Risk of Thromboembolism Varies, Depending on Category of Immobility in Outpatients

Introduction 

Immobility can cause stasis to venous blood flow and thus predispose to the development of venous thromboembolism.1 The categorical presence of immobility increases the probability a symptomatic patient will have venous thromboembolism.2 However, in practice, defining immobility presents a challenge inasmuch as immobility can occur in many forms and degrees of severity. For example, emergency department (ED) patients often have had recent bed confinement as a result of acute or chronic illnesses, dementia, or massive obesity. Traumatic injury, surgery, and orthopedic interventions usually result in some degree of body or limb immobility. Healthy patients experience immobility during long-distance travel for work or vacation. It could be reasonably conjectured that these different types of immobility confer differing degrees of risk of venous thromboembolism.

Previous research has examined immobility as a whole unit, without considering the importance of different causes. To our knowledge, no precedent literature has directly measured and compared the magnitude of risk of venous thromboembolism stratified by the cause of immobility in a symptomatic ambulatory population. We hypothesized that different causes of immobility would confer different degrees of risk for the outcome of venous thromboembolism in a large cohort of symptomatic ED patients who were evaluated for clinically suspected pulmonary embolism. Our overall purpose is to evaluate the overall risk of immobility in patients presenting to the ED who are evaluated for pulmonary embolism.

Materials and Methods 

This was a prospective, noninterventional, multicenter study of patients presenting to the ED in 12 hospitals in the United States. The study protocol was approved by the institutional review boards for the conduct of research on humans and the privacy boards of all 12 hospitals. Patients were enrolled from July 1, 2003, until November 30, 2006, as previously described.3 Briefly, patients were enrolled either consecutively, requiring greater than 85% capture, or during randomly selected shifts. At one site, a convenience sample used a paper instrument (n=150) as previously published to test for interobserver variability.4 Eligibility for enrollment required an order for an objective diagnostic test for pulmonary embolism, written by or under the supervision of a board-certified emergency physician. The decision to order this test was based on information obtained from the initial history and physical examination and medical records immediately available in the ED. Pulmonary vascular imaging study (computed tomographic [CT] angiography, pulmonary angiography, or scintillation ventilation-perfusion lung scanning) or a D-dimer assay order was considered an acceptable criterion to evaluate for possible pulmonary embolism. Diagnostic testing ordered solely to evaluate for suspected deep venous thrombosis did not trigger enrollment, but patients who had previous positive Doppler results for deep venous thrombosis could be included in enrollment. Exclusions were (1) clinician knowledge of a diagnostic positive pulmonary vascular imaging study result performed within the previous 7 days; (2) the patient unable or unwilling to indicate with certainty that he or she would return to the same hospital for reevaluation in event of persistent problems; and (3) a circumstance that suggested a high probability that patient would be lost to follow-up.

Data were entered into a Web-based, 128-key bit encrypted, electronic data collection form that encoded 74 data fields. A full description of the content, appearance, and methodology for this data collection system has been published.5 Data were entered while the patient was in the ED, using the information and beliefs of the clinicians who had ordered the diagnostic test that triggered enrollment. All fields had explicit, written definitions, which were visible in cursor-activated hover boxes. The immobility pull-down box coded for 6 mutually exclusive possibilities. These are listed below in the order shown on the screen, and the exact definitions that were visible to the user from the hover boxes are provided in quotations5:

The enrollment procedure specified that the Web-based electronic form had to be completed before the results of testing were known and required the clinician to input his or her knowledge of any test results pertinent to diagnosis of venous thromboembolism at the time the form was populated. The server side-code would not allow the form to be uploaded with missing data. Using a combination of telephone follow-up, chart review, and death records, we recorded patient outcomes at 45 days.3 Outcomes were recorded on a second, more extensive, Web-based follow-up form that encoded 143 data fields based on scripted telephone interviews and were supplemented by review of each patient's medical record using previously described methods.6

We tested interobserver reliability for the overall designation of immobility in a convenience sample of 150 patients at Carolinas Medical Center. For this measurement, a clinician of equal training level to the clinician who populated the Web-based data collection instrument reapproached these 150 patient-participants to repeat the process of asking and documenting the bivariate designation of immobility as either present or absent. This repeated result was recorded on a paper form. The second clinician had to check a box to verify that he or she was not aware of the beliefs of the first clinician about immobile status. A study associate collected these forms on a regular basis and recorded their results.

The criterion standard outcome of venous thromboembolism included was established by an adjudication process requiring independent agreement of 2 physicians who were blinded to immobility status but were given all imaging results, treatment plans, death certificates, autopsy reports, and results of telephone follow-up.6 Adjudicators used explicit criteria to determine venous thromboembolism positive (+) outcome, which required image-proven pulmonary embolism or deep venous thrombosis, together with written evidence of intention to treat or actual administration of systemic anticoagulation for greater than 89 days or insertion of vena caval filter in patients with a contraindication to anticoagulation. Diagnosis of pulmonary embolism required either a high-probability ventilation-perfusion scan or CT angiogram or conventional pulmonary angiogram that demonstrated a pulmonary arterial filling defect interpreted as positive for acute pulmonary embolism within 45 days. Diagnosis of deep venous thrombosis required extremity venous duplex Doppler-ultrasonography or CT venography result interpreted as positive for acute venous thrombosis in the popliteal, femoral, or axillary (but not calf) veins within 45 days, with intention to treat. All imaging results were from the final interpretations signed by a board-certified radiologist. Additionally, pulmonary embolism or deep venous thrombosis from autopsy within 45 days was considered definitive.

Results 

The study cohort included 7,940 patients with the clinical characteristics shown in Table 1. Five hundred forty-five patients were venous thromboembolism+, representing 6.9% (545/7,940) of the cohort. Of the 545 venous thromboembolism+ patients, 354 were diagnosed with pulmonary embolism only, 72 had isolated deep venous thrombosis, and 119 had pulmonary embolism and deep venous thrombosis.

Table 1. Clinical features of the cohort of 7,940 ED patients with suspected pulmonary embolism.

	Feature	Mean or n	(SD) or %
	Demographic data
	Age, y	49	(17)
	Black	2,704	34
	White	4,541	57
	Latino or Hispanic⁎	482	6
	Asian	74	1
	Other race	131	2
	Female sex	5,328	67
	Symptoms
	Pleuritic chest pain†	3,660	46
	Substernal chest pain	2,909	37
	Dyspnea	4,260	54
	Syncope	479	6
	Cough	2,324	29
	Hemoptysis	227	3
	Risk factors for venous thromboembolism
	Active malignancy‡	489	6
	Immobility	1,394	18
	Recent surgery§	520	7
	Prior pulmonary embolism or deep venous thrombosis	858	11
	Physical findings
	Highest pulse rate, beats/min	92	(21)
	Highest respiratory rate, breaths/min	20	(5)
	Lowest systolic blood pressure, mm Hg	131	(24)
	Lowest room air pulse oximetry, %	97	(4)
	Temperature, °C	37(99.6°F)	(1)
	Body mass index, kg/m2	29	(8)
	Wheezing	822	10
	Unilateral leg swelling	710	9

⁎ Recorded as a race. † Nonsubsternal and worse with breathing or cough. ‡ Currently active cancer or metastatic cancer. § Surgery or trauma within 4 weeks and requiring endotracheal intubation.

Clinicians classified 1,394 of 7,940, or 18% of the cohort, as having at least 1 of the categories of immobility, leaving 6,546 with no immobility. Interobserver reliability data from the convenience sample of 150 patients at one center demonstrated moderate agreement for the designation of the presence or absence of immobility, as evidenced by overall raw agreement of 85%, with κ=0.48 (95% CI 0.32 to 0.68).

Table 2 shows a breakdown of each classification of immobility according to outcome of venous thromboembolism+ or venous thromboembolism-. The most frequently observed category of immobility was travel, accounting for 631 of 1,394 patients, or 45% of all patients coded as having immobility.

Table 2. Breakdown of differing immobilities associated with venous thromboembolism+/−.

	Category	VTE (+)	VTE (−)	%	PE (+)	%	DVT (+)	%	PE(+), DVT(+)	%
	No immobility	386	6,157	6	258	4	49	1	79	1
	General	56	323	14	31	8	9	2	16	4
	Limb	27	122	17	17	11	3	2	7	5
	Travel	42	589	6	29	5	3	0	10	2
	Neurologic	8	45	13	6	11	0	0	2	4
	Other	26	156	13	13	7	8	4	5	3

VTE, Venous thromboembolism; PE, pulmonary embolism; DVT, deep venous thrombosis.

The relative risk ratios for each category of immobility are presented in Table 3. Four categories of immobility demonstrated relative risk ratios with associated 95% CIs above unity: limb, general, neurologic, and other. Travel had the lowest relative risk of all 6 categories (relative risk 1.13 [95% CI 0.83 to 1.54]). Risk ratios in Table 3 were computed as the incidence of venous thromboembolism in the patients in each category of immobility divided by the incidence of venous thromboembolism in the 6,546 patients without immobility.

Table 3. Relative risk ratios⁎ of the patients with different categories of immobility developing venous thromboembolism.

	Category	VTE(+)	PE(+)	DVT(+)	PE(+), DVT(+)
	General	2.50(1.93–3.25)	1.47(1.05–2.11)	0.46(0.24–0.88)	0.79(0.49–1.30)
	Limb	3.07(2.16–4.38)	2.05(1.31–3.27)	0.40(0.13–1.25)	0.91(0.44–1.90)
	Travel	1.13(0.83–1.54)	0.79(0.55–1.15)	0.09(0.03–0.27)	0.28(0.15–0.52)
	Neurologic	2.56(1.34–4.88)	1.97(0.93–4.25)	0.00(NA)	0.72(0.19–2.81)
	Other	2.42(1.68–3.50)	1.29(0.77–2.22)	0.82(0.42–1.64)	0.52(0.22–1.25)

⁎ Denominator incidence computed only for patients free of immobility.

Results of multivariate logistic regression analysis are presented in Table 4. For the immobility variables, the odds ratios mirrored the results of univariate analysis. Limb and other immobility were most strongly associated with the outcome of venous thromboembolism+ (OR 2.25 [95% CI 1.40 to 3.60]; OR 1.97 [95% CI 1.25 to 3.09] respectively), followed by general immobility (OR 1.76 [95% CI 1.27 to 2.44]). Neurologic immobility had borderline significance (OR 2.23; 95% CI 1.01 to 4.92). Travel demonstrated no evidence of significant association with the outcome of venous thromboembolism+ (OR 1.19 [95% CI 0.85 to 1.67]). The absence of any immobility was associated with significantly lower odds of the outcome of venous thromboembolism+ within 45 days.

Table 4. Odds ratio data from multivariate logistic regression.

	Feature	OR	95% CI
	Age >50 y	1.50	1.25–1.82
	Female sex	0.62	0.51–0.74
	Unilateral leg swelling	2.68	2.13–3.37
	Metastatic or active cancer	2.24	1.69–2.95
	Surgery within 4 weeks	2.12	1.61–2.81
	Trauma in previous 4 weeks	0.78	0.37–1.62
	Personal history of VTE	3.00	2.41–3.71
	Body mass index >40 kg/m2	1.13	0.85–1.49
	Travel	1.19	0.85–1.67
	Neurologic	2.23	1.01–4.92
	Limb	2.25	1.40–3.60
	Other	1.97	1.25–3.09
	General	1.76	1.27–2.44

Limitations 

Limitations of this study include the fact the population was preselected according to a clinical presentation that prompted an emergency physician to initiate a formal evaluation for pulmonary embolism. Thus, we can make no inference about the risk of immobility for the outcome of venous thromboembolism for ED patients who did not elicit an evaluation for pulmonary embolism. Patients who had a known deep venous thrombosis but for whom the physician believed an investigation was not warranted for pulmonary embolism were not included in the data set. Our data do not exclude the possibility of increased risk of venous thromboembolism conferred by longer or more constraining types of travel such as long-haul flights lasting greater than 12 hours. Additionally, although it is our belief that our follow-up procedure correctly codified all patients who had diagnosed venous thromboembolism within 45 days, it remains possible that additional patients had pulmonary embolism or deep venous thrombosis that was not diagnosed. Areas of uncertainty include the preponderance of women in the cohort. This was a relatively uniform phenomenon, observed at 10 of 12 of the enrolling sites. Additionally, further research will be required to parse the significance of the “other” classification. This category represents a black box; it might reflect the product of gestalt reasoning or it could have been a catch-all category that included morbidly obese and very frail patients.

Discussion 

In this large, multicenter sample of symptomatic ED patients who underwent clinical evaluation for suspected pulmonary embolism, we found that different categories of immobility conferred variable risk for the outcome of venous thromboembolism at 45 days' follow-up. Patients who had generalized and limb immobility had significantly increased risk for venous thromboembolism that remained significant in univariate and multivariate analysis. Conversely, the absence of immobility significantly reduced the probability of venous thromboembolism within 45 days. We believe the finding that exposure to travel conferred only a minor significant increase in risk of venous thromboembolism+ and represents an important and potentially controversial observation. To our knowledge, this study represents the first prospective study to examine the significance of these categories of immobility as risk factors for venous thromboembolism in the ED setting. These data provide evidence that clinicians should consider immobility as a heterogeneous entity that should be parsed according to its root cause. For example, our data support the notion that clinicians are not obligated to consider every traveler who presents to the ED with symptoms suggestive of pulmonary embolism as high risk without other concrete variables that would be indicative of pulmonary embolism.

In the data entry form, any type of travel modality was allowable as long as the criterion of greater than 8 hours causing prolonged immobility in the previous 7 days was met. The transporting vehicle (eg, automobile, train, long-haul truck, or airplane) was not recorded, and not known was the percentage of air travel to other modalities. To our knowledge, all previous evidence showing increase risk of venous thromboembolism to travel has been derived from travelers in airplanes,7 as opposed to travel in land-based vehicles. Although many experts would assert a risk exists between venous thromboembolism and air travel, we would point out that no study has yet specifically examined the risk of venous thromboembolism conferred by air travel in a large symptomatic cohort. Unfortunately, the present study does not allow a specific inference about the risk of transcontinental air travel, nor can a comparison of air travel versus land travel be made. Moreover, the exact mechanism(s) of venous thromboembolism risk from air travel remain uncertain and could include other influences, including barometric pressure changes and dehydration.8, 9 Because different modalities of travel were not distinguished, if there had been an overwhelming bias of air travel venous thromboembolism compared with other modalities, it would not have been elucidated in this study. Further investigation will be necessary to confirm whether air travel specifically represents a risk factor for the outcome of venous thromboembolism in symptomatic ED patients.

Methodological strengths of this study include efforts to reduce sampling bias that can accompany convenience sampling. Patients were obtained from urban, rural, academic, and community practice settings, resulting in a data set mimicking the US population in terms of race and ethnicity. Explicit, written definitions of the categories of immobility were used. Our data collection system ensured that data were collected in real time in the ED and that data collection forms contained no missing or nonsense data.

In conclusion, data from a large cohort of symptomatic ED patients found that 3 categories of immobility, limb, neurologic, and whole-body immobility, significantly increased risk of venous thromboembolism, whereas travel greater than 8 hours did not confer increased risk of venous thromboembolism. These data show the importance of clarifying the cause of immobility in risk assessment of venous thromboembolism.