Risk Factors Associated with Delayed Diagnosis of Acute Pulmonary Embolism

Article Outline

• Abstract
• Introduction
• Methods
• Results
• Discussion
  • Limitations
• Conclusions
• References
• Copyright

Abstract 

Background

Prompt diagnosis and treatment of acute pulmonary embolism (PE) is essential to reduce mortality. Risk factors for PE are well known, but factors associated with delayed diagnosis are less clear.

Objective

Our objective was to identify clinical factors associated with delayed diagnosis of patients with acute PE presenting to a tertiary-care emergency department (ED).

Methods

We studied 400 consecutive adults who presented to our ED with acute, symptomatic PE. All patients were diagnosed by computed tomography (CT) angiography. Early diagnosis was defined as CT diagnosis<12h from ED arrival, and delayed diagnosis as CT diagnosis>12h. Univariate and multiple logistic regression models were used to identify factors associated with delayed diagnosis. Odds ratios with 95% confidence intervals are reported.

Results

The median time from arrival to diagnosis was 2.4h (interquartile range 1.4–7.6), and 73 (18.3%) patients had delayed diagnosis. Patients aged>65 years and those with coronary artery disease or congestive heart failure had longer times from ED arrival to CT diagnosis, whereas patients with recent immobility had shorter times. Patients diagnosed>12h were older and had higher rates of morbid obesity and coronary artery disease, whereas patients diagnosed<12h had higher rates of tachycardia. In multiple regression modeling, tachycardia and recent immobility remained associated with early diagnosis, whereas morbid obesity remained associated with delayed diagnosis.

Conclusions

Older patients with cardiovascular comorbidities had longer times from ED arrival to CT diagnosis. Our data suggest that these patients represent more of a diagnostic challenge than those presenting with traditional risk factors for PE, such as tachycardia and recent immobilization. Physicians should consider these factors to diagnosis acute PE promptly in the ED.

Keywords: pulmonary embolism, risk factors, timing, delay, diagnosis

Back to Article Outline

Introduction 

Acute pulmonary embolism (PE) is a common and potentially fatal disorder if not promptly diagnosed and treated 1, 2, 3, 4, 5. Even with anticoagulation, the 14- and 90-day mortality rates are approximately 10% and 20%, respectively 6, 7, 8, 9. Studies have previously shown potential for significant outpatient delays from initial symptom onset to the diagnosis of PE in the emergency department (ED) 1, 2, 3. Kline et al. reported that patients diagnosed within 48h of ED arrival had improved outcomes, and we recently reported that anticoagulation within 24h of ED arrival is associated with reduced mortality (4,5). Indeed, prompt diagnosis and treatment with anticoagulation are essential to improve outcomes and reduce mortality for patients with acute PE.

Current guidelines suggest the initiation of anticoagulation therapy even before confirmatory diagnosis if clinicians feel that the probability of PE is high, especially in hemodynamically unstable patients 1, 3, 10. However, many patients present without hemodynamic instability, and so emergency physicians must risk-stratify patients to make a prompt diagnosis. Traditional risk factors for PE have been well established, but factors associated with the timing of diagnosis have not been as well studied. We therefore conducted a retrospective review of patients with acute PE to identify clinical factors associated with delayed diagnosis.

Back to Article Outline

Methods 

We conducted a retrospective review of consecutive adult patients who presented to a single, tertiary-care ED with acute, symptomatic PE between June 17, 2002 and September 6, 2005 (4). Patients were identified based upon review of International Classification of Diseases, 9^th Revision codes 415.1–415.19. We included only patients with symptoms compatible with acute PE (i.e., chest pain, dyspnea, hypoxia, pre-syncope, or syncope), and patients were excluded if an asymptomatic PE was incidentally diagnosed during evaluation for symptoms other than those identified. Asymptomatic patients were excluded from our analyses to focus upon the diagnostic process rather than incidental diagnosis. Symptoms must have developed acutely and immediately before ED presentation so that lead-time bias would be minimized. Patients were included only if their PE diagnosis was made with computed tomography (CT) angiography at our institution; and patients diagnosed before arrival were excluded. Acute PE was defined as a filling defect on CT angiography that the attending radiologist felt was consistent with an acute rather than subacute or chronic process. In accordance with our ED’s clinical practice at that time, all patients were admitted and treated with intravenous, weight-based unfractionated heparin. Patients were excluded if anticoagulation was contraindicated.

The primary aim was to identify clinical factors (i.e., demographics, comorbidities, vital signs, or laboratory data) associated with the timing of CT diagnosis. Drs. Smith and Geske reviewed all patient records in an unblinded fashion for the ED evaluation and index hospitalization, including those from ED and hospital attending physicians, residents, and nurses. Primary data were collected based upon the information known and documented at the time of ED arrival, including symptoms, vital signs, medications, and past medical history. Clinical factors included: initial vital signs, age, gender, body mass index (BMI, kg/m²), history of prior venous thromboembolism (VTE), known hypercoagulopathy, active malignancy, chronic obstructive pulmonary disease (COPD), coronary artery disease (CAD), congestive heart failure (systolic or diastolic, CHF), diabetes mellitus, end-stage renal disease (ESRD) requiring dialysis, oral contraceptive pill (OCP) use, active nicotine use, and recent immobility. The following definitions were established: tachycardia as a heart rate>100 beats/min; hypotension as a systolic blood pressure<100mm Hg; morbid obesity (World Health Organization class III) as BMI>40; leukocytosis as a white blood count>10×10⁹ cells/mL; positive troponin-T as>0.01 ng/mL; positive D-dimer (high-sensitivity turbidimetric immunoassay) as>500 ng/mL. Recent immobilization was defined as documented hospitalization, surgery, or travel within 30 days before ED evaluation. Early diagnosis was defined as having the confirmatory CT<12hours from ED arrival, and delayed diagnosis was defined as a CT>12hours from arrival.

Median values were reported with interquartile ranges (IQR) because data were not normally distributed. Fisher’s exact tests and odds ratios with 95% confidence intervals were used to evaluate the association of categorical variables with the timing of diagnosis. A nominal multiple logistic regression model was used to estimate the probability of delayed diagnosis. Clinical factors were included in the multiple regression if they achieved statistical significance in univariate analysis as defined by a p<0.1. The type I error rate was otherwise set at 0.05 (2-sided) a priori, and no correction factor was applied to account for multiple comparison issues. Data were analyzed with JMP 9.0 (2010, SAS Institute Inc., Cary, NC). Statistical analyses were aided by the Center for Translational Science Activities at our institution. This study was approved by our Institutional Review Board.

Back to Article Outline

Results 

We identified 400 consecutive adult patients with a median age of 68.0 years (IQR 54.0–76.0); 195 patients (48.8%) were male (Table 1). The median time from ED arrival to CT diagnosis was 2.4h (IQR 1.4–7.6) (Table 1). Seventy-three (18.3%) patients had delayed diagnosis (Table 2, Table 3).

Table 1. Hours from Arrival to Diagnosis

Patients aged>65 years and those with CAD or CHF had longer times from ED arrival to CT diagnosis, whereas patients with recent immobility had shorter times.

BMI=body mass index; CHF=congestive heart failure; OCP=oral contraceptive pill; COPD=chronic obstructive pulmonary disease; VTE=venous thromboembolism; DVT=deep vein thrombosis; ESRD=end-stage renal disease.

Table 2. Continuous Variables Associated with Delayed Diagnosis

Continuous Variable	Diagnosis<12 Hours	Diagnosis>12 Hours
Age (years)	67 (53–75)	74 (62–81)∗∗∗
Temperature (°C)	36.8 (36.3–37.3)	36.9 (36.1–37.3)
Heart rate (beats/min)	93 (79–108)	83 (73–100)
SBP (mm Hg)	134 (118–154)	135 (117–153)
Respirations (breaths/min)	20 (18–24)	20 (18–27)
BMI (kg/m2)	29.5 (25.9–36.4)	29.5 (25.3–35.0)
WBC (109 cells/mL)	9.2 (7.4–12.3)	9.8 (7.9–13.3)
D-dimer (ng/mL)	1100 (588–2000)	1300 (650–2000)
Troponin T (ng/mL)	0.01 (0.01–0.02)	0.01 (0.01–0.03)

Patients diagnosed>12h from arrival were older than patients diagnosed<12h. Continuous variables are given with interquartile ranges (IQR).

SBP=systolic blood pressure; BMI=body mass index; WBC=white blood cell count.

Table 3. Categorical Variables Associated with Delayed Diagnosis

Categorical Variable	Diagnosis<12 Hours (n=327)	Diagnosis>12 Hours (n=73)	OR (95% CI)
Age>65 years	176 (53.8%)	53 (72.6%)∗∗∗	2.27 (1.30 – 3.97)
Male	160 (48.9%)	35 (48.0%)	0.96 (0.58 – 1.60)
Tachycardia	132 (40.7%)	20 (27.8%)∗∗	0.55 (0.32 – 0.98)
Hypotension	22 (7.0%)	4 (5.7%)	0.80 (0.27 – 2.40)
BMI >40	32 (10.3%)	14 (20.0%)∗∗	2.19 (1.10 – 4.36)
Smoking	33 (10.1%)	7 (9.6%)	0.94 (0.40 – 2.23)
Coronary disease	58 (17.7%)	23 (31.5%)∗∗	2.13 (1.21 – 3.77)
CHF	26 (8.0%)	10 (13.7%)	1.84 (0.84 – 4.00)
Coagulopathy	19 (5.8%)	3 (4.1%)	0.69 (0.20 – 2.41)
OCP use	22 (6.7%)	1 (1.4%)∗	0.19 (0.03 – 1.45)
COPD	34 (10.4%)	8 (11.0%)	1.06 (0.47 – 2.40)
History of VTE	60 (18.4%)	11 (15.1%)	0.79 (0.39 – 1.59)
Current DVT	48/171 (28.1%)	14/43 (32.6%)	1.24 (0.60 – 2.54)
Malignancy	108 (33.0%)	21 (28.8%)	0.82 (0.47 – 1.43)
Hemoptysis	12 (3.7%)	2 (2.7%)	0.74 (0.16 – 3.38)
Immobility	145 (44.3%)	24 (32.9%)∗	0.61 (0.36 – 1.05)
Diabetes	46 (14.1%)	14 (19.2%)	1.44 (0.75 – 2.80)
ESRD	3 (0.9%)	1 (1.4%)	1.50 (0.15 – 14.58)
Leukocytosis	136 (41.7%)	35 (48.0%)	1.20 (0.77 – 2.14)
D-dimer >500 ng/mL	128/158 (81.0%)	32/34 (94.1%)∗	3.75 (0.85 – 16.52)
Troponin T >0.01 ng/mL	68/251 (27.1%)	19/61 (31.2%)	1.22 (0.66 – 2.24)

Patients diagnosed>12h were older and had higher rates of morbid obesity and coronary artery disease, whereas patients diagnosed<12h had higher rates of tachycardia. Not all patients were assessed for DVT, and not all patients had D-dimer and Troponin-T laboratory tests obtained.

OR=odds ratio; CI=confidence interval; BMI=body mass index; CHF=congestive heart failure; OCP=oral contraceptive pill; COPD=chronic obstructive pulmonary disease; VTE=venous thromboembolism; DVT=deep vein thrombosis; ESRD=end-stage renal disease.

Patients with age>65 years, CAD, and CHF had significantly longer times from arrival to diagnosis (Table 1), although only age>65 years and CAD were univariate predictors of delayed diagnosis (Table 2, Table 3). Patients with COPD tended to have prolonged times from arrival to diagnosis (4.9 [IQR 1.6–8.7] vs. 2.3 [IQR 1.4–7.2] h, p=0.062), although COPD was not a univariate predictor of delayed diagnosis. Patients with recent immobility had reduced times from arrival to diagnosis (1.9 [IQR 1.3–4.6] vs. 2.9 [IQR 1.7–8.8] h, p=0.032), although recent immobility was not a univariate predictor of early diagnosis. Patients with OCP use tended to have reduced times to diagnosis (1.8 [IQR 1.2–3.3] vs. 2.4 [IQR 1.4–7.7] h, p=0.099), but OCP use was not a univariate predictor of early diagnosis.

The multiple regression model included those seven univariate factors that achieved statistical significance with p<0.1 for delayed diagnosis. These seven factors provided an appropriately fit model because 73 patients had delayed diagnosis (i.e., 10 events per one risk factor). Morbid obesity remained associated with delayed diagnosis, whereas tachycardia and recent immobility remained associated with early diagnosis (Table 4).

Table 4. Multiple Regression Model for Characteristics Associated with Delayed Diagnosis

Characteristic	OR (95% CI)	p Value
Age>65 years	1.38 (1.09 – 1.75)	0.173
Tachycardia	0.57 (0.45 – 0.73)	0.021
BMI>40	1.90 (1.45 – 2.49)	0.018
Coronary disease	0.86 (0.67 – 1.10)	0.532
OCP use	0.69 (0.39 – 1.20)	0.503
Immobility	0.63 (0.50 – 0.79)	0.039
D-dimer>500 ng/mL	1.90 (1.27 – 2.83)	0.111

Tachycardia and recent immobility remained associated with early diagnosis, whereas morbid obesity remained associated with delayed diagnosis.

OR=odds ratio; CI=confidence interval; BMI=body mass index; OCP=oral contraceptive pill.

Back to Article Outline

Discussion 

Early diagnosis is essential to reduce the high morbidity and mortality associated with acute PE. It is therefore essential for clinicians to recognize not only risk factors for PE but also the clinical factors that may delay management. In our study we identified several such factors that were associated with delayed diagnosis.

Previous studies have examined the demographics and comorbidities associated with the timing of the diagnosis of acute PE. Kline et al. studied 161 patients with acute PE who were diagnosed either within 48h or after 48h from ED arrival (5). They found that patients with delayed diagnosis were older, whereas patients with early diagnosis more frequently had recent surgery. Pineda et al. also described a trend towards younger age in a cohort where PE was diagnosed more accurately (6). Our data concur, and we hypothesize that older patients present more of a diagnostic challenge, given increased medical complexity and comorbid conditions.

Kline et al. considered a composite definition of various comorbid conditions (i.e., CAD, CHF, COPD, atrial fibrillation, history of VTE, ESRD, sleep apnea, sarcoidosis, or diaphragm weakness), and they found that patients with early or delayed diagnosis had similar rates of comorbid conditions (5). We considered comorbidities separately and found that CAD and CHF were associated with delayed management. Pineda et al. also found that PE diagnosis was made less accurately in patients with CAD. There was a non-significant trend towards delayed diagnosis in patients with COPD in our cohort. Several studies found that PE may be diagnosed less accurately in patients with COPD 11, 12. Indeed, one study suggested that 25% of patients hospitalized with a presumed COPD exacerbation may actually have PE (7). It stands to reason that when patients with a history of cardiopulmonary disease present with an acute PE, clinicians may attribute their symptoms to their known cardiopulmonary pathophysiology rather than acute PE.

Ageno et al. studied 542 patients with PE and found that patients diagnosed within 5 days of symptom onset had similar demographics and comorbidities as compared to patients diagnosed after 5 days (8). However, patients in their study who were diagnosed within 5 days were more likely to have transient risk factors for PE, such has recent immobilization or OCP use. Many studies have identified immobility and estrogens as risk factors for VTE 8, 9, 10, 13, 14, 15, 16, 17. We therefore hypothesize that when patients present with these well-known risk factors for PE, physicians are more likely to make an expedited diagnosis of PE.

The Wells and Geneva scores are established systems that use well-known risk factors to determine the pre-test probability that a patient has a PE 11, 14. We studied all of the clinical factors considered in the Wells and Geneva scores (i.e., age>65 years, tachycardia, hemoptysis, recent immobility, history of VTE, symptoms of current DVT, and malignancy) except whether an alternative diagnosis was more or less likely. A retrospective review of charts may not be sufficiently accurate for determining whether the clinicians felt that an alternative diagnosis was more or less likely than PE. Although tachycardia and recent immobility were associated with the timing of diagnosis, we found that a history of VTE, current DVT, hemoptysis, and malignancy were not.

Although it is not included in the Wells or Geneva scoring systems, obesity has been associated with PE 12, 18, 19. Our regression modeling found that morbid obesity is a clinical factor strongly associated with delayed diagnosis. These data suggest that, in addition to being a risk factor for developing PE, morbid obesity complicates the diagnostic evaluation of acute PE. The body habitus associated with morbid obesity presents a unique challenge to CT scanning, and we theorize that the logistics of CT scanning may have contributed to the diagnostic delay observed in our cohort (20). A decisive physical examination is also more challenging in the morbidly obese patient, which may have complicated diagnostic evaluations.

Our study differs in design from previous ones because we considered only time from ED arrival rather than the time from symptom onset. ED arrival time is a more accessible, universal, and accurate time-point because patient recollection and gradual or non-specific symptoms may confound the timing of symptom onset. Furthermore, our timeframe allows for characterization of the clinical factors that affect the timing of ED and early hospital evaluation, which do not begin until a patient arrives at the ED.

Only 214 (53.5%) patients were screened for acute DVT with ultrasonography, and D-dimer testing was obtained in only 192 (48.0%) patients. Neither concurrent DVT nor positive D-dimer were associated with the timing of CT diagnosis. However, far fewer patients had DVT or D-dimer testing if their CT diagnosis was delayed. Although not conclusive, this supports the hypothesis that ED clinicians were entertaining cardiopulmonary diagnoses other than VTE in patients for whom CT diagnosis was delayed.

Limitations 

Our study is limited as a single-center, non-randomized retrospective review that identifies associations rather than establishes cause and effect between clinical factors and delayed diagnosis. This was a chart review with unblinded authors, and no inter-rater variability was assessed. Comorbidities were defined solely by provider documentation in the chart review rather than independent assessment. The severity of PE was assessed only by the presence of tachycardia and hypotension, and echocardiography data were not abstracted. Indeed, if patients present with cardiogenic shock, and if their clinicians have high suspicion for PE, then formal CT diagnosis may be delayed while patients are stabilized and given empiric anticoagulation.

Back to Article Outline

Conclusions 

This study was designed to identify clinical factors associated with delayed diagnosis of acute PE. Patients with age>65 years, cardiovascular disease, and those with morbid obesity had delayed diagnosis, whereas those with recent immobility and tachycardia were diagnosed more expediently. Therefore, clinicians should be aware of these factors to provide expedient management of acute PE and to reduce the morbidity and mortality associated with such delays.

Back to Article Outline

References 

1. Elliott CG, Goldhaber SZ, Jensen RL. Delays in diagnosis of deep vein thrombosis and pulmonary embolism. Chest. 2005;128:3372–3376
   • View In Article
   • MEDLINE
   • CrossRef
2. Bulbul Y, Ozsu S, Kosucu P, Oztuna F, Ozlu T, Topbas M. Time delay between onset of symptoms and diagnosis in pulmonary thromboembolism. Respiration. 2009;78:36–41
   • View In Article
   • CrossRef
3. Ozsu S, Oztuna F, Bulbul Y, et al. The role of risk factors in delayed diagnosis of pulmonary embolism. Am J Emerg Med. 2011;29:26–32
   • View In Article
   • Abstract
   • Full Text
   • Full-Text PDF (160 KB)
   • CrossRef
4. Smith SB, Geske JB, Maguire JM, Zane NA, Carter RE, Morgenthaler TI. Early anticoagulation is associated with reduced mortality for acute pulmonary embolism. Chest. 2010;137:1382–1390
   • View In Article
   • CrossRef
5. Kline JA, Hernandez-Nino J, Jones AE, Rose GA, Norton HJ, Camargo CA. Prospective study of the clinical features and outcomes of emergency department patients with delayed diagnosis of pulmonary embolism. Acad Emerg Med. 2007;14:592–598
   • View In Article
   • CrossRef
6. Pineda LA, Hathwar VS, Grant BJ. Clinical suspicion of fatal pulmonary embolism. Chest. 2001;120:791–795
   • View In Article
   • MEDLINE
   • CrossRef
7. Rizkallah J, Man SF, Sin DD. Prevalence of pulmonary embolism in acute exacerbations of COPD: a systematic review and metaanalysis. Chest. 2009;135:786–793
   • View In Article
   • CrossRef
8. Ageno W, Agnelli G, Imberti D, et al. Factors associated with the timing of diagnosis of venous thromboembolism: results from the MASTER registry. Thromb Res. 2008;121:751–756
   • View In Article
   • Abstract
   • Full Text
   • Full-Text PDF (114 KB)
   • CrossRef
9. Al Otair H, Chaudhry M, Shaikh S, Bahammam A. Outcome of patients with pulmonary embolism admitted to the intensive care unit. Ann Thorac Med. 2009;4:13–16
   • View In Article
10. Le Gal G, Righini M, Roy PM, et al. Differential value of risk factors and clinical signs for diagnosing pulmonary embolism according to age. J Thromb Haemost. 2005;3:2457–2464
    • View In Article
    • MEDLINE
    • CrossRef
11. Le Gal G, Righini M, Roy PM, et al. Prediction of pulmonary embolism in the emergency department: the revised Geneva score. Ann Intern Med. 2006;144:165–171
    • View In Article
12. Goldhaber SZ, Grodstein F, Stampfer MJ, et al. A prospective study of risk factors for pulmonary embolism in women. JAMA. 1997;277:642–645
    • View In Article
    • MEDLINE
13. Anderson FA, Spencer FA. Risk factors for venous thromboembolism. Circulation. 2003;107(23 Suppl. 1):I9–16
    • View In Article
14. Wells PS, Anderson DR, Rodger M, et al. Derivation of a simple clinical model to categorize patients probability of pulmonary embolism: increasing the models utility with the SimpliRED D-dimer. Thromb Haemost. 2000;83:416–420
    • View In Article
    • MEDLINE
15. Farmer RD, Lawrenson RA, Thompson CR, Kennedy JG, Hambleton IR. Population-based study of risk of venous thromboembolism associated with various oral contraceptives. Lancet. 1997;349:83–88
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (76 KB)
    • CrossRef
16. van Hylckama Vlieg A, Helmerhorst FM, Vandenbroucke JP, Doggen CJ, Rosendaal FR. The venous thrombotic risk of oral contraceptives, effects of oestrogen dose and progestogen type: results of the MEGA case-control study. BMJ. 2009;339:b2921
    • View In Article
17. Poulter NR. Risk of fatal pulmonary embolism with oral contraceptives. Lancet. 2000;355:2088
    • View In Article
    • Full Text
    • Full-Text PDF (41 KB)
    • CrossRef
18. Holst AG, Jensen G, Prescott E. Risk factors for venous thromboembolism: results from the Copenhagen City Heart Study. Circulation. 2010;121:1896–1903
    • View In Article
    • CrossRef
19. Stein PD, Beemath A, Olson RE. Obesity as a risk factor in venous thromboembolism. Am J Med. 2005;118:978–980
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (96 KB)
    • CrossRef
20. Ginde AA, Foianini A, Renner DM, Valley M, Camargo CA. The challenge of CT and MRI imaging of obese individuals who present to the emergency department: a national survey. Obesity (Silver Spring). 2008;16:2549–2551
    • View In Article