomes Associated With Infective Endocarditis in Pregnancy

Abstract

Background

Infective endocarditis (IE) is a rare but serious infection that complicates pregnancy. Little is known about IE management and outcomes in this population.

Methods

The National Readmissions Database was used to obtain data between October 2015 and October 2018. Billing codes identified admissions for IE in female patients of reproductive age. Demographic characteristics, comorbidities, and outcomes were compared between patients with maternity-associated and nonmaternity-associated IE and obstetric patients who delivered with and without IE. Weighted regressions were used to examine outcomes in adjusted models.

Results

We identified 12 602 reproductive-aged female patients with a diagnosis of IE, of which 382 (weighted national estimate, 748) were maternity-associated. Of these cases, 117 (weighted national estimate, 217) occurred during a delivery admission. Compared with patients with nonmaternity-associated IE, maternity-associated infection was associated with younger age (mean, 29.0 vs 36.6 years; P < .001), Medicaid coverage (72.5% vs 47.2%; P < .001), and drug use (76.2% vs 59.8%; P < .001). Mortality was comparable (8.1% vs 10.6%; adjusted rate ratio [aRR], 1.03; 95% confidence interval [CI]: .71–1.48). Compared with patients who delivered without IE, IE complicating delivery was associated with worse maternal and fetal outcomes, including maternal mortality (17.2% vs <0.01%; aRR, 323.32; 95% CI: 127.74–818.37) and preterm birth (55.7% vs 10.1%; aRR, 3.61; 95% CI, 2.58–5.08).

Conclusions

Maternity-associated IE does not appear to confer additional risk for adverse outcome over nonmaternity-associated infection. Patients who deliver with IE have worse maternal and fetal outcomes than those whose deliveries are not complicated by IE.

Infective endocarditis (IE) during pregnancy is rare. Using a large cohort, we found that maternity-associated IE does not confer additional risk over nonmaternity-associated infection. However, IE during an admission for delivery was associated with poor maternal and fetal delivery outcomes.

Infective endocarditis (IE) is a life-threatening infection with an incidence rate of 1.4–6.2 cases per 100 000 person-years in developed countries [1]. In the United States, the IE incidence has increased [2], reflecting an increase in intravenous drug use (IVDU), hemodialysis dependence, long-term indwelling cardiovascular prostheses, and elderly patients with age-related cardiovascular disease [1, 3]. As the demographic characteristics of patients with IE shift, evidence suggests that endocarditis incidence is increasing in women [4]. Although rare, IE can occur during pregnancy and is associated with both maternal and fetal morbidity and mortality [5, 6]. Pregnancy in general is a time of greater susceptibility to infection given relative immunosuppression. It is also a time in which the pregnant body undergoes considerable hemodynamic change with respect to cardiovascular function and a time of increased cardiovascular risk, both during pregnancy and the weeks following delivery. However, the available literature provides little insight into this clinical entity, and the incidence, management, and outcomes are poorly understood. Using a national cohort, we had the following 2 objectives: to describe characteristics, clinical outcomes, and healthcare resource use of patients with maternity-associated IE and to evaluate maternal and fetal outcomes of deliveries complicated by IE.

METHODS

Data Source and Study Population

Data were obtained from the National Readmissions Database (NRD), a project of the US Agency for Healthcare Research and Quality Healthcare Cost and Utilization Project [7]. The NRD is an all-payer administrative dataset designed to be a representative sample of short-stay, nonfederal hospital discharges across the United States (excluded hospitals include Veterans Affairs and Department of Defense facilities, rehabilitation hospitals, and long-term acute care hospitals).

As of 2018, the NRD records data from 28 states, accounting for 58.7% of such hospital discharges. It includes weighting and stratification variables to allow for generation of national estimates (the estimated outcomes for patients in all 50 US states). The NRD includes patient demographics, hospital-level variables, International Classification of Diseases (ICD) diagnosis and procedure codes, length of stay, discharge disposition, and patient billing/charges. Readmissions can be identified for patients, provided they occur in the same US state and calendar year.

We limited our study to discharges that occurred between October 2015 and October 2018. These dates were selected based on data availability and to ensure use of the tenth revision of International Classification of Diseases, Clinical Modification, codes (ICD-10) adopted in October 2015. We excluded hospitalizations with discharges in November and December to allow for 60 days of follow-up. Patients were included in the study if they were female sex, of reproductive age (aged 12–55 years), and had at least 1 ICD-10 code indicating IE diagnosis (Supplementary Table 1). For patients with multiple IE-related admissions, their first admission in the dataset was considered their index hospitalization. All analyses were performed using the index hospitalization and readmissions that occurred within 60 days of discharge. ICD-10 codes pertaining to pregnancy (Supplementary Table 1) was used to identify hospitalizations for maternity-associated IE (Figure 1A). A subset of these pregnancy-related codes was used to identify pregnant patients who were hospitalized for delivery regardless of IE status (Figure 1B). Our first analysis compared patients with maternity-associated and nonmaternity-associated IE. Our second analysis compared patients with IE complicating delivery to those without IE at delivery.

Figure 1.

Sample selection and comparison groups.

Definitions

A combination of data provided directly in the NRD datasets and what could be constructed using ICD-10 codes were used to identify outcomes of interest. We divided primary payor into 5 categories: Medicaid (government-funded and covers low-income individuals), Medicare (government-funded and covers patients aged >65 years, as well as younger individuals with disabilities), private insurance (commercial providers), self-pay, and other insurance. For the comparison of maternity-associated IE to nonmaternity-associated IE, outcomes included in-hospital mortality, valve repair and replacement procedures, intubation/ventilation, occurrence of a thromboembolic event, length of stay, and total inpatient costs. For the comparison of IE complicating delivery to deliveries not complicated by IE, obstetric outcomes including stillbirth, preterm birth, and cesarean (rather than vaginal) delivery were identified, as were maternal outcomes including in-hospital mortality, occurrence of any of the established severe maternal morbidity composite outcomes [8, 9], length of stay, and total inpatient costs. A complete list of these outcomes and the codes used in their ascertainment can be found in Supplementary Table 2.

Statistical Analyses

Because the NRD is a weighted and stratified dataset intended to be representative of admissions across the United States, statistical analyses were performed using methods appropriate for such data. Categorical variables were analyzed using weighted χ² tests. Continuous variables were compared using weighted linear regression. Multivariate regression models were used to compare outcomes. In selecting covariates for adjustment, we considered the possibility for confounding from several potential sources. These included differences in demographics (age) and socioeconomic factors (zip-code, median household income, and primary payor); these factors were salient given that pregnancy itself affects access to health insurance in the United States (ie, Medicaid eligibility), which can affect access to care. Because the combination of pregnancy and IE is an uncommon situation that requires care from a diverse range of subspecialists, we hypothesized that these patients might be treated preferentially at larger, referral hospitals, and adjusted for bed size, teaching, and urban status. With respect to medical history, we began with a previously validated comorbidity index for pregnancy, updated for ICD-10 by the current authors [10]. Because some of the comorbid conditions are specific to pregnancy (eg, twin pregnancy), these were removed from comparisons between maternity-associated and nonmaternity-associated stays. Similarly, there were some conditions included in the obstetric comorbidity index (eg, cardiac valvular disease), which could itself be the result of IE, and these were removed as well. Substance use in pregnancy is typically categorized as being related to alcohol, tobacco, or other substances in the comorbidity index. However, given the clear potential differences in IE risk by substance in this study, we identified the specific category of substance used whenever possible (eg, opioid, cannabis, sedative). Finally, we adjusted for history of prosthetic valves and other cardiac procedures for all comparisons and for etiologic organisms (for the comparison of IE-associated hospitalizations only; Supplementary Table 3).

Weighted logistic regression was used for binary outcomes, and Poisson and gamma/log-link models were used for length of stay and inpatient costs, respectively. Because the odds ratios generated by logistic regression are commonly misinterpreted as relative risks, and for common outcomes this misinterpretation can be a significant overestimation of the risk (ie, the odds ratio is typically numerically larger than the relative risk), they were converted to relative risks [11]. The odds ratios are available in Supplementary Table 5. Due to the large imbalance between cases and controls in the comparison of delivery hospitalizations, we additionally performed propensity score matching using the same covariates as above and matching each case to 1000 controls given the rarity of mortality. These results are presented in Supplementary Table 6. There were minimal missing data for most variables, including primary payer and zip-code household income (<1%), and thus these variables had the most common value imputed for observations with missing values. For etiologic organism, unknown organism was treated as a discrete category in the analysis. A 2-sided alpha value of 0.05 was prespecified to be statistically significant. Data were analyzed using Stata Statistical Software, Version 16.1 (Statacorp, College Station, TX). Given this was a retrospective analysis using an existing limited dataset, it was determined to be exempt from review by the Duke Health Institutional Review Board.

RESULTS

Sample Population

During our study period, there were 57.1 million hospital discharges recorded in the NRD. We identified 12 602 female patients of reproductive age with a diagnosis of IE, of whom 382 (weighted national estimate, 748) met the criteria for maternity-associated IE. The majority of maternity-associated IE occurred antepartum (39.1%). Postpartum IE accounted for 19.0% and early/abnormal pregnancies constituted 12.5%. Delivery cases comprised 117 cases (weighted national estimate, 217), that is, 29.4% of maternity-associated IE occurred during an admission for delivery. During the same period, there were approximately 5 million admissions for delivery in patients without IE.

Comparing Maternity-Associated With Nonmaternity-Associated IE

Population Characteristics

Patients with maternity-associated IE were significantly younger than those with nonmaternity-associated disease (29.0 vs 36.6 years; P < .001; Table 1). Although the distribution of patients across median household income quartiles did not differ, patients with maternity-associated infection were more likely to be covered by Medicaid (72.5% vs 47.2%; P < .001). There were lower rates of medical comorbidities such as diabetes, hypertension, and chronic kidney disease among maternity-associated IE admissions. In addition, rates of cardiac comorbidities were lower in patients with maternity-associated IE, including history of valve replacement, the presence of a cardiac prosthesis, and ischemic heart disease. Although rates of drug use were high in both groups, they were significantly higher in patients with maternity-associated infection (76.2% vs 59.8%; P < .001). The most common etiologic organism in both groups was Staphylococcus aureus, which accounted for approximately half of all IE cases. This was followed by polymicrobial infections, Streptococcus, and Enterococcus. Methicillin-resistant S. aureus was more commonly associated with maternity-associated IE (29.8% vs 22.7%; P < .001).

Table 1.

Baseline Patient Characteristics of Infective Endocarditis Hospitalizations Stratified by Maternity Association

Characteristic	Overall IE (N = 12 602) (Nweighted = 24 132)	Nonmaternity-Associated IE (N = 12 220) (Nweighted = 23 384)	Maternity-Associated IE (N = 382) (Nweighted = 748)	P Value
	Mean (Standard Deviation) or Count (%)
Age, years	36.4 (7.7)	36.6 (7.7)	29.0 (3.8)	<.001
Primary expected payera				<.001
Medicaid	11 563 (48.0)	11 023 (47.2)	540 (72.5)
Private insurance	4324 (18.0)	4211 (18.0)	113 (15.2)
Medicare	3678 (15.3)	3666 (15.7)	12 (1.6)
Self-pay	4324 (18.0)	4211 (18.0)	113 (15.2)
No charge	611 (2.5)	599 (2.6)	12 (1.6)
Other	802 (3.3)	774 (3.3)	29 (3.9)
Median household income national quartile for patient zip-codeb				.48
0 to 25th percentile	9260 (38.9)	8981 (39.0)	279 (37.8)
25th to 50th percentile	6809 (28.6)	6603 (28.6)	206 (28.0)
50th to 75th percentile	4941 (20.8)	4794 (20.8)	147 (19.9)
75th to 100th percentile	2776 (11.7)	2671 (11.6)	106 (14.3)
Comorbidities
General medical
Pregestational diabetes	3459 (14.3)	3441 (14.7)	18 (2.4)	<.001
Chronic hypertension	7001 (29.0)	6939 (29.7)	62 (8.3)	<.001
Chronic kidney disease	3741 (15.5)	3721 (15.9)	20 (2.7)	<.001
Living with human immunodeficiency virus	c	406 (1.7)	c	.17
Drug use	14 561 (60.3)	13 991 (59.8)	570 (76.2)	<.001
Specific substance (if known, allowing for duplicates)
Opioid	10 872 (45.1)	10 445 (44.7)	427 (57.0)	<.001
Cannabis	1756 (7.3)	1690 (7.2)	67 (8.9)	.33
Sedative	517 (2.1)	487 (2.1)	30 (4.0)	.02
Cocaine	3074 (12.7)	2959 (12.7)	115 (15.4)	.14
Stimulant	3504 (14.5)	3383 (14.5)	121 (16.2)	.43
Other	3630 (15.0)	3507 (15.0)	122 (16.4)	.51
Alcohol use	1263 (5.2)	1248 (5.3)	15 (2.0)	.009
Tobacco use disorder	12 629 (52.3)	12 237 (52.3)	392 (52.4)	.97
Cardiac
Congenital heart disease	1583 (6.6)	1527 (6.5)	56 (7.5)	.47
Valve replacement	2303 (9.5)	2258 (9.7)	45 (6.0)	.03
Other cardiac prosthesis	2417 (10.0)	2391 (10.2)	26 (3.5)	<.001
Ischemic heart disease	1632 (6.8)	1617 (6.9)	15 (2.0)	<.001
Congestive heart failure	2200 (9.1)	2160 (9.2)	40 (5.4)	.03
Etiologic organism				<.001
Methicillin-sensitive Staphylococcus aureus	5642 (23.4)	5418 (23.2)	224 (29.9)
Methicillin-resistant Staphylococcus aureus	4508 (23.0)	4324 (22.7)	184 (29.8)
Polymicrobic	2244 (9.3)	2172 (9.3)	72 (9.7)
Streptococcus	2107 (8.7)	2065 (8.8)	41 (5.5)
Enterococcus	956 (4.0)	932 (4.0)	24 (3.2)
Other Staphylococcus	c	608 (2.6)	c
Gram-negative rods	578 (2.4)	567 (2.4)	12 (1.6)
Fungal	c	166 (0.7)	c
Other	2648 (11.0)	2590 (11.1)	58 (7.7)
Unknown	2927 (12.1)	2807 (12.0)	120 (16.0)

Abbreviation: IE, infective endocarditis.

^a Missing data: 28 observations.

^b Missing data: 186 observations.

^cN <10, value cannot be displayed due to data use restrictions.

Antepartum, delivery, and postpartum cases of maternity-associated IE were also compared (Supplementary Table 4). Although population characteristics were largely similar across these 3 groups, antepartum IE was associated with higher rates of drug and tobacco use, while postpartum IE cases were more likely associated with pregestational diabetes.

Outcomes

Outcomes of maternity-associated and nonmaternity-associated IE are described in Figure 2. Including readmissions within 60 days of discharge from the patient’s first IE-related hospital stay, in-hospital mortality rates were similar for maternity-associated (8.1%) and nonmaternity-associated (10.6%) infection. After adjustment, the difference continued to be not statistically significant (adjusted rate ratio [aRR], 1.03; 95% confidence interval [CI]: .71–1.48). Similarly, rates of valve procedures and mechanical ventilation were similar in both unadjusted and covariate-adjusted analyses. Rates of cardiac valve procedures did not differ significantly between these 2 groups. Thromboembolic events were common. However, while the difference between pregnant and nonpregnant patients was not statistically significantly different in the unadjusted analyses, it was significant in the adjusted analysis (54.7% maternity-associated vs 58.1% nonmaternity-associated; aRR, 0.73; 95% CI: .63–.84). Last, length of stay and inpatient cost were similar for both groups.

Figure 2.

Unadjusted and adjusted outcomes among infective endocarditis patients stratified by maternity association. All outcomes include readmissions within 60 days of hospital discharge from the patient's first endocarditis-related hospitalization. All outcomes adjusted for age, primary payor, zip-code income quartile, hospital bed size, urban/rural, teaching status, alcohol use, asthma, ischemic heart disease, chronic kidney disease, congenital heart disease, cystic fibrosis, diabetes mellitus, drug use and specific drug class used, human immunodeficiency virus/AIDS, chronic hypertension, obesity, pulmonary hypertension, sickle cell disease, systemic lupus erythematosus, tobacco use, etiologic organism, history of mechanical valve, and history of other heart procedure.

When comparing antepartum, delivery, and postpartum maternity-associated IE, 60-day mortality was highest in the delivery subgroup (16.9%; P = .004), while rates of valve replacements were highest in postpartum cases (22.2%; P = .03; Supplementary Table 4). Patients with postpartum IE experienced cerebrovascular thrombosis (23.7%; P < .001), gastrointestinal clots (11.9%; P = .007), and obstetric-specific clots (12.9%; P < .001) at significantly higher rates. Patients with antepartum IE had significantly lower inpatient costs. Results were similar when presented as odds ratios (Supplementary Table 5).

Comparing Delivery With and Without IE

Population Characteristics

In our study sample, IE incidence complicating delivery was 1:45 560 deliveries. There was no difference in age between patients delivering with and without IE (28.7 vs 28.7 years; P = .93; Table 2). Patients with IE complicating delivery were more likely to live in areas with lower median household income, with 43.3% in the lowest quartile compared with 27.7% in patients without IE (P = .004). Similarly, patients with IE were more likely to be covered by Medicaid (73.9% vs 41.9%; P < .001). They also had higher rates of medical comorbidities including chronic kidney disease and chronic hypertension, as well as cardiac comorbidities such as a history of valve disease (38.3% vs 0.2%), congenital heart disease (7.3% vs 0.1%), and congestive heart failure (9.0% vs <0.1%; P < .001 for all comparisons; Table 2). Drug use was more common in patients who delivered with IE (72.9% vs 2.7%; P < .001).

Table 2.

Baseline Patient Characteristics of Delivery Hospitalizations Stratified by Associated Diagnosis of Infective Endocarditis

Characteristic	All Delivery Admissions (N = 5 059 674) (Nweighted = 9 585 312)	No Associated IE (N = 5 059 557) (Nweighted = 9 585 095)	Associated IE (N = 117) (Nweighted = 217)	P Value
	Mean (Standard Deviation) or Count (%)
Demographics
Maternal age, years	28.7 (5.8)	28.7 (5.8)	28.7 (5.0)	.93
Gestational agea	38.4 (2.2)	38.4 (2.2)	33.3 (4.8)	<.001
Multiple gestations	b	174 848 (1.8)		.96
Primary expected payerc				<.001
Private insurance	5 074 870 (53.0)	5 074 836 (53.0)	35 (16.0)
Medicaid	4 007 306 (41.9)	4 007 146 (41.9)	160 (73.9)
Self-pay	133 572 (1.4)	133 563 (1.4)	10 (4.4)
Medicare	b	76 395 (0.8)	b
No charge	b	4 285 (0.0)	b
Other	b	277 070 (2.9)	b
Median household income national quartile for patient zip-coded				.004
0 to 25th percentile	2 631 721 (27.7)	2 631 628 (27.7)	93 (43.3)
25th to 50th percentile	2 509 634 (26.4)	2 509 590 (26.4)	44 (20.6)
50th to 75th percentile	2 399 098 (25.2)	2 399 048 (25.2)	50 (23.3)
75th to 100th percentile	1 974 383 (20.8)	1 974 355 (20.8)	28 (12.8)
Comorbidities
Total number of comorbid conditions	0.7 (0.9)	0.7 (0.9)	2.6 (1.4)	<.001
General medical
Pregestational diabetes	b	111 903 (1.2)	b	.66
Chronic hypertension	219 224 (2.3)	219 209 (2.3)	15 (6.8)	.007
Chronic kidney disease	b	10 914 (0.1)	b	.01
Living with human immunodeficiency virus	b	8916 (0.1)	b	.79
Drug use	262 042 (2.7)	261 884 (2.7)	158 (72.9)	<.001
Specific substance (if known, allowing for duplicates)
Opioid	75 769 (0.8)	75 648 (0.8)	121 (55.9)	<.001
Cannabis	137 708 (1.4)	137 692 (1.4)	16 (7.3)	<.001
Sedative	3666 (0.0)	3657 (0.0)	9 (4.4)	<.001
Cocaine	18 089 (0.2)	18 065 (0.2)	24 (11.2)	<.001
Stimulant	35 281 (0.4)	35 247 (0.4)	34 (15.8)	<.001
Other	20 275 (0.2)	20 249 (0.2)	26 (12.0)	<.001
Alcohol use	b	14 274 (0.1)	b	.04
Tobacco use disorder	548 236 (5.7)	548 140 (5.7)	96 (44.2)	<.001
Cardiac
Congenital heart disease	9879 (0.1)	9863 (0.1)	16 (7.3)	<.001
Cardiac valve disease	15 927 (0.2)	15 844 (0.2)	83 (38.3)	<.001
Valve replacement	1160 (0.0)	1150 (0.0)	11 (4.9)	<.001
Other cardiac prosthesis	b	4322 (0.0)	b	<.001
Congestive heart failure	1231 (0.0)	1212 (0.0)	20 (9.0)	<.001
Ischemic heart disease	3275 (0.0)	3265 (0.0)	10 (4.6)	<.001
Obstetric
Gestational hypertension	521 786 (5.4)	521 765 (5.4)	21 (9.8)	.04
Gestational diabetes	b	709 654 (7.4)	b	.04
Obesity in pregnancy	1 031 403 (10.8)	1 031 382 (10.8)	22 (9.9)	.80
Preeclampsia without severe features	324 476 (3.4)	324 466 (3.4)	10 (4.7)	.41
Preeclampsia with severe features or eclampsia	229 060 (2.4)	229 050 (2.4)	10 (4.7)	.09
Prior cesarean section	1 679 206 (17.5)	1 679 179 (17.5)	27 (12.4)	.20

Abbreviation: IE, infective endocarditis.

^a Missing data: 66 617 observations.

^bN <10, value cannot be displayed due to data use restrictions.

^c Missing data: 5832 observations.

^d Missing data: 39 122 observations.

Outcomes

The outcomes of delivery admissions with and without concurrent IE are displayed in Figure 3. When delivery was complicated by IE, patients experienced worse maternal outcomes including higher rates of in-hospital mortality (17.2% vs <0.1%; aRR, 323.32; 95% CI: 127.74–818.37), severe maternal morbidity (77.5% vs 1.6%; aRR, 47.06; 95% CI: 38.10–58.14), and cesarean delivery (56.1% vs 32.2%; aRR, 2.04; 95% CI: 1.65–2.52). Similarly, concurrent IE was associated with worse fetal outcomes such as stillbirths (aRR, 2.96; 95% CI: 1.21–7.24) and preterm deliveries (55.7% vs 10.1%; aRR, 3.61; 95% CI: 2.56–5.08). Healthcare resource use was also higher for patients with IE with longer hospitalizations (mean [standard deviation], 30.0 [24.6] vs 2.7 [2.4] days; aRR, 7.86; 95% CI: 6.57–9.41) and greater inpatient costs (mean [standard deviation], $72 400 [$67 600] vs $5400 [$5400]; aRR, 10.98; 95% CI: 9.05–13.33). Results were similar when analysis was performed using propensity score matching rather than covariate adjustment (Supplementary Table 6).

Figure 3.

Unadjusted and adjusted outcomes among patients admitted for delivery stratified by associated diagnosis of infective endocarditis. * Including readmissions within 60 days of hospital discharge from the patient's first endocarditis-related hospitalization. All outcomes adjusted for age, primary payor, zip-code income quartile, hospital bed size, urban/rural, teaching status, alcohol use, asthma, ischemic heart disease, chronic kidney disease, congenital heart di sease, cystic fibrosis, diabetes mellitus (pregestational vs gestational), drug use and specific drug class used, human immunodeficiency virus/AIDS, chronic hypertension, gestational hypertension and preecclampsia, multiple gestations, obesity, adherent placenta and placenta previa, prior cesarean delivery, pulmonary hypertension, sickle cell disease, systemic lupus erythematosus, tobacco use, history of mechanical valve, and history of other heart procedure.

DISCUSSION

IE is a rare infection that complicates pregnancy, and its incidence and the associated outcomes are largely unknown. The estimated incidence of IE during pregnancy is 0.006% [12, 13], making its investigation difficult. We used a large cohort to explore the reciprocal effects of IE and pregnancy. In our study, 3% of all IE cases among women of reproductive age were maternity-associated, and IE was present in 0.00 002% of delivery hospitalizations. We found that maternity-associated IE does not appear to carry any additional risk for adverse outcomes compared with IE alone. However, IE complicating an admission for delivery is strongly associated with adverse maternal and fetal outcomes.

In-hospital mortality did not differ significantly between patients with maternity-associated and nonmaternity-associated IE, and our reported in-hospital mortality for both groups is lower than reported in the general IE population (ranges between 15% and 20%) [3]. This is likely due to exclusion of patients aged >55 years, as increasing age has been associated with mortality in IE [14]. Additionally, it is possible that mortality from maternity-associated IE is decreasing due to medical advances in diagnostics and intervention, as early reports estimated mortality to be 33% [15], whereas newer reviews report maternal mortality to be closer to 11% [6].

When examining the effects of IE on maternal and fetal delivery outcomes, patients with IE complicating delivery experienced dramatically worse maternal and fetal outcomes. IE was associated with higher rates of severe maternal morbidity and in-hospital mortality. Furthermore, in a subgroup analysis of maternity-associated IE, patients diagnosed with IE during an admission in which delivery occurred experienced the highest rates of mortality. This may be due in part to an emergent need for delivery in pregnant patients who are critically ill. In fact, approximately half of the patients with IE at delivery were delivered by cesarean section, which suggests that rapid delivery of the fetus may have been indicated. More than half of deliveries complicated by IE were preterm with an average gestational age at delivery of 32.4 ± 4.8 weeks. This is comparable to a review that found a mean gestational age of 34.2 ± 5.1 weeks [6]. We hypothesize that this represents a mixture of spontaneous preterm birth due to systemic inflammation from the infection and medically indicated deliveries for maternal or fetal indications. The stillbirth rate was also higher in patients who delivered with IE. These findings underscore that IE can have devastating effects on both mother and fetus when compared with outcomes of a typical maternity patient.

Patients with maternity-associated IE and IE complicating delivery reported higher rates of drug use. It is important to note that in the ICD-10 coding system, route of administration of drugs used is not differentiated. However, IVDU is a well-known risk factor for IE in the general population [1, 3] and specifically in pregnant patients [15]. Therefore, in line with previous research, we hypothesize that IVDU contributes to the higher rates of drug use in maternity-associated IE. We also found that patients with maternity-associated IE and IE complicating delivery were more likely to be covered by Medicaid. Similarly, patients who delivered with IE were more likely to come from the lowest income quartile than those who delivered without IE. These findings highlight the potential interplay of behavioral risk factors and social determinants of health. Indeed, previous research has begun to elucidate this relationship. For example, studies have found that the incidence of IVDU-related IE is associated with household income and a reliance on Medicaid [16, 17]. Finally, we found that among patients with IE, pregnancy was associated with younger age and lower rates of both general and cardiac comorbidities. Conversely, patients who had concurrent IE during their delivery admission were generally less healthy and were more likely to have a history of cardiac disease and prostheses. Taken together, these findings suggest that the risk of experiencing IE during pregnancy reflects a complex interplay among biologic, socioeconomic, and behavioral variables, while supporting previous claims that IE during pregnancy is almost always associated with IVDU or preexisting cardiac disease [6, 15].

Our study has limitations. The NRD relies on use of ICD-10 codes. In addition, although the NRD tracks patients over time, the data structure does not allow tracking across calendar year or state lines. The NRD only captures inpatient encounters, while much pregnancy care occurs in the outpatient setting. Finally, our study was limited to the data collected and made available through the NRD. As the NRD reports only a fixed selection of data points, we were unable to fully characterize patient demographics (ie, race/ethnicity), risk/prognostic factors, clinical characteristics of IE episodes, and certain aspects of treatment (ie, cardiac surgery, antibiotic therapy). Furthermore, this limits our ability to validate our ICD-10–based definition of IE against standardized clinical definitions such as the modified Duke criteria. Despite these limitations, we used a large sample size and provide what we believe to be the largest analysis of IE during pregnancy to date. Furthermore, our data comes from a nationally representative sample of hospitalizations, improving generalizability of our results.

CONCLUSIONS

Maternity-associated IE is rare and does not appear to convey additional risk of adverse outcomes when compared with nonmaternity-associated IE. However, IE complicating delivery is associated with higher rates of severe maternal morbidity, preterm births, and both maternal and fetal mortality compared with uncomplicated deliveries. Further research is needed to fully describe this disease process and generate reliable data to guide the management and prevention in this unique population.

Supplementary Data

Supplementary materials are available at Clinical Infectious Diseases online. Consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or comments should be addressed to the corresponding author.

Notes

Acknowledgments. The authors appreciate the HCUP Data Partners who contribute data to the National Readmissions Database. A complete list of partners can be found at (www.hcup-us.ahrq.gov/hcupdatapartners.jsp).

Financial support. Work contained here was made possible by the following grants from the National Institute of Allergy and Infectious Diseases: K24-AI093969, V. G. F.; R01-AIO68804, V. G. F.; T32-AI100851, E. M. E.; and the National Center for Advancing Translational Sciences TL1-TR002555, J. J. F.; all to institution. Data acquisition was also supported by funding from the Foundation for Women and Girls with Blood Disorders to J. J. F.’s institution.

Potential conflicts of interest. V. G. F. reports personal fees for consultancy from Novartis, Novadigm, Durata, Debiopharm, Genentech, Achaogen, Affinium, Medicines Co, Cerexa, Tetraphase, Trius, MedImmune, Bayer, Theravance, Basilea, Affinergy, Janssen, xBiotech, Contrafect, Regeneron, Basilea, Destiny, Amphliphi Biosciences, Integrated Biotherapeutics, and C3J; grants to their institution from NIH, MedImmune, Cerexa/Forest/Actavis/Allergan, Pfizer, Advanced Liquid Logics, Theravance, Novartis, Cubist/Merck, Medical Biosurfaces, Locus; Affinergy, Contrafect; Karius, Genentech, Regeneron, Basilea, and Janssen; personal/educational fees from Green Cross, Cubist, Cerexa, Durata, Theravance, Debiopharm; royalties from UpToDate; and a patent for sepsis diagnostics pending (host gene expression signature diagnostic for sepsis). All remaining authors: No reported conflicts of interest. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.