Abstract
Introduction:
Approximately one-quarter of intensive care unit admissions include a severe maternal morbidity event. We hypothesized that non-severe maternal morbidity intensive care unit admissions would have an increased need for intensive care unit-specific services compared to admissions with severe maternal morbidity.
Methods:
We conducted a retrospective cohort study using the Premier Database to examine delivery hospitalizations that included intensive care unit admissions (10/2015–12/2020). We used ICD-10 codes to identify severe maternal morbidity and charge codes to identify services usually provided in intensive care units. We performed descriptive analyses for intensive care unit service codes and multivariable logistic regression to assess factors associated with admissions in patients with no codes for severe maternal morbidity.
Results:
Among 4.8 million delivery hospitalizations, 35,837 patients were admitted to the intensive care unit. Of intensive care unit admissions, 9,803 (27.4%) met severe maternal morbidity criteria. Intensive care unit-specific charge codes were more commonly charged to patients with severe maternal morbidity compared to those without severe maternal morbidity: invasive blood pressure monitoring (21.0% vs 3.8%) and invasive hemodynamic monitoring (2.0% vs 0.2%). Codes for intravenous fluid amounts suggest that severe maternal morbidity patients received higher volumes of IVF. Median intensive care unit length of stay was similar between groups. Comorbidities such as cardiomyopathy and preeclampsia were associated with increased non-severe maternal morbidity intensive care unit admission.
Conclusion:
Intensive care unit admission may be driven by concern for clinical deterioration among patients with high-risk comorbidities rather than the presence of severe maternal morbidity.
Introduction:
Admission to the intensive care unit (ICU) during a maternal delivery hospitalization is rare, occurring in less than 1% of deliveries.1–3 ICU admission is often used in research as an indicator of clinical deterioration and need for intervention. However, recent studies demonstrate that only about one-quarter of delivery-hospitalization ICU admissions include Centers for Disease Control and Prevention (CDC) criteria for a severe maternal morbidity (SMM) event.4,5 Similarly, data from Godeberge et al. show that 15.5% of obstetric patients admitted to the ICU had no documented severe acute maternal morbidity.6 These findings suggest that ICU admissions may not always be driven by an SMM event. ICU admission instead may be chosen in some institutions for obstetric patients with comorbid conditions who require ICU-specific interventions or invasive monitoring. To address this gap, we aim to characterize the reasons for ICU admission in delivery hospitalizations requiring ICU care but not involving an SMM event. We hypothesized that non-SMM ICU admissions would have an increased need for ICU-specific services compared to admissions with SMM and that patient medical and obstetric comorbidities would be associated with increased non-SMM ICU admission.
Methods:
We conducted a retrospective cohort study using the Premier Database to examine delivery hospitalizations that included ICU admissions from October 1, 2015, to December 31, 2020. This study was determined exempt by Duke University Health System IRB. The Premier database is an all-payer, nationally representative hospital administrative database encompassing more than 700 hospitals. The dataset includes comprehensive administrative, clinical, and billing information, such as diagnosis and procedure codes, demographics, and facility characteristics. We used ICD-10 codes to identify SMM according to CDC criteria. We identified charge codes for ICU-specific services and compared rates of those charge codes between patients with and without SMM.
We performed descriptive analyses of ICU admissions stratified by SMM status to examine differences in ICU-level resource utilization. Multivariable logistic regression models identified factors associated with non-SMM ICU admissions. Covariates examined were patient and hospital characteristics, payor status, location of care, and comorbidities.
Our primary outcome was ICU-specific services among patients admitted to ICU with and without SMM. Our secondary outcomes were patient and hospital factors associated with non-SMM ICU admission.
Results:
We analyzed 4,832,179 delivery hospitalizations in the Premier database from October 1, 2015, to December 31, 2020. Of these, 35,837 patients were admitted to the ICU, and 9,803 (27.4%) met SMM criteria (Figure 1).5 ICU-specific service codes were more commonly charged to patients with SMM compared to those without SMM. These included invasive blood pressure monitoring (21.0% vs. 3.8%) and invasive hemodynamic monitoring (2.0% vs. 0.2%). Intravenous fluids (IVF) charge codes suggest larger-volume resuscitation in SMM patients, with a median of 7.1L compared to 3.5L in non-SMM patients. Median ICU length of stay was comparable between groups (Table 1).
Figure 1.
Cohort derivation
Table 1.
ICU-level resource utilization in patients with and without SMM
| SMM (including transfusion) (n=9803) | No SMM (n=26034) | |
|---|---|---|
| Invasive BP monitoring | 2059 (21.0%) | 980 (3.8%) |
| Telemetry | 234 (2.4%) | 226 (0.9%) |
| Invasive hemodynamic monitoring | 196 (2.0%) | 40 (0.2%) |
| ICU nursing | 62 (0.6%) | 245 (0.9%) |
| Blood product administration | 5555 (56.7%) | 0 (0.0%) |
| IV fluid administration | 9109 (92.9%) | 23354 (89.7%) |
| Volume (L) | 7.1 [2.7, 13.1] | 3.5 [1.1, 6.3] |
| >= 3L | 7161 (73.0%) | 14215 (54.6%) |
| >= 5L | 6059 (61.8%) | 8999 (34.6%) |
| >= 10L | 3489 (35.6%) | 2810 (10.8%) |
| Diuresis | 836 (8.5%) | 600 (2.3%) |
| 2 [1, 3] | 2 [1, 2] |
ICU: Intensive Care Unit; SMM: Severe Maternal Morbidity; RBC: Red Blood Cell; BP: Blood Pressure
Data are N (%), mean ± standard deviation, or median [25th percentile, 75th percentile]
In the multivariable analysis (Table 2), the medical and obstetric comorbidities with strong association with non-SMM ICU admission were cardiomyopathy (OR 12.77; 95% CI: 11.07, 14.72), preeclampsia with severe features (8.05; 7.72, 8.40), and pulmonary hypertension (7.67; 5.65, 10.41). Additionally, hospital-level characteristics with association with non-SMM ICU admission were identified. Smaller hospitals (<400 beds) were associated with decreased non-SMM ICU admission compared to larger hospitals.
Table 2.
Model for factors associated with ICU admission among patients without SMM
| Odds Ratio (95% CI) | p-value | |
|---|---|---|
| Mode of delivery | ||
| Vaginal delivery | 1.0 (ref) | N/A |
| Cesarean delivery | 1.28 (1.22, 1.33) | <0.001 |
| Intrapartum cesarean delivery | 1.59 (1.53, 1.65) | <0.001 |
| Age (per year) | 0.96 (0.96, 0.97) | <0.001 |
| Primary Payor | ||
| Managed care | 1.02 (0.99, 1.06) | 0.258 |
| Medicaid | 1.0 (ref) | N/A |
| Medicare | 1.03 (0.90, 1.18) | 0.693 |
| Other | 0.84 (0.81, 0.88) | <0.001 |
| Race | ||
| Asian | 1.06 (0.98, 1.14) | 0.139 |
| Black | 1.04 (1.00, 1.08) | 0.078 |
| Other | 1.00 (0.95, 1.06) | 0.966 |
| Unknown | 1.09 (1.02, 1.16) | 0.008 |
| White | 1.0 (ref) | N/A |
| Ethnicity | ||
| Hispanic | 0.94 (0.90, 0.99) | 0.017 |
| Non-Hispanic | 1.0 (ref) | N/A |
| Unknown | 1.35 (1.26, 1.45) | <0.001 |
| Urban hospital | 1.08 (0.81, 1.46) | 0.594 |
| Teaching hospital | 0.77 (0.59, 1.01) | 0.059 |
| Bed size | ||
| 000–099 | 0.42 (0.27, 0.66) | <0.001 |
| 100–199 | 0.54 (0.37, 0.79) | 0.002 |
| 200–299 | 0.59 (0.41, 0.86) | 0.006 |
| 300–399 | 0.60 (0.41, 0.87) | 0.007 |
| 400–499 | 0.74 (0.49, 1.11) | 0.147 |
| 500+ | 1.0 (ref) | N/A |
| Region of country | ||
| Midwest | 0.69 (0.52, 0.91) | 0.008 |
| Northeast | 0.80 (0.57, 1.13) | 0.209 |
| South | 1.0 (ref) | N/A |
| West | 0.79 (0.59, 1.06) | 0.119 |
| Condition | ||
| Cardiomyopathy | 12.77 (11.07, 14.72) | <0.001 |
| Preeclampsia with severe features | 8.05 (7.72, 8.40) | <0.001 |
| Pulmonary hypertension | 7.67 (5.65, 10.41) | <0.001 |
| Arrhythmias | 4.05 (3.57, 4.60) | <0.001 |
| Placenta accreta spectrum | 6.43 (5.28, 7.83) | <0.001 |
| Preterm birth | 3.90 (3.78, 4.03) | <0.001 |
| Congenital heart disease | 4.13 (3.45, 4.94) | <0.001 |
| Valvular heart disease | 3.53 (3.05, 4.09) | <0.001 |
| Coronary/ischemic/atherosclerotic | 2.12 (1.61, 2.77) | <0.001 |
| Chronic renal disease | 2.35 (2.07, 2.67) | <0.001 |
| Aortopathies | 2.76 (1.86, 4.11) | <0.001 |
| Other cardiac disease | 2.41 (1.23, 4.75) | 0.011 |
| Preeclampsia | 2.35 (2.27, 2.44) | <0.001 |
| Chronic hypertension | 2.11 (2.01, 2.21) | <0.001 |
| Neuromuscular disease | 1.63 (1.52, 1.76) | <0.001 |
| HIV/AIDS | 1.68 (1.26, 2.24) | 0.001 |
| Bleeding disorder, preexisting | 1.48 (1.38, 1.59) | <0.001 |
| Preexisting diabetes mellitus | 1.85 (1.74, 1.97) | <0.001 |
| Advanced maternal age | 1.49 (1.42, 1.57) | <0.001 |
| Asthma, acute or moderate/severe | 1.20 (1.15, 1.26) | <0.001 |
| Placental abruption | 1.63 (1.50, 1.77) | <0.001 |
| Substance use disorder | 1.31 (1.25, 1.37) | <0.001 |
| BMI greater or equal to 40 | 1.30 (1.23, 1.37) | <0.001 |
| Thyrotoxicosis | 1.49 (1.24, 1.80) | <0.001 |
| Gestational diabetes mellitus | 1.31 (1.25, 1.37) | <0.001 |
| Gastrointestinal disease | 1.11 (1.05, 1.16) | <0.001 |
| Major mental health disorder | 1.17 (1.12, 1.23) | <0.001 |
| Bariatric surgery | 1.28 (1.07, 1.54) | 0.008 |
| Preexisting anemia | 1.14 (1.09, 1.18) | <0.001 |
| Multiples gestation | 1.21 (1.13, 1.29) | <0.001 |
| Connective tissue or autoimmune disease | 1.10 (0.90, 1.35) | 0.343 |
| Prior cesarean birth | 0.92 (0.89, 0.96) | <0.001 |
| Placenta previa, complete or partial | 1.12 (0.97, 1.31) | 0.133 |
Discussion:
We examined non-SMM ICU admissions and found that these patients generally received fewer invasive monitoring procedures and ICU-specific interventions than patients with SMM. However, we did find significant associations between high-risk comorbidities and non-SMM ICU admission. This was particularly significant for patients with conditions like cardiomyopathy, preeclampsia with severe features, and pulmonary hypertension. Taken together, these findings suggest obstetric ICU admission may not always reflect clinical severity as defined by the CDC and instead reflect concern for potential clinical deterioration and need for close observation.
Additionally, we found that smaller hospitals are less likely to have non-SMM ICU admissions. This is consistent with prior work showing that smaller hospitals are less likely to have ICU admissions generally.5 This suggests that more complex patients may be preemptively transferred to larger hospitals or that smaller hospitals reserve their ICU beds for patients with established SMM.
ICU admission itself has important consequences.7 Transfer to the ICU may limit access to obstetric-specific care (e.g., fundal assessments, lactation support).8 Qualitative studies also describe ICU admission as a traumatic experience for mothers. Separation from their newborns during hospitalization, which may occur when a birthing person is admitted to the ICU instead of the postpartum unit, has been described as ‘terrible’ and ‘painful’.9 In Hinton et al., participants reported confusion about the reason for ICU admission and distress related to their physical condition.10
These qualitative studies reinforce the importance of thoughtful decision-making regarding ICU admission given the potential trauma for the birthing person, their family, and members of the healthcare team who may also be affected by these events. Internal audits may be necessary to evaluate the appropriateness of ICU admissions. This is especially relevant when considering the similar length of ICU stay between patients with and without SMM. If patients without SMM are receiving less ICU-specific services, their length of stay should in theory be shorter. However, providers may be hesitant to discharge patients from ICU care if they do not have a concrete way to determine clinical improvement, since these patients may not always be decompensated on admission. This helps demonstrate why managing these admissions can be challenging for clinicians. The resources necessary to facilitate an ICU admission should also be considered in these audits.
It should be noted that the American College of Obstetricians and Gynecologists (ACOG) has released its own list of SMM which includes both (1) Preexisting cardiac disease (congenital or acquired) with intensive care unit admission for treatment and (2) Peripartum cardiomyopathy.11 Including these pathologies as part of the CDC definition of SMM would encapsulate many of the non-SMM ICU admissions presented in our dataset. These findings suggest that at present, the CDC’s SMM definition may incompletely capture cardiac-related severe morbidity. However, given the range of presentations a pregnant patient with cardiac comorbidities could present with, future work should examine whether ICU admission for these conditions reliably indicates the degree of acute morbidity that the SMM classification is intended to reflect. For example, Majmundar et al. reported an 11.3% prevalence of cardiovascular disease in pregnancy using data from the Nationwide Readmission Database which included over 39 million deliveries.12 While they included lower acuity morbidities including gestational hypertension and preeclampsia in this estimation, their stratification of congenital heart disease is relevant. Simple congenital cardiac disease had a ten-fold higher prevalence when compared to complex congenital heart disease (6.0 vs. 0.6 per 10,000, respectively). These findings highlight the possible implications of incorporating broad categories of morbidity into CDC definitions. Simply expanding SMM definitions could reduce the specificity of SMM as a marker of acute illness.
The strength of our study lies in the use of Premier, a large and nationally representative database with strong external validity. However, this study has several limitations. Although we thoughtfully considered and included charge codes that we believed capture relevant ICU services, charge codes alone are unable to fully encapsulate the indication nor the severity of the clinical course of patients admitted to the ICU. Additionally, the use of an administrative database has limitations. Billing and ICD-10 procedure codes are subject to human error in that if something is not coded, it is not recorded as part of the delivery hospitalization. Future studies and obstetric service lines may consider using an electronic health record-based system to intentionally track, accurately document and describe the etiology of maternal ICU admissions.
Highlights:
Approximately 1 in 4 intensive care unit admissions during delivery hospitalization meet severe maternal morbidity criteria
Intensive care unit-specific services were more common in patients with severe maternal morbidity than in those without
Cardiomyopathy and preeclampsia were strongly associated with non-severe maternal morbidity admissions
Funding:
Dr. Federspiel is supported by the Eunice Kennedy Shriver National Institute of Child Health and Human Development under Award Number K12-HD103083. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.
Disclosures:
Dr. Federspiel discloses consulting for Noma AI, Grifols, UCB, and Sanofi, and research funding from Koko Medical. The other authors report no disclosures.
References:
- 1.Aoyama K, Pinto R, Ray JG, et al. Variability in intensive care unit admission among pregnant and postpartum women in Canada: a nationwide population-based observational study. Crit Care. 2019;23(1):381. doi: 10.1186/s13054-019-2660-x [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Wanderer JP, Leffert LR, Mhyre JM, Kuklina EV, Callaghan WM, Bateman BT. Epidemiology of obstetric-related ICU admissions in Maryland: 1999–2008*. Crit Care Med. 2013;41(8):1844–1852. doi: 10.1097/CCM.0b013e31828a3e24 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Maiden MJ, Finnis ME, Duke GJ, et al. Obstetric admissions to intensive care units in Australia and New Zealand: a registry-based cohort study. BJOG Int J Obstet Gynaecol. 2020;127(12):1558–1567. doi: 10.1111/1471-0528.16285 [DOI] [PubMed] [Google Scholar]
- 4.CDC. Identifying Severe Maternal Morbidity (SMM). Maternal Infant Health. May 20, 2024. Accessed May 13, 2025. https://www.cdc.gov/maternal-infant-health/php/severe-maternal-morbidity/icd.html [Google Scholar]
- 5.Himmelwright R, Fuller M, Krishnamoorthy V, Bauer M, Habib AS, Meng ML. Maternal intensive care unit utilization by mode of delivery in the United States: a retrospective cohort study. Presented at: Society for Obstetric Anesthesia and Perinatology (SOAP) Annual Meeting; 2026. Accessed March 30, 2026. https://sync.meetingfiles.com/PF/01250ba2b6c2fc96e7dec16f3365f7a5/01250ba2b6c2fc96e7dec16f3365f7a5.pdf [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Godeberge C, Deneux-Tharaux C, Seco A, et al. Maternal Intensive Care Unit Admission as an Indicator of Severe Acute Maternal Morbidity: A Population-Based Study. Anesth Analg. 2022;134(3):581–591. doi: 10.1213/ANE.0000000000005578 [DOI] [PubMed] [Google Scholar]
- 7.Herridge MS, Azoulay É. Outcomes after Critical Illness. Hardin CC, ed. N Engl J Med. 2023;388(10):913–924. doi: 10.1056/NEJMra2104669 [DOI] [PubMed] [Google Scholar]
- 8.Sareen J, Olafson K, Kredentser MS, et al. The 5-Year Incidence of Mental Disorders in a Population-Based ICU Survivor Cohort. Crit Care Med. 2020;48(8):e675–e683. doi: 10.1097/CCM.0000000000004413 [DOI] [PubMed] [Google Scholar]
- 9.Biskop E, Thernström Blomqvist Y, Diderholm B, Skalkidou A, Grandahl M. Separation between mothers and newborns directly after birth: mothers’ experiences of separation - a qualitative analysis from a cohort study. Arch Dis Child Fetal Neonatal Ed. Published online November 12, 2025:fetalneonatal-2025-329190. doi: 10.1136/archdischild-2025-329190 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Hinton L, Locock L, Knight M. Maternal critical care: what can we learn from patient experience? A qualitative study. BMJ Open. 2015;5(4):e006676. doi: 10.1136/bmjopen-2014-006676 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.American College of Obstetricians and Gynecologists and the Society for Maternal–Fetal Medicine, Kilpatrick SK, Ecker JL. Severe maternal morbidity: screening and review. Am J Obstet Gynecol. 2016;215(3):B17–22. doi: 10.1016/j.ajog.2016.07.050 [DOI] [PubMed] [Google Scholar]
- 12.Majmundar M, Doshi R, Patel KN, Zala H, Kumar A, Kalra A. Prevalence, trends, and outcomes of cardiovascular diseases in pregnant patients in the USA: 2010–19. Eur Heart J. 2023;44(9):726–737. doi: 10.1093/eurheartj/ehac669 [DOI] [PubMed] [Google Scholar]