Antenatal Pulmonary Embolism Diagnostics in Pregnant Patients with SARS-CoV-2 Infection in Community Hospitals

Abstract

Background

The diagnostic evaluation of pulmonary embolism (PE) in pregnancy is challenging, as the physiologic changes of pregnancy can mimic several PE symptoms. Acute respiratory infections introduce additional diagnostic complexity by producing systemic inflammation, altering vital signs, and, in some cases, elevating D-dimer levels. As a case in point, we examined how SARS-CoV-2 infection affected presentation, pretest probability, and diagnostic testing in pregnant patients with possible PE.

Methods

We performed a retrospective cohort study across 21 community medical centers from October 1, 2021, to March 30, 2023. We included pregnant outpatients ≥ 18 years evaluated for suspected PE with D-dimer testing, compression ultrasonography, computed tomography pulmonary angiography (CTPA), or lung scintigraphy. We compared patients with and without COVID-19 using bivariate analysis.

Results

Among 860 patients, the median age was 30.0 years; 39.1% were in the third trimester. COVID-19 was present in 147 (17.1%). Compared with non-COVID-19 patients, those with COVID-19 more often had fever (36.1% vs. 4.2%), tachycardia ≥ 110 bpm (66.0% vs. 34.2%), and oxygen saturation < 95% (12.2% vs. 4.8%), but less often reported chest pain (49.7% vs. 65.5%; all p  < 0.001). Nearly all patients had low-to-intermediate pretest probability, but intermediate classification was more common with COVID-19 patients (63.3% vs. 39.0%; p  < 0.001). COVID-19 patients more often had elevated D-dimer > 1.0 mg/L (49.1% vs. 36.4%; p  < 0.001) and more commonly underwent chest radiography (61.9% vs. 50.1%; p  = 0.009). Among patients who underwent advanced imaging ( n  = 393), CTPA predominated in both cohorts. Overall, PE was rare ( n  = 6; 0.7%), and mortality was low ( n  = 3; 0.3%).

Conclusion

COVID-19 in pregnancy was associated with worse vital signs, higher pretest probability, higher D-dimer values, and increased diagnostic testing. These findings illustrate how acute respiratory infections may recalibrate PE risk assessment in pregnancy and highlight the need to refine diagnostic strategies when infection-related physiologic changes are present.

Introduction

Pulmonary embolism (PE) has an estimated incidence of approximately 1 in 3,000 pregnancies and is a leading cause of maternal death in developed countries. ¹ Timely and accurate diagnosis is essential, as delays in treatment can result in devastating consequences for both patient and fetus. However, diagnosing PE during pregnancy presents unique challenges. Physiological changes during pregnancy can make the diagnosis of PE more complicated, as symptoms such as shortness of breath and leg swelling, along with elevated D-dimer levels, are common during pregnancy. ^{2 3}

Acute respiratory infections introduce an additional layer of diagnostic complexity in the evaluation of suspected PE. Infections such as influenza, respiratory syncytial virus (RSV), bacterial pneumonias, and SARS-CoV-2 can produce systemic inflammation, tachycardia, hypoxia, and, in some cases, elevated D-dimer levels, clinical findings that overlap with both pregnancy and PE. ^{4 5 6}

The emergence of COVID-19 brought the intersection of pregnancy, respiratory infection, and possible PE to the forefront. SARS-CoV-2, the virus responsible for COVID-19, is primarily a respiratory pathogen but also exerts systemic effects, particularly on the vascular endothelium. ^{7 8 9} COVID-19 is now recognized as a prothrombotic condition known to increase the risk of thrombosis through endothelial damage, cytokine-driven inflammation, and activation of the coagulation cascade. ⁷ Among those requiring inpatient care, immobility, hypoxia, and severe systemic illness further amplify this risk. ^{10 11} Most venous thromboembolism (VTE) events in COVID-19–positive patients occur in those with more severe illness, both during and following hospitalization. ^{12 13} Despite the hypercoagulable state of pregnancy, gravid outpatients with COVID-19 may not have an appreciably higher risk of PE than their noninfected outpatient counterparts. ¹⁴

Guidelines have been published to assist in the diagnosis of PE in patients with COVID-19. ^{15 16} How the diagnostic approach might vary in pregnant patients has not been well studied. To better understand how acute respiratory infection may influence PE evaluation in pregnancy, we conducted a multi-site retrospective cohort study across U.S. community medical centers comparing pregnant outpatients with and without confirmed SARS-CoV-2 infection. Examining how one respiratory infection altered presentation, structured pretest probability classification, and imaging patterns may offer broader insight into the need for PE diagnostic recalibration in the setting of acute respiratory infection in pregnancy.

Methods

Study Design and Setting

We conducted a retrospective cohort study, Antenatal PE Diagnostics (APED), across 21 Kaiser Permanente Northern California (KPNC) community medical centers and affiliated outpatient obstetric and primary care clinics. ^{17 18 19 20 21 22} KPNC serves over 4.5 million members with demographic and socioeconomic characteristics comparable to those of the broader local and state populations. ²³ The study was approved by the KPNC Institutional Review Board, which waived the requirement for informed consent.

Diagnostic evaluations for suspected acute PE in ambulatory pregnant patients were conducted at the discretion of treating clinicians. Hospital-based clinicians in the emergency department (ED) and labor and delivery unit (LDU) had 24-hour access to routine laboratory tests, including D-dimer, COVID-19 polymerase chain reaction (PCR) test, chest radiography (CXR), computed tomography pulmonary angiography (CTPA), and specialty consultation. Formal compression ultrasonography (CUS) was available from hospital-based radiology departments daily between 07:00 and 21:00 hours, while lung scintigraphy (ventilation/perfusion [V/Q] scanning) was available from nuclear medicine between 08:00 and 17:00 hours. Off-hours CUS and lung scintigraphy availability varied by facility and typically required on-call specialty approval. During pregnancy, the nuclear medicine department first performed low-dose Q scanning, adding ventilation scans only as indicated. ^{24 25} Ventilation scans were not routinely employed in patients with COVID-19. ^{26 27 28}

Clinicians based in outpatient clinics also had the ability to order any PE diagnostic test, including pulmonary vascular imaging. However, outpatient access to on-site diagnostic testing was more limited and varied by location.

Population

The larger APED study population consisted of pregnant health plan members ≥ 18 years of age who underwent one of four diagnostic tests: D-dimer, CUS, CTPA, or V/Q scan for suspected PE from October 1, 2021, to March 30, 2023, as previously described. ^{17 18 19 20} PE-suggestive symptoms included new or worsening dyspnea, chest or thoracic pain, hemoptysis, palpitations, syncope, and presyncope. ²⁹ We identified patients who were pregnant using the KPNC Division of Research Perinatal Research Unit's Perinatal Obstetric Database. We manually reviewed the electronic health records of PE diagnostic test recipients to determine study eligibility. We excluded those not known to be pregnant at the time of diagnostics, those known to have recent or impending pregnancy loss, those who had known VTE and were currently on anticoagulation, or those who left before completing the agreed-upon diagnostic evaluation. We also excluded patients whose diagnostic evaluation began during an inpatient stay, as the study focused on outpatients seeking care for PE-related symptoms. Pregnant patients with COVID-19 undergoing PE diagnostics were excluded from prior APED studies, as the presence of infection might have altered the diagnostic process. ^{17 18 19 20 30} This analysis focuses on those who had confirmed COVID-19. Patients were classified as COVID-19–positive if they had (1) a positive SARS-CoV-2 PCR test obtained during the index ED or LDU encounter or (2) documentation of a positive test from a home or healthcare setting within 5 days prior to presentation. Across the study period, California experienced a transition in circulating SARS-CoV-2 variants, moving from the Delta variant in late 2021 to sustained predominance of the Omicron variant for the remainder of the study. ³¹ For this study, we undertook a secondary review of the cohort to confirm eligibility and proper categorization (COVID-19 vs. not) and recategorized patients as indicated.

Data Collection, Definitions, and Outcome

As previously described, abstractors, all serving as APED co-investigators, conducted manual reviews of electronic health records after completing standardized training in data collection methods using a computerized tool. ^{17 18 19 20 32} Any questions that arose during abstraction were addressed with the principal investigator. The collected patient characteristics included demographics, comorbidities, symptoms related to PE and deep vein thrombosis (DVT), vital signs, COVID-19 status, and laboratory and imaging results. Race and ethnicity data, which were self-reported, were collected to reflect population diversity. Low socioeconomic status was defined as residence in a census block group where at least 25% of adults had not completed high school, or at least 20% of households had an income below the federal poverty level. ³³ For vital signs, we recorded the most concerning values observed during the diagnostic encounter, but prior to orders of pulmonary vascular imaging. We reported D-dimer values using the tripartite categories of the pregnancy-adapted YEARS algorithm from the Artemis study: <0.5, 0.5 to 1.0, and >1.0 mg/L, ^{34 35 36} which were used in prior APED studies ¹⁹ and have been recommended by recent society guidelines in Europe and North America. ^{37 38 39} We estimated PE pretest probability using the Pregnancy-Adapted Geneva (PAG) score, reporting both age and vital sign categories from this scoring system. ⁴⁰

Statistical Analysis

We presented categorical data as frequencies and proportions, and continuous variables as medians with interquartile ranges (IQRs). We used bivariate analysis to compare patients with and without COVID-19. p -Values were calculated using Pearson's chi-square test, Wilcoxon rank-sum test, or Fisher's exact test as indicated. A p -Value of < 0.05 was considered statistically significant. Data management was conducted in SAS (version 9.4), and all analyses were conducted in R Version 4.3.1.

Results

We originally identified 720 APED-eligible cases without COVID-19 and 144 cases with presumed COVID-19. During a second screening to confirm accurate categorization for this current analysis, we recategorized 13 cases, 4 of which were excluded from the study ( Fig. 1 ). The remaining 860 constitute the study cohort: 713 without and 147 with COVID-19. Among the total cohort of 860 patients, the median age was 30.0 years (IQR: 25.9–34.2), and 336 (39.1%) were in the third trimester. Most ( n  = 741; 86.2%) were evaluated in the ED. COVID-19 was present in 147 (17.1%) patients. Among patients with COVID-19, the median age was 30.0 years, and nearly half were in their third trimester at presentation ( Table 1 ). Most COVID-19 patients ( n  = 127; 86.4%) were evaluated in the ED, while the remainder presented to the LDU. Symptom characteristics differed between COVID-19 and non-COVID-19 patients. While patients with COVID-19 were significantly more likely to present with fever (36.1% vs. 4.2%; p  < 0.001) and cough without blood (70.1% vs. 18.8%; p  < 0.001), they reported chest pain and palpitations less frequently than their non-COVID-19 counterparts ( Table 1 ). Similarly, classic DVT symptoms such as unilateral leg pain or swelling were less common in the COVID-19 cohort (2.0% vs. 7.7%; p  = 0.013). COVID-19 patients also exhibited more pronounced vital sign derangements, with higher rates of tachycardia and low oxygen saturations (<95%; p  < 0.001). While nearly all patients ( n  = 829; 96.4%) had low-to-intermediate pretest probability, the COVID-19 cohort had a significantly higher prevalence of intermediate pretest probability than the non-COVID-19 cohort (63.3% vs. 39.0%; p  < 0.001).

Fig. 1.

Cohort assembly of pregnant patients with COVID-19 undergoing pulmonary embolism diagnostic testing. APED, antenatal pulmonary embolism diagnostics; CT, computed tomography; CUS, compression ultrasonography; PE, pulmonary embolism; V/Q, ventilation/perfusion. ^a Net changes: The non-COVID cohort lost 7 encounters (subtracted 8, added 1) and the COVID cohort gained 3 (added 8, subtracted 4, subtracted 1).

Table 1. Patient and clinical setting characteristics of pregnant patients with suspected pulmonary embolism stratified by COVID-19 status.

	Overall	Active COVID-19
Characteristic	n  = 860	Yes, n  = 147	No, n  = 713	p -Value
Age (y), median (IQR)	30.0 (25.9–34.2)	30.0 (26.5–33.5)	30.0 (25.5–34.5)	>0.9
Age category (y)				0.7
<35	645 (75.0)	112 (76.2)	533 (74.8)
≥35	215 (25.0)	35 (23.8)	180 (25.2)
Race/ethnicity				0.6
Asian	148 (17.2)	29 (19.7)	119 (16.7)
Black or African American	107 (12.4)	15 (10.2)	92 (12.9)
Hispanic or Latino	303 (35.2)	48 (32.7)	255 (35.8)
Non-Hispanic White	259 (30.1)	49 (33.3)	210 (29.5)
Other race categories a	43 (5.0)	6 (4.1)	37 (5.2)
Insurance category				0.11
Commercial	671 (78.0)	122 (83.0)	549 (77.0)
Medicaid	189 (22.0)	25 (17.0)	164 (23.0)
Low socioeconomic status	208 (25.6)	32 (22.4)	176 (26.3)	0.3
Unknown	48	4	44
Prepregnancy body mass index (kg/m 2 )				0.7
<30	411 (56.7)	74 (54.0)	337 (57.3)
30–39	258 (35.6)	53 (38.7)	205 (34.9)
≥40	56 (7.7)	10 (7.3)	46 (7.8)
Unknown	135	10	125
Gravidity				0.4
1	217 (25.3)	41 (28.1)	176 (24.7)
≥2	641 (74.7)	105 (71.9)	536 (75.3)
Unknown	2	1	1
Parity				0.4
None	332 (38.6)	61 (41.5)	271 (38.0)
≥1	528 (61.4)	86 (58.5)	442 (62.0)
History venous thromboembolism	35 (4.1)	2 (1.4)	33 (4.6)	0.068
Current gestational diabetes	32 (3.7)	9 (6.1)	23 (3.2)	0.091
Gestational age, trimester (wk)				0.024
First (<14)	180 (20.9)	24 (16.3)	156 (21.9)
Second (14–27)	344 (40.0)	51 (34.7)	293 (41.1)
Third (≥28)	336 (39.1)	72 (49.0)	264 (37.0)
PE-related symptoms
Shortness of breath	637 (74.1)	107 (72.8)	530 (74.3)	0.7
Thoracic pain	540 (62.8)	73 (49.7)	467 (65.5)	<0.001
Cough without blood	237 (27.6)	103 (70.1)	134 (18.8)	<0.001
Palpitations	174 (20.2)	19 (12.9)	155 (21.7)	0.015
Fever	83 (9.7)	53 (36.1)	30 (4.2)	<0.001
Syncope or presyncope	56 (6.5)	3 (2.0)	53 (7.4)	0.016
Hemoptysis	14 (1.6)	3 (2.0)	11 (1.5)	0.7
DVT-related symptoms				0.013
None	802 (93.3)	144 (98.0)	658 (92.3)
Unilateral leg symptoms	58 (6.7)	3 (2.0)	55 (7.7)
Duration of PE-related symptoms (h)				0.019
<48	517 (60.7)	76 (52.1)	441 (62.5)
≥48	335 (39.3)	70 (47.9)	265 (37.5)
Unknown	8	1	7
Heart rate ≥ 110 beats/min	341 (39.7)	97 (66.0)	244 (34.2)	<0.001
Pulse oximetry (%)				<0.001
95–100	801 (93.9)	129 (87.8)	672 (95.2)
<95	52 (6.1)	18 (12.2)	34 (4.8)
Unknown	7	0	7
Max temperature ≥ 38.0°C	33 (3.8)	20 (13.6)	13 (1.8)	<0.001
Pregnancy-Adapted Geneva classification				<0.001
Low suspicion	458 (53.3)	48 (32.7)	410 (57.5)
Intermediate suspicion	371 (43.1)	93 (63.3)	278 (39.0)
High suspicion	31 (3.6)	6 (4.1)	25 (3.5)
Site of care				0.14
Emergency department	741 (86.2)	127 (86.4)	614 (86.1)
Labor and delivery unit	103 (12.0)	20 (13.6)	83 (11.6)
Clinic (in person or virtual)	16 (1.9)	0 (0.0)	16 (2.2)
Number of consultations				0.6
0	622 (72.3)	105 (71.4)	517 (72.5)
1	231 (26.9)	40 (27.2)	191 (26.8)
2	7 (0.8)	2 (1.4)	5 (0.7)

Abbreviations: DVT, deep vein thrombosis; IQR, interquartile range; PE, pulmonary embolism.

Note: n (col %) throughout, except where noted.

^a Other race categories include American Indian/Alaskan native ( n  = 4), native Hawaiian or Other Pacific Islander ( n  = 13), decline to state ( n  = 26).

During diagnostic evaluation, COVID-19 patients more often had elevated D-dimer values > 1.0 mg/L (49.1% vs. 36.4%; p  < 0.001) and more commonly underwent CXR (61.9% vs. 50.1%; p  = 0.009; Table 2 ). Among CXR recipients, infiltrates were more commonly found in COVID-19 patients than non-COVID-19 patients (33.0% vs. 3.4%; p  < 0.001). The relationship between D-dimer levels and the use of pulmonary vascular imaging showed a consistent trend in both cohorts: very few patients with low D-dimer levels (<0.5 mg/L) underwent advanced imaging, while higher D-dimer levels were associated with more frequent use of advanced imaging. Among the patients who underwent advanced imaging, CTPA was used predominantly, accounting for 90.3% of studies in the non-COVID-19 cohort and 93.2% in the COVID-19 cohort. Five COVID-19 patients underwent low-dose Q scans rather than CTPA. Four had prior CXR, which was negative in 3. Overall, PE was diagnosed in fewer than 1% of patients, with the prevalence of PE not significantly differing between cohorts. All-cause mortality was low, with three total deaths among the whole study cohort ( Table 1 ).

Table 2. Diagnostic testing and imaging characteristics of pregnant patients with suspected pulmonary embolism stratified by COVID-19 status.

	Overall	Active COVID-19
Characteristic	n  = 860	Yes, n  = 147	No, n  = 713	p -Value
ED COVID-19 testing				<0.001
Not performed	750 (87.2)	53 (36.1)	697 (97.8)
Performed and positive	94 (10.9)	94 (63.9)	0 (0.0)
Performed and negative	16 (1.9)	0 (0.0)	16 (2.2)
CXR performed				0.009
No	412 (47.9)	56 (38.1)	356 (49.9)
Yes	448 (52.1)	91 (61.9)	357 (50.1)
CXR result				<0.001
Infiltrate	42 (9.4)	30 (33.0)	12 (3.4)
Negative completely	384 (85.7)	56 (61.5)	328 (91.9)
Other	22 (4.9)	5 (5.5)	17 (4.8)
Not performed	412	56	356
D-dimer performed				0.21
No	229 (26.6)	33 (22.4)	196 (27.5)
Yes	631 (73.4)	114 (77.6)	517 (72.5)
D-dimer result (mg/L)				<0.001
<0.5	167 (26.5)	12 (10.5)	155 (30.0)
Advanced imaging done	4 (0.6)	0 (0.0)	4 (0.8)
No advanced imaging done	163 (25.8)	12 (10.5)	151 (29.2)
0.5–1.0	220 (34.9)	46 (40.4)	174 (33.7)
Advanced imaging done	81 (12.8)	19 (16.7)	62 (12.0)
No advanced imaging done	139 (22.0)	27 (23.7)	112 (21.7)
>1.0	244 (38.7)	56 (49.1)	188 (36.4)
Advanced imaging done	176 (27.9)	38 (33.3)	138 (26.7)
No advanced imaging done	68 (10.8)	18 (15.8)	50 (9.7)
Advanced pulmonary imaging				0.4
CTPA	357 (90.8)	69 (93.2)	288 (90.3)
Lung scintigraphy	36 (9.2)	5 (6.8)	31 (9.7)
Not performed	467	73	394
Advanced pulmonary imaging result				>0.9
PE negative	854 (99.3)	146 (99.3)	708 (99.3)
PE positive	6 (0.7)	1 (0.7)	5 (0.7)
All-cause mortality	3 (0.3)	1 (0.7)	2 (0.3)	0.4

Abbreviations: CXR, chest radiography; CTPA, computed tomography pulmonary angiography; ED, emergency department; PE, pulmonary embolism.

Note: n (col %) throughout.

Discussion

In this retrospective cohort study, we observed that SARS-CoV-2 infection was associated with differences in presenting symptoms, structured pretest probability, and diagnostic testing patterns among pregnant patients undergoing PE diagnostics. This work extends our previous APED analyses of clinical practice by exploring how an acute respiratory infection (i.e., COVID-19) influences clinician assessment and testing patterns. ^{17 18 19} Pregnant patients with COVID-19 were more likely to present with systemic and respiratory symptoms such as fever, cough, tachycardia, and hypoxemia, while those without COVID-19 more frequently exhibited classic thromboembolic unilateral leg symptoms and chest pain. Despite these differences in presentation, the prevalence of confirmed PE was < 1% in both groups.

One of the most notable differences we observed between our COVID-19 and non-COVID-19 cohorts was the higher prevalence of D-dimer values > 1.0 mg/L among infected patients. Several D-dimer-based diagnostic algorithms estimate the pretest probability of PE to help clinicians identify which patients suspected of acute PE warrant pulmonary vascular imaging. ^{37 41} Algorithm performance has faltered, however, when used among patients hospitalized with COVID-19, with some investigators suggesting that new prediction rules are needed in this unique, acutely ill population. ^{42 43} Though less attention has been directed to the performance of PE diagnostics algorithms among ambulatory patients with COVID-19 outside the hospital, some research suggests that “the strategy to safely exclude PE in COVID-19 outpatients should not differ from that used in non-COVID-19 patients.” ⁴⁴ This line of inquiry has not included pregnant patients.

COVID-19 itself is known to elevate D-dimer levels, particularly among patients with greater disease severity. ⁴⁵ In hospitalized and critically ill patients, elevated or rising D-dimer values frequently reflect a true increase in thromboembolic risk, including PE, and often justify closer monitoring, diagnostic testing, or empiric anticoagulation. Large population-based studies further support this severity-dependent risk: symptomatic patients with SARS-CoV-2 had a significantly higher 30-day incidence of VTE than those who tested negative, driven almost entirely by hospital-associated events (5.8 vs. 3.0 per 1,000, p  < 0.001), while outpatient VTE incidence remained comparable between groups. ¹² These findings underscore that the greatest thromboembolic risk occurs during severe COVID-19 illness requiring hospitalization, whereas in mild or ambulatory COVID-19 patients, elevated D-dimer may more often reflect systemic inflammation and virally mediated coagulopathy rather than true underlying PE. ⁴⁶

In our study, we observed that gravid patients with COVID-19 had a higher prevalence of elevated D-dimer values > 1.0 mg/L despite being largely ambulatory. This suggests that even in lower-risk, noncritically ill pregnant patients, acute COVID-19 infection may be sufficient to raise D-dimer levels independent of thromboembolic disease. The overlap between infection-related coagulopathy and physiologic pregnancy-associated increases in D-dimer values creates a diagnostic challenge, as thresholds commonly used to stratify PE risk may lack specificity in this infected pregnant population. These findings reinforce the importance of cautious interpretation of D-dimer values in the setting of concurrent pregnancy and respiratory infection, particularly given that pregnancy-adapted PE diagnostic algorithms were not designed with infection-related inflammatory states such as COVID-19 in mind. ^{34 40 47}

Differences in heart rate between the cohorts were notable, with COVID-19 patients having a higher prevalence of tachycardia. Infection is an established cause of tachycardia and is common among ambulatory patients with COVID-19, even those not requiring inpatient care. ^{48 49} Heart rate is a key driver of multiple PE pretest probability frameworks, including the Wells criteria, revised Geneva score, four-level PE Clinical Probability Score, and, for pregnant patients, the PAG score. In our study, patients with COVID-19 were approximately twice as likely to present with tachycardia compared with noninfected patients. A heart rate ≥ 110 beats per minute accounts for 25% of the overall PAG score, 5 out of a possible 20 points, and is sufficient on its own to place otherwise low-risk patients within the intermediate-risk category (defined as 2–6 points). ⁴⁰ In our COVID-19 cohort, 97 of 99 patients who were classified as intermediate or high risk met the tachycardia criterion. This finding, when taken with the fact that the prevalence of confirmed PE did not actually differ between the two groups, suggests that the heart rate-dependent pretest probability algorithms, such as the PAG score, may overestimate PE risk in patients with acute respiratory infections like COVID-19, where tachycardia may reflect systemic infectious stress rather than thromboembolic pathology. It is reasonable to conclude that this inflation of pretest probability may increase clinicians' suspicion for PE, leading to more diagnostic testing and pulmonary vascular imaging. It is important to note that while this study used the PAG score, as it has been specifically validated in pregnant patients, the use of alternative scoring systems, such as the Wells criteria, may have yielded different pretest probability classifications, as their variable weightings differ from those of the PAG score.

COVID-19 infection may also decrease clinicians' suspicion for PE in the same clinical scenario presented above. When a patient's symptoms and signs (e.g., dyspnea, cough, tachycardia, hypoxemia) can be attributed to a confirmed COVID-19 diagnosis, PE may no longer be considered “the most likely diagnosis,” as emphasized in the pregnancy-adapted YEARS algorithm. ³⁴ The same findings that elevate risk in one framework can simultaneously reduce it in another, depending on the diagnostic model employed. This dual effect highlights a key diagnostic challenge: clinicians must balance overlapping physiologic responses to infection and risk for thrombosis when applying pretest probability tools in the context of COVID-19.

Diagnostic imaging patterns further reflected this diagnostic uncertainty. COVID-19 patients underwent significantly more CXRs than the non-COVID-19 cohort. While CXRs have limited diagnostic value for ruling in or out PE, they are often the first recommended radiation-associated imaging modality for gravid patients with suspected PE, as they can identify other diagnoses that cause dyspnea and thoracic pain (e.g., pneumonia and pneumothorax). ^{37 50} Additionally, according to guidelines, CXRs are crucial in guiding the selection of appropriate pulmonary vascular imaging. ³⁷ Notably, only approximately half of patients underwent CXR despite presenting with symptoms concerning for PE, highlighting variability in real-world practice and potential opportunities to standardize early diagnostic evaluation. ¹⁹ Given the combined diagnostic uncertainty of COVID-19 and suspected PE, it is reasonable that clinicians would more frequently order CXRs in COVID-19 patients, not only to evaluate for alternative or coexisting diagnoses, but also to assess pulmonary involvement from the virus and help inform the choice between CTPA and lung scintigraphy. ^{37 50} This is further supported by our finding that COVID-19 patients had 10 times as many pulmonary infiltrates on CXR compared with their non-COVID-19 counterparts. Importantly, identification of pneumonia on CXR does not exclude concurrent PE. Thromboembolism has been reported in patients with radiographic pneumonia, likely reflecting infection-related systemic inflammation and endothelial activation. ^{51 52} Thus, abnormal chest imaging may simultaneously provide an alternative explanation for symptoms while not fully resolving concern for embolism, reinforcing the diagnostic complexity in this population. Given that COVID-19 pneumonia is a common complication of the infection, the higher prevalence of infiltrates is expected, but it also underscores clinicians' inclination to obtain CXRs in this cohort as part of the initial diagnostic work-up.

This study was undertaken during the COVID-19 pandemic. The findings from this study, however, have implications beyond the pandemic. SARS-CoV-2 is now part of the new postpandemic landscape, establishing a place among numerous respiratory pathogens. This means that clinicians will continue to encounter patients like those described in this study. But COVID-19 is not the only respiratory infection that can complicate PE diagnostics. We found that patients with COVID-19 suspected of a concurrent PE diagnosis were commonly febrile and tachycardic and sometimes hypoxemic with infiltrates on CXR. This clinical profile is not dissimilar to that of the influenza virus, RSV, or Streptococcal pneumonia. ^{6 53 54 55 56} The challenges we observed in diagnosing PE in gravid patients with COVID-19 are likely to be encountered when evaluating gravid patients with other respiratory infections who are suspected of concurrent PE. The identification of a non-PE cause of a patient's presenting signs and symptoms does not always curtail the diagnostic pursuit of PE, as studies among nongravid patients have shown. The diagnosis of PE is commonly pursued even when other explanatory diagnoses are thought more likely. ^{57 58 59 60}

Our findings raise important questions regarding the calibration of clinical decision-making during concurrent pregnancy and respiratory infection. Elevated D-dimer values, tachycardia, and respiratory findings, common to pregnancy, COVID-19, and PE, can both heighten and diminish clinical suspicion for PE, depending on whether these abnormalities are attributed to infection or embolism. As the healthcare system continues to manage overlapping infectious and obstetric concerns, there is a pressing need to refine and validate clinical decision tools that can better guide appropriate use of pulmonary vascular imaging in these complex scenarios. For pregnant patients with mild or stable COVID-19 illness, close monitoring and more holistic pretest assessments may help reduce unnecessary imaging and associated downstream risks.

Next Steps

Future research should prospectively validate PE diagnostic algorithms in pregnant patients with concurrent respiratory infections and examine whether clinical decision tools such as the PAG score require modification to account for infection-related physiologic changes. Additionally, qualitative studies exploring clinician perspectives on PE evaluation during periods of diagnostic uncertainty, particularly in the setting of other respiratory infections like influenza, could provide deeper insight into real-world clinical decision-making and support the development of more targeted risk stratification tools.

Limitations

This study has several limitations. As a retrospective analysis, our findings depend on the completeness and accuracy of clinicians' documentation, which may vary across settings. This study lacked systematic data on the severity of COVID-19 illness, which may have influenced clinician decision-making and could have added valuable context regarding imaging choices. As this was an observational study of real-world practice patterns, we were unable to confirm that every patient in the non-COVID-19 cohort tested negative, as not all patients were tested. Lack of universal testing may have misclassified COVID-19 patients as non-COVID patients, which may have blunted the between-group differences we observed in patient symptoms and vital signs. Additionally, clinicians aware of a patient's COVID-19 status may have had a lower threshold to initiate PE evaluation, introducing potential detection bias that could have influenced observed diagnostic testing patterns. Testing was performed at the clinicians' discretion based on the patient's clinical presentation. Finally, as this study was conducted within a single integrated healthcare system, our findings may not be generalizable to other settings with differing diagnostic workflows, resource availability, and clinical thresholds for PE evaluation.

Conclusion

Among ambulatory pregnant patients undergoing evaluation for suspected PE, COVID-19 was associated with more abnormal vital signs, higher D-dimer values, and higher objective pretest probability classifications despite similarly low rates of confirmed PE. These findings highlight how respiratory infections may influence PE diagnostic pathways in pregnancy and suggest that current pregnancy-adapted risk assessment approaches may require refinement for patients with concurrent respiratory infections.

Funding Statement

Funding Information The Kaiser Permanente Northern California Community Health Research Program.

What Is Known About This Topic?

• Diagnosing pulmonary embolism (PE) in pregnant patients is challenging because normal physiologic changes of pregnancy overlap with PE symptoms and elevate PE biomarkers such as D-dimer.

• Acute respiratory infections may further complicate diagnostics in pregnancy by causing symptoms and vital sign abnormalities that mimic PE.

What Does This Study Add?

• Pregnant patients with COVID-19 undergoing PE diagnostics more frequently had tachycardia, hypoxemia, elevated D-dimer values, and higher objective pretest probability classifications compared with non-COVID-19 pregnant patients, despite similarly low PE prevalence.

• Clinicians should be mindful how COVID-19 may significantly alter the diagnostic landscape of ambulatory pregnant patients with suspected PE; refinement of diagnostic strategies in pregnant patients with concurrent respiratory infections may be needed.