Travel and venous thrombosis: a systematic review

Authors


S. Middeldorp, MD PhD internist, Department of Clinical Epidemiology, C9-P, Leiden University Medical Center, PO Box 9600, 3500 RC Leiden, the Netherlands.
(fax: +31 71 5266994; e-mail: s.middeldorp@lumc.nl).

Abstract

Abstract.  Kuipers S, Schreijer AJM, Cannegieter SC, Büller HR, Rosendaal FR, Middeldorp S (Leiden University Medical Center, Leiden; and Academic Medical Center, Amsterdam; and Leiden University Medical Center, Leiden, the Netherlands). Travel and venous thrombosis: a systematic review (Review). J Intern Med 2007; 262: 615–634.

In the past decade, numerous publications on the association between venous thrombosis (VT) and travel have been published. Relative and absolute risks of VT after travel, and particularly after travel by air, have been studied in case–control and observational follow-up studies, whereas the effect of prophylaxis has been studied through intervention trials of asymptomatic clots. The mechanism responsible for the association between travel and VT was addressed in pathophysiologic studies. Here, we systematically reviewed the epidemiologic and pathophysiologic studies about the association between travel and VT.

We conclude that long-distance travel increases the risk of VT approximately two to fourfold. The absolute risk of a symptomatic event within 4 weeks of flights longer than 4 h is 1/4600 flights. The risk of severe pulmonary embolism (PE) occurring immediately after air travel increases with duration of travel, up to 4.8 per million in flights longer than 12 h. The mechanism responsible for the increased risk of VT after (air) travel has insufficiently been studied to draw solid conclusions, but one controlled-study showed evidence for an additional mechanism to immobilization that could lead to coagulation activation after air travel.

Introduction

Venous‘thrombosis (VT) is a serious disease that affects approximately 2–3 per 1000 persons per year [1, 2]. Both genetic and environmental factors are known to increase the risk of VT and these are mainly associated with procoagulant changes of the blood or immobilization. Prevalent genetic risk factors for VT are factor V Leiden mutation [3] and prothrombin G20210A mutation [4], each present in several percent of the population. Environmental factors that increase the risk of VT include oral contraceptive use, pregnancy, recent surgery, major trauma, immobilization and malignant diseases [5]. In the last decade, it has become clear that long-distance travel increases the risk of VT as well.

The first four cases of VT associated with air travel were described in 1951 [6]. Since then, many case-reports and case-series have been published on VT associated with not only air travel, but also travel by train, bus or car and even tractor driving [7]. The term economy class syndrome was coined in 1977 [8] and the first controlled study was published in 1986. Sarvesvaran and colleagues studied the causes of death occurring at a large international airport and concluded that pulmonary embolism (PE) occurred more often in the arrival hall than in the departure hall [9]. More controlled studies were not conducted until a young woman died of PE at Heathrow Airport in 2000. Since then, numerous reports have published results of case–control, follow-up and intervention studies on the association between air travel and VT. Furthermore, several studies have looked into the mechanism responsible for VT after air travel. A number of investigators have studied the effect of prolonged immobilization with or without the combination with flight-related factors, such as hypobaric hypoxia or dehydration.

The objective of this systematic review is to quantify the risk of VT after long-distance travel, when possible to assess the effect of various prophylactic measures on this risk and to summarize the available literature about potential mechanisms of the association between air travel and VT.

Methods

A systematic literature search was performed to identify all studies that included data on long-distance travel and VT. Studies that included epidemiological data on absolute and relative risks of VT after any kind of travel, randomized controlled trials that assessed the effect of prophylactic measures and publications that described pathophysiological studies were included.

Search strategy

Publications were identified through an extensive search, using PubMed, Embase, Web of Science and the Cochrane Central Register of Controlled Trials. The complete search strategy is shown in the Appendix. We did not apply a language restriction and searched all databases until 1 January 2007. Two reviewers independently screened the titles of all retrieved records for obvious exclusions. The same two reviewers read all remaining abstracts to identify eligible studies. Differences were solved by discussion.

Exposures of interest and outcomes

The main exposure of interest was travel, irrespective of mode of transportation and duration of travel. Studies that assessed the effect of prolonged immobilization and hypobaric hypoxia were evaluated as well. The main outcome of interest was symptomatic deep vein thrombosis and pulmonary embolism, diagnosed by objective methods (ultrasound, venography, ventilation-perfusion scanning, spiral CT-scanning, angiography or at autopsy). We also considered asymptomatic VT (diagnosed by objective methods), although of unclear clinical significance, and the effect of prophylactic interventions. The effect of any of the exposures of interest on coagulation parameters was the main outcome in the pathophysiological studies.

Quality assessment

All studies were judged on both internal and external validity by two reviewers independently, according to the guidelines of the Cochrane Collaboration Handbook [10]. Disagreement was solved by discussion and when no consensus could be reached a third reviewer was consulted.

We considered case–control studies to have a low risk of bias (i.e. to have a good internal validity) when selection bias of cases and controls was unlikely (when they came from the same population and travel frequency did not influence the likelihood of inclusion in the study), when travel frequency was assessed in the same way in cases and controls, when recall bias was minimized, when VT was diagnosed by objective means and when cases were consecutive, unselected patients with a first thrombotic event.

Follow-up studies that assessed the risk of VT in groups of travellers were considered to have a low risk of bias when loss to follow-up was <10%, when details of the exposure of interest were mentioned (mode of transportation, duration of travel and number of flights) and when the outcome of interest was assessed in the same way in all study participants. Symptomatic VT had to be diagnosed by the objective methods as described above.

Intervention studies were included when they assessed the effect of prophylactic measures on the risk of VT (both symptomatic and asymptomatic). They were considered to have a low risk of bias when randomization procedure and allocation concealment were adequate, when outcome assessors were blinded for the exposure status of the participants and when loss to follow-up was described and <10%. Ideally, study participants were blinded as well.

We included pathophysiological publications when they contained original data of studies about the effect of either travel or one of its specific factors (such as immobilization or hypobaric hypoxia) on thrombin generation or fibrinolysis in humans. Ideally, pathophysiological studies assessed the effect of an exposure of interest as compared with a control situation that would be exactly the same as the exposure situation except for the exposure itself. This would rule out other effects, such as circadian rhythm.

Data extraction:

For epidemiological studies, we used standardized forms for extraction of the following data:

  • • Case–control studies: source population of cases and controls, number of cases and controls, methods of diagnosis, disease characteristics (types of thrombotic events that were included), general characteristics of cases and controls (age, sex, prevalence of risk factors for VT), frequency of travel in both study groups and when possible mode of transportation, duration of travel and time interval between travel and event or index date.
  • • Follow-up studies: method of selection and inclusion of the study participants, numbers of participants (when applicable per subgroup), presence of a nontravelling control population, general characteristics (age, sex and prevalence of other risk factors for VT), outcome assessment, frequency of all relevant outcomes (symptomatic VT and asymptomatic thrombi) and numbers lost to follow-up.
  • • Prophylactic intervention studies: method of recruitment of participants, details of the treatment (type of stockings, dosage and frequency of any pharmacological treatment), use of placebo, method of randomization, concealment of allocation, method of outcome assessment, frequency of all relevant outcomes per treatment group, occurrence of adverse outcomes per treatment group (such as hemorrhagic complications when antithrombotics were studied) and numbers lost to follow-up.
  • • Pathophysiological studies: general characteristics, presence of a control population, intervention (immobilization, hypobaric hypoxia or travel), outcomes, assessment of outcome of interest (methods and timing) and main results.

Statistical analysis

All reported odds ratios from case–control studies were extracted. When possible, we pooled odds ratios to estimate relative risks for both air travel and travel by other modes of transportation. Pooling was performed using the inverse-variance-weighted average of the log odds ratios from the individual studies.

From follow-up studies, we calculated the absolute risk of symptomatic VT per flight. We also calculated the risk of asymptomatic thrombi per flight. When possible, we calculated incidence rates of VT within a few weeks after a long-haul flight. When data on different modes of transportation and duration of travel were available, we calculated risks per flight and incidence rates for each mode of transportation and duration of travel separately. We did not attempt to pool the data from follow-up studies, because of the anticipated differences in study design and participants.

From prophylactic intervention studies, we calculated absolute risks of thrombotic events per flight per intervention group. Furthermore, relative risks of the treated groups versus the control groups were calculated and, when possible, the data were pooled. To assess heterogeneity, we calculated the I2-statistic. This describes the percentage of the variability in effect estimates that is because of heterogeneity rather than chance. We considered heterogeneity present when I2 was >50%.

Because of the diversity of the study designs, no attempt to pool data for pathophysiological studies was made.

Results

At the first search (see Appendix for strategy) we found a total of 4154 titles. Based on the title, 3626 papers were excluded. We screened the abstracts of 528 publications, after which 10 publications with case–control data, 20 papers describing observational follow-up studies, 11 reports on intervention studies, 113 with case-reports or small case-series and 14 describing studies on the possible mechanism causing VT after long-distance travel were identified (Fig. 1). The remaining publications were comments, reviews, letters or editorials or did not concern VT and long-distance travel.

Figure 1.

 Results of the literature search. *These publications were editorials, reviews or comments or did not concern travel-related thrombosis.

Case–control studies – estimate of relative risks

We identified 10 publications in which travel frequency of cases with symptomatic VT was compared with a control population without VT (Table 1) [11–20]. One publication was excluded because the data were also used in a subsequent more extensive publication [11]. There were four studies with an increased potential of bias. In one study, cases were self-reported, without verification of the diagnosis and both cases and controls were selected frequent travellers [14]. In three studies, individuals with suspected VT in whom the diagnosis was ruled out were used as control persons [15–17], which may have caused over-representation of travel exposure in the controls and thus underestimation of the effect of travel. In these studies with a potential of bias, the odds ratios for any travel ranged from 0.5 to 1.3, whereas in the other studies, the odds ratios ranged from 1.8 to 4.0[12, 13, 18–20]. The pooled odds ratio of all studies together was 1.7 (95% CI 1.4–2.1). After exclusion of the four studies with a potential bias, this increased to 2.3 (95% CI 1.8–2.9). Six studies contained data on air travel only or separately for air travel and other modes of travel [15–20]. The pooled odds ratio for air travel of any duration of all studies was 1.4 (95% CI 0.9–2.0). After exclusion of three studies with a potential bias [15–17], the pooled odds ratio of the remaining studies [18–20] was 1.9 (95% CI 1.2–2.8). Three publications showed data on long-distance air travel [15, 18, 19], defined as flights longer than 8 h, with odds ratios ranging from 1.3 to 7.9 and a pooled odds ratio of 1.9 (1.1–3.6). After exclusion of one study with a potential bias [15], this pooled odds ratio increased to 3.9 (95% CI 1.4–10.7).

Table 1.   Case–control studies
First author/year of publication [reference]CasesControlsTravel dataPotential biasesn cases/ controls Travel cases/ controlsOR (95% CI)
  1. VT, venous thrombosis; DVT, deep vein thrombosis; PE, pulmonary embolism; GP, general practitioner.

  2. aThe study of Ten Wolde includes data from a previous publication by Kraaijenhagen and colleagues [11].

  3. **OR calculated with conditional logistic regression analysis.

Ferrari 1999 [12]Hospitalized at cardiology department for DVT or PEHospitalized for the first time at same department for other reasons Any travel
>4 h
<4 weeks
Selection of cases and controls (severe cases are hospitalized and patients hospitalized for other reasons may have travelled less) → underestimation of the effect160/16039/123.98 (1.9–8.4)
Samama 2000 [13]Consecutive DVTPatients of GP with flu-like symptomsNot specifiedSelection of controls: those visiting GP may have lower travel frequency → overestimation of the effect636/63662/312.4 (1.5–3.8)
Dimberg 2001 [14]Self-reported or insurance-claimRandom from insurance records‘Corporate air travel’
<30 days
Cases not objectively confirmed (self-reported), selection of frequent travellers → under or overestimation of the effect30/8913/1630.5 (0.1–1.6)
Arya 2002 [15]Consecutive DVTSuspected DVTAny travel
>3 h
<4 weeks
Selection of controls: travel creates symptoms similar to those in VT (oedema) → overrepresentation of travel in control population → underestimation of the effect185/38320/311.3 (0.6–2.8)
Hosoi 2002 [17]Consecutive DVTSuspected DVTAny travel
<2 weeks
101/10615/121.2 (0.6–2.8)
Ten Woldea 2003 [16]Consecutive, DVT and PESuspected DVT/PEAny travel
>3 h
<4 weeks
477/147032/1050.9 (0.6–1.4)
Martinelli 2003 [18]PE and DVT, visiting thrombosis centre for thrombophilia screenPartners and friendsAir travel
>4 h
<4 weeks
Selection of cases who underwent thrombophilia screen. Selection of controls: partners/friends often travel together, no matched analysis performed → underestimation of the effect210/21031/162.1 (1.1–4.0)
Parkin 2006 [19]Fatal PERandomly from electoral rollAir travel
>3 h
<4 weeks
Selection of only fatal cases → recall bias → overestimation of the effect89/3345/91.8 (0.5–7.1)
Cannegieter 2006 [20]Consecutive DVT and PE registered at anticoagulation clinicPartners of casesAny travel
>4 h
<8 weeks
Selection of controls: partners often travel together, but a matched analysis was performed1906/1096233/1822.1 (1.5–3.0)**

Observational follow-up studies – estimates of absolute risks

Twenty publications reported data on observational follow-up studies [14, 19, 21–38]. Two publications [36, 37] contained data that were also used in another publication [28]. From three studies, no absolute risks could be calculated, because either the number of events or the number of flights was not provided [14, 34, 35]. Of one paper [38], no full text version could be retrieved. The remaining 14 publications are listed in Table 2.

Table 2.   Observational follow-up studies
First author, year of pub [reference]Study populationExaminationsFlight durationLimitationsOutcomeAbsolute risk (95% CI) And time window
  1. aUS, ultrasonography; DVT, deep vein thrombosis; VT, venous thrombosis; FU, follow-up; PE, pulmonary embolism.

  2. bDVT and superficial thrombophlebitis (STF) were mainly asymptomatic, detected by ultrasound.

Outcome: asymptomatic DVT
 Arfvidsson 2001 [21]83 visitors of conference in Hawaii, voluntary participationUSa in all passengersMean 9 hResponse 30%, incomplete follow up, only asymptomatic thrombosesDVTb4 weeks
1.2% (0–3.6)
 Belcaro 2001 [26]355 passengers at low risk for VT and 389 passengers at higher risk. Method of recruitment unclearUS of all passengers,10–15 hSelection of participants unclear, overlapping inclusion and exclusion criteria, unclear whether any event was symptomaticDVTb24 h
1.5%
 Schwarz 2002 [72]160 passengers making ≥2 flights and 160 controls, recruitment through advertisements
No use of stockings or anticoagulants
US in all travellers and controls. ≥8 hOnly asymptomatic thrombosesDVTb travellers nontravellers48 h after return flight
0%
0%
 Hughes 2003 [24]878 individuals making ≥2 long-distance flights in 6 weeks, recruited through media, no severe risk factors, no increased d-dimers at baselineD-dimer and clinical probability US when either high, FU by telephone after 3 months>4 hMainly asymptomatic casesDVTb2 weeks after return flight
1% (0.4–1.7)
 Schwarz 2003 [23]964 passengers making ≥2 flights and 1213 nontravelling controls, recruitment through advertisements
No use of stockings or anticoagulants, no thrombus at baseline examination
Thrombophilia screen, d-dimers and US≥8 hOnly asymptomatic casesDVTb
 travellers nontravellers
48 h
0.7%
0.2%
 Jacobsen 2003 [25]Passengers flying London– Johannesburg, recruited at check-in, 180 business and 719 economy classQuestionnaire, d-dimers, thrombophilia screen and optional US. Follow up after 6 months>8 hIncomplete follow up in 52%, time window unclearDVTbTime window unclear
0%
Outcome: symptomatic DVT/PE
 Kuipers 2005 [33]8 755 frequently travelling employees of international companies and organizationsQuestionnaire and confirmation through medical chart review>4 hOnly healthy, working populationSymptomatic DVT and PE8 weeks 215/million
 Gajic 2005 [27]223 travellers undergoing surgery and 8637 nontravelling controls with the same type of surgeryMedical chart review>4 hMethod of prophylaxis unclear, unclear how complete the FU was, incomparability of travellers and nontravellersDVT
travellers nontravellers
28 days
4.9%
0.15%
Outcome: severe PE
 Kelman 2003 [28]4.8 million Australians and 4.6 million non-Australians arriving after intercontinental flights in Western AustraliaRecords of all patients admitted with DVT or PE in Western Australia in 1981–1999 internationalFatal or ambulatory treated cases were missedHospitalized for VT
 Australians
 Visitors
28 days
9.6/million
43.5/million
 Lapostolle 2001 [29]All 134.29 million passengers arriving at Charles de Gaulle Airport 1993–2000Review of all records of patients requiring medical help for PE immediately after arrival at the airport<3 h
3–6 h
6–9 h
9–12 h
>12 h
Very short-time window: only cases that seek help within a few hours after landing were includedSevere PEFew hours
0
0.1/million
0.4/million
2.7/million
4.8/million
 Hertzberg 2003 [30]All 6.58 million passengers arriving at Sydney AirportReview of all records of patients admitted with PE at two hospitals>9 hCases who went to other hospitals than the study hospital were missedSevere PEFew hours
2.6/million
 Perez-Rodriguez 2003 [31]All 4.1 million passengers arriving at Madrid Barajas AirportAll patients admitted to one hospital with PE coming directly from the airport<6 h
6–8 h
>8 h
Very short-time window: only cases that seek help within a few hours after landing were includedSevere PEFew hours
0
0.3/million
1.7/million
Outcome: fatal PE
 Parkin 2006 [19]All passengers arriving at New Zealand, 55.8 million residents of New Zealand and 11.2 million overseas visitorsReview of death records and interviews with relatives to identify fatal cases of PE with international air travel within 4 weeks>3 hOnly fatal cases of PE and age limit 15–59Fatal PE residents
 visitors
Few hours
0.6/million
0.5/million
 Kline 2002 [32]All 1.1 million passengers arriving after international flights at Charlotte-Douglas International AirportReview of records of all passengers with cardiac arrest or unstable patients at the airport.InternationalOnly fatal cases that caused severe symptoms immediately after the flight were includedFatal PEImmediately after the flight
0%

Six publications concerned studies in which passengers were systematically screened for the presence of asymptomatic VT after a long-haul flight [21–26]. In one study, the methods were inadequately described [26] and in another, the follow-up was incomplete. [25]. The risk of mainly asymptomatic thrombosis in air travellers as found by screening ranged from 0 (no events in 160 passengers) to 1.5% (11 events in 744 passengers). Only two studies included a nontravelling control population [22, 23]. In the first study, none of the 160 control persons developed a thrombus and in the second study, 2 out of 1213 (0.2%) nontravelling participants developed deep vein thrombosis.

The absolute risk of symptomatic VT was assessed in a study of approximately 9000 employees of international companies and organizations [33]. A total of 22 events occurred within 8 weeks of flights longer than 8 h, yielding an absolute risk of symptomatic VT of 215 per million travellers (95% CI 133–316 per million) after flights longer than 4 h. This was equivalent to a risk of 1/4600 flights. The risk increased with travel duration, up to 793 (95% CI 198–1784) per million travellers after flights longer than 16 h. The results of this study may not be generalizable to all travellers, since the study was conducted in a healthy, working population.

One retrospective follow-up study [39] assessed the frequency of deep vein thrombosis in high risk surgical patients that had to travel long distances prior to their operation and found a risk of 4.9% (95% CI 2.1–7.8) within 2 weeks of the operation, as compared with 0.2% (95% CI 0.1–0.2%) in patients undergoing the same types of high-risk surgery without prior air travel. In this study, the use of prophylaxis for thrombosis was unclear and travelling and nontravelling study participants came from different source populations, because the nontravellers were all US citizens, whereas all travellers were non-US citizens visiting the country for the surgical procedure.

Three studies assessed the risk of pulmonary embolism, requiring medical care immediately after a long-distance air travel [29–31]. Two studies found a dose-response relationship between the frequency of pulmonary embolism and duration of travel [29, 31]. In one study, the risk ranged from no events in 74.2 million flights shorter than 3 h to 13 in 2.7 million flights (4.8 per million, 95% CI 2.2–7.4 per million) in flights longer than 12 h [29]. In a similar Spanish study, no PE was seen after 28.0 million flights shorter than 6 h and in 9.1 million flights longer than 8 h, 15 cases of severe pulmonary embolism occurred (absolute risk 1.7 per million, 95% CI 0.8–2.5) [31]. In another study using a similar design, the risk of PE immediately after a flight longer than 9 h was 2.6 per million (95% CI 1.4–3.8 per million) [40]. In all three studies, the time window in which a traveller could become a case was extremely small, since only persons who developed severe symptoms immediately after arrival were included.

One study assessed the risk of hospital admission for pulmonary embolism within 2 weeks of international flights to Australia, which was found to be 9.6 per million (95% CI 7.0–12.6 per million) for 4.9 million passengers who were residents of Australia and 43.5 per million (95% CI 37.5–49.8 per million) for 4.6 million passengers who were visiting Australia [28]. In this study, travellers who died before reaching the hospital or patients who were treated ambulatory were not included, which may partly explain the difference between residents and nonresidents of Australia.

The risk of fatal pulmonary embolism after air travel was assessed in two studies [19, 32]. One study found no fatal PE’s in 1.1 million passengers arriving after international flights to Charlotte-Douglas Airport, Charlotte, NC in the USA, whereas another study found 11 cases of fatal PE within 4 weeks of 19.3 million flights longer than 3 h, yielding an absolute risk of 0.6 per million passengers (95% CI 0.2–0.9 per million). In both studies, patients who were not sent to the study hospital were missed.

Randomized controlled trials – estimate of the effect of interventions

A total of 11 randomized trials were conducted to assess the effect of various prophylactic measures on the risk of VT after air travel [26, 41–50]. The main results of these trials are shown in Table 3. All studies had a similar design: a number of air travellers, varying from 148 to 833, making long-haul flights (>7 h) were randomized to either a control group or an intervention group that received elastic compression stockings, aspirin, heparin, venoruton (hydroxyethylrutosides), pycnogenol (pine tree extract containing procyanidins, bioflavonoids and organic acids) or FLITE tabs (containing pycnogenol and nattokinase, a soybean extract). All passengers were routinely screened for VT by ultrasound after their flight (within a maximum of 48 h). All but one of these studies was conducted by the same research group. In these publications, the methods of the study were inadequately described or even contradictory. Most striking was that inclusion and exclusion criteria were frequently overlapping. Furthermore, the method of recruitment of participants and whether study participants and outcome assessors were blinded for the treatment group was unclear. The majority of the thrombotic events in all trials were asymptomatic, which may partly explain the high prevalence in the control population. The number of symptomatic events was not clearly described, but is likely to be much lower. These drawbacks, as well as a report from the Medical Research Council’s Fitness to Practice Panel, judging it proved that these papers named co-authors who had not approved the papers [51], hamper the credibility of these trials. We therefore will not discuss the results of these trials in this systematic review. In the only remaining trial [50], the effect of elastic compression stockings was assessed in 231 airline passengers travelling at least 8 h. None of the 100 passengers who were randomized to the elastic compression stockings group developed VT, whereas 12 of the 100 control passengers did, yielding a relative risk of 0.04 (95% CI 0–0.6). However, four passengers wearing elastic compression stockings developed superficial thrombophlebitis, whereas none of the control passengers did.

Table 3.   Randomized controlled trials
First author/ year of publication [reference]ParticipantsInterventionExaminationsLimitationsOutcomes Frequency outcomes (%)a Relative riskb (95% CI)
  1. DVT, deep vein thrombosis; VT, venous thrombosis; US, ultra sonography.

  2. In all studies, most DVTs were asymptomatic.

  3. aNumber of passengers with the outcome of interest (%).

  4. bRelative risk of the intervention group as compared with the control passengers.

  5. cIn these studies, all by the same research group, only asymptomatic events were assessed, the method of selection of participants were unclear and inclusion and exclusion criteria were frequently overlapping. Furthermore, the credibility of the authors of these trials was seriously questioned by the Medical research Council’s Fitness to Practice Panel.

Scurr 2001 [50]200 unselected travellers flying making at least 2 flights >8 h100 stockings
100 no intervention
US all passengers <48 h of the return flightOnly asymptomatic thrombosesDVT
No intervention
Stockings
12 (12)
 0 (0)
0.04 (0–0.6)
Belcaro 2001 [26]833 passengers at increased risk for VT making one flight 10–15 h422 no intervention
411 stockings
US all passengers <24 hcDVT
No intervention
Stockings
19 (4.5)
 1 (0.2)
0.05 (0–0.4)
Belcaro 2002 [41]629 travellers at low risk making 2 flight 7–12 h314 no intervention
315 stockings
US all travellers before and after flightcDVT
No intervention
Stockings
 7 (2.2)
 0 (0)
0.07 (0–1.2)
Cesarone 2002 [45]249 passengers at increased risk making one flight 7–8 h83 No Intervention
84 Aspirin 400 mg 3 days
82 LMWH therapeutic dose once preflight
US all travellers within a few hours of the flightcDVT
No intervention
Aspirin
Heparin
 4 (4.8)
 3 (3.6)
 0 (0)
0.7 (0.2–3.4)
0.1 (0.01–2.0)
Belcaro 2003 [42]151 Passengers with varicose veins making 1one flight 8 h73 No intervention
78 Venoruton
US all travellers within a few hours of the flightcDVT
No intervention
Venoruton
 0 (0)
 0 (0)
 
Belcaro 2003 [43]205 passengers at increased risk making one flight 11.5–12 h102 No intervention
103 Stockings
US all travellers <90 min of the flightcDVT
No intervention
Stockings
 6 (6)
 0 (0)
0.07 (0–1.3)
Cesarone 2003 [48]341 passengers at low-medium risk making one flight 7–12 h169 no intervention
172 Stockings
US all travellers within a few hours of the flightcDVT
No intervention
Stockings
0 (0)
0 (0)
Cesarone 2003 [46]148 passengers with varicose veins making one flight 7–8 h79 no intervention
69 Venoruton 3 days
US all travellers within a few hours after the flightcDVT
No intervention
Venoruton
0 (0)
0 (0)
Cesarone 2003 [49]274 passengers at low-medium risk making one flight 7–12 h138 No intervention
136 Stockings
US all travellers within a few hours of the flightcDVT
No intervention
Stockings
2 (1.4)
0
0.2 (0.01–4.2)
Cesarone 2003 [47]186 passengers at increased risk making one flight 7–8 h92 No intervention
94 FLITE tabs
US all travellers within a few hours of the flightcDVT
No intervention
FLITE tabs
5 (5.4)
0 (0)
0.08 (0–1.5)
Belcaro 2004 [44]198 passengers at increased risk making 1 flight 8 h97 No intervention
101 Pycnogenol
US all travellers <2 h of the flightcDVT
No intervention
Pycnogenol
1 (1)
0 (0)
0.3 (0.01–7.9)

Mechanism of travel-related thrombosis

There are several explanations fore the increased risk of VT after air travel. Apart from immobilization, flight specific factors, such as hypobaric hypoxia may affect the coagulation system. Various investigators have examined the effect of air travel, or one of its specific aspects (e.g. immobilization and hypobaric hypoxia) on thrombin generation and fibrinolysis.

The studies differed much in participant characteristics, duration of exposure, type of exposure and statistical analyses. Most studies determined changes in various parameters before and after specific exposures in volunteers. Table 4 summarizes the relevant aspects of the studies and the direction of the changes in the most commonly used coagulations parameters during the different exposures.

Table 4.   Pathophysiological studies
MechanismFirst author/year of publication [ref]Volunteers (number of women)MethodsMarkers of thrombin generationaMarkers of fibrinolysisa
TATF1 + 2D-dimerPAItPA
  1. VT, venous thrombosis; FVL, Factor V Leiden mutation; OC, oral contraceptive/hormone use; TAT, thrombin-antithrombin.

  2. aThis table roughly indicates the changes in the most commonly used markers of coagulation activation and fibrinolyis during the several exposures. When studies took blood from both arm and leg veins results from arm veins are shown.

ImmobilizationTardy I 1996 [54]31 (28) elderly with varicose veins
 9 (7) nontravelling controls
8-h bus trip, freedom of walking during bus tripTravellers versus controls: no differenceTravellers versus controls no differenceTravellers versus controls no difference  
Tardy II 1 996 [54]23 (20) elderly with varicose veins16-h bus trip, freedom of walking during bus tripAfter versus before: increaseAfter versus before: no changeAfter versus before: no changeAfter versus before: no changeAfter versus before: no change
Stricker 2003 [56]40 (20) healthy6-h immobilization After versus before: decreaseAfter versus before: no change  
Schobersberger 2004 [57]19 (11) healthyReturn bus trip Innsbruck to Rome, 10 h per trip, two nights stop over in RomeAfter versus before: no changeAfter versus before: increaseAfter versus before: no changeAfter versus before: decreaseAfter versus before: decrease
Ansari 2006 [58]10 (0) healthy8-h immobilizationAfter versus before: no changeAfter versus before: no changeAfter versus before: no changeAfter versus before: decreaseAfter versus before: no change
Stricker 2006 [60]20 (9) healthy6-h immobilization After versus before: decreaseAfter versus before: no change  
HypoxiaBendz 2000 [59]20 (0) healthy8-h hypobaric hypoxia
 ∼2400 m
After versus before: increaseAfter versus before: increaseAfter versus before: no change  
Crosby 2003 [61]8 (not stated)  healthyCrossover study
 8-h isocapnic hypoxia
 ∼3600 m and 8-h normobaric normoxia
Hypoxia versus control: no difference Hypoxia versus control: no difference Hypoxia versus control: no difference   
Hodkinson 2003 [62]6 (0) healthyCrossover study
 3-h normobaric hypoxia
 ∼3660 m and 3-h normobaric normoxia
 Hypoxia versus control: no difference  Hypoxia versus control: no difference  
Schobersberger 2006 [63]12 (3) healthy10-h normobaric hypoxia
∼2400 m
After versus before: no changeAfter versus before: no changeAfter versus before: no changeAfter versus before: no changeAfter versus before: decrease
Toff 2006 [64]73 (42):49 (22) without risk factors (OC or age >50) and 24 (20) with risk factors Crossover study
8-h hypobaric hypoxia
∼ 2438 m and 8-h normobaric normoxia
Hypoxia versus control: no difference Hypoxia versus control: no difference Hypoxia versus control: no difference Hypoxia versus control: no difference Hypoxia versus control: no difference
Air travelSchobersberger 2002 [65]20 volunteers:
10 (5) healthy
10 (6) with risk factors
(>40 yrs, OC, fobesity or venous insufficiency)
Return flight Innsbruck to Washington with stop over of 2 h in Vienna
Flight time: 8 h 20 min one way (Vienna–Washington)
 two night stays in Washington
After versus before: no change After versus before: no changeAfter versus before: increaseAfter versus before: decrease
Boccalon 2005 [66]30 (0) healthy men11-h flightAfter versus before: decreaseAfter versus before: decreaseAfter versus before: no change After versus before: no change
Schreijer 2006 [67]71 volunteers:
30 (15) without risk factors
41 (41) with risk factors
(OC, FVL or both)
Crossover study
 8-h flight compared with 8-h movie marathon and 8 h of daily activities
All three parameters increased in more participants during the flight than during the immobilized or ambulant situation    

Parameters of thrombin generation and fibrinolysis

Thrombin generation and fibrinolysis are regulated by its activators and inhibitors (Figs 2 and 3). The amount of thrombin generated is often assessed by levels of activation peptide prothrombin fragment 1 + 2 (F1 + 2), and its inhibitor complex thrombin-antithrombin (TAT). Activation of the fibrinolytic system is reflected by the increased levels of D-dimers (fibrin degradation products), which by definition also points to thrombin generation and hence may be its simple consequence. [52].

Figure 2.

 Thrombin generation. Dashed lines reflect inhibition, solid lines reflect activation, and substances in italic are products of the process.

Figure 3.

 Fibrinolysis. Dashed lines reflect inhibition, solid lines reflect activation.

Increased levels of activated coagulation factors (such as FVIIa and FVIIIa) are also known to increase thrombin generation [52, 53].

Immobilization

The most obvious explanation for air-travel-related thrombosis is immobilization. Passengers are restricted to limited space, resulting in a cramped position during long-haul flights and are even more immobilized when they are asleep.

Several studies investigated the effect of prolonged immobilization on thrombin generation, but conflicting results were shown [54–67] (Table 4). Stricker et al. found a decrease in markers of thrombin generation during 6 h of immobilization in 40 volunteers, whereas they found no change in 18 participants during the ambulant situation [56]. In a subsequent study with a similar design, the investigators found similar results in 20 volunteers and also found evidence for down-regulation of the protein C system, one of the inhibitors of the coagulation system [60]. Others found evidence for thrombin activation during a 10-h bus journey in a group of 19 healthy volunteers [57]. In a similar study, but with 23 elderly with varicose veins, TAT increased after a 16-h bus trip (especially in two high responders), indicating thrombin generation. However, F1 + 2 remained unchanged after the bus trip and FVIIa decreased [54]. The same study group found no difference in a comparable study in markers of thrombin generation between 31 bus travellers and a nontravelling control group (n = 9). More recently, Ansari et al. found no change in markers of thrombin generation in 10 healthy volunteers after 8 h of prolonged sitting, although F1 + 2 levels showed a tendency to decrease [58]. In an artificial model of immobilization, with 30 min of blood stasis provoked by a pressure cuff around the thigh, F1 + 2 decreased [68].

The effect of immobilization on the fibrinolytic system was also contradictory (Table 4). In some studies, markers of fibrin generation and fibrinolysis remained unchanged [54, 56, 60, 68]. Others found a decrease in levels of tissue plasminogen activator (tPA) (an activator of the fibrinolytic system) and plasminogen activator inhibitor (PAI) (an inhibitor of the fibrinolytic system). D-dimers remained unchanged in this study, with the exception of two subjects in whom D-dimers increased (>0.5 mg l−1) [57]. In a more recent study, PAI decreased after 8 h of immobilization, whereas D-dimer and tPA levels did not clearly differ from baseline, although both parameters had a tendency to decrease [58].

In the majority of the studies, no control population was used to assess whether any observed effect was the result of immobilization or other factors, such as circadian rhythm. Only in one study, coagulation parameters in passengers after an 8-h bus trip were compared with those in a nontravelling control population [54]. No differences in TAT, D-dimers or F1 + 2 were observed between the two groups.

Hypoxia during simulated air travel

During air travel, cabin pressure drops to 75.8 kPa, which is equivalent to an altitude of 2400 m above sea level. Consequently, oxygen saturation can drop as low as 90–93% and even to 80% in passengers who are asleep [69, 70]. To separate the effect of sole hypoxia from hypobaric hypoxia, both the effects of hypobaric as well as normobaric hypoxia on human coagulation have been investigated.

Already in 1976, Maher et al. studied the effect of acute hypobaric hypoxia (corresponding to an altitude of 4400 m above sea level) on human coagulation [71]. He found a shortening of the partial thromboplastin time (aPTT) after 1 and 24 h in a hypobaric chamber whereas fibrinogen and FVIII levels returned to baseline at 24 h after an abrupt decrease after 1 h. Fibrin degeneration products (e.g. D-dimers) were transiently increased in three subjects.

Years later, when more laboratory assays became available, the effect of hypoxia on both thrombin generation and fibrinolysis was further investigated (Table 4). Bendz et al. exposed 20 volunteers to 8 h of hypobaric hypoxia (76 kPa) by natural elevation and found an increase in thrombin generation (reflected by TAT and F1 + 2), with a maximum increase after 2 h of hypobaric hypoxia [59]. These changes were accompanied by an increase in FVIIa (measured as FVIIa-tissue factor complex), whilst FVII antigen and tissue factor pathway inhibitor (TFPI) (antigen and activity) decreased [59]. However, in a controlled experiment with eight participants, Crosby et al. found no evidence for thrombin generation during exposure to 8 h of isocapnic hypoxia compared with 8 h of normobaric normoxia [61]. Also short-term exposure to normobaric hypoxia did not seem to affect markers of thrombin generation in a crossover study [62]. In a much larger study, Toff and colleagues exposed 73 volunteers alternately to hypobaric hypoxia and normobaric normoxia. They found no difference between the changes in markers of thrombin generation during hypobaric or normobaric exposures [64]. These findings were confirmed by Schobersberger and colleagues [63].

Markers of fibrinolysis mainly remained unchanged during hypoxia in most studies [59, 61, 62, 64], although Schobersberger found a decrease in t-PA after 10 h of hypoxia [63].

Air travel

Only few studies have investigated the effect of actual air travel on the coagulation system (Table 4). No evidence for increased thrombin generation was found in a study with 20 volunteers (including 10 volunteers who were obese, aged >40 years, used oral contraceptives or had venous insufficiency) after a return flight from Vienna to Washington [65]. In a similar study, Boccalon found a reduction in thrombin generation after an 11-h return flight with 30 healthy male volunteers [66]. In a tightly controlled crossover study with 71 volunteers (including 41 women with risk factors for VT, i.e. factor V Leiden mutation, oral contraceptive use or both) evidence was found for thrombin generation in 17% of individuals during an 8-h flight, whereas this was found in only 3% during an 8-h movie marathon and in 1% during the ambulant situation. The effect was most evident in women with Factor V Leiden mutation (FVL) who use oral contraceptives [67].

Schobersberger found evidence for suppressed fibrinolysis during air travel [65], whereas markers of fibrinolysis remained unchanged in the study by Boccalon et al. [66]. In the crossover study, the fibrinolytic system was activated in more participants during the flight then during the immobilized or ambulant situation [67].

Drawbacks of the studies

Thus, the results in all three settings (immobilization, hypoxia and air travel) were conflicting. There are several possible explanations for these discrepancies. First, since it is plausible that only some individuals are susceptible to coagulation activation during air travel, people with risk factors for VT may react differently than those without. To control for the large inter-individual normal range of coagulation parameters, data are best analysed on an individual level in a crossover design. Most studies that have been conducted so far included few participants (mostly without risk factors for VT) and presented data on group level only. Secondly, when only pre- and postexposure data are compared, effects of circadian rhythm are not accounted for. Only four studies compared changes in coagulation parameters during air travel or hypobaric hypoxia to a control situation and may be considered to have yielded valid data [61, 62, 64, 67].

Discussion

Long-distance travel increases the risk of VT 2- to 4- fold. Only one study assessed the risk of symptomatic VT in a frequently travelling working population and found a risk of 1/4600 travellers within 4 weeks of flights longer than 4 h. The risk of pulmonary embolism occurring immediately after air travel increases with duration of travel, from 0 in flights shorter than 3 h up to 4.8 per million in flights longer than 12 h. The risk of fatal PE immediately after arrival, which was assessed in two studies, is estimated at <0.6 per million passengers in flights longer than 3 h. The risk of VT is not increased after travel shorter than 3–4 h. The risk of asymptomatic VT after long-haul flights ranged up to 12%.

In one randomized controlled trial comparing the prevalence of asymptomatic VT in control travellers to that in travellers assigned to wear stockings, stocking effectively prevented the development of asymptomatic thrombi. The effect of stockings, low molecular weight heparin and aspirin has been studied in a series of trials by a single research group. These studies had serious methodological flaws and the scientific integrity of the authors was questioned by the Medical Research Council’s Fitness to Practice Panel [51].

Although several studies have addressed the mechanism responsible for the association between VT and long-distance travel, differences in design, analysis and interventions do not allow us to draw solid conclusions. One controlled study that included volunteers with risk factors for VT showed evidence for an additional mechanism to immobilization that could lead to coagulation activation after air travel, especially in susceptible individuals.

Future research should focus on the mechanism responsible for coagulation activation because of (air) travel, identification of individuals at high risk, and prophylactic measures that prevent symptomatic VT and outweigh its potential harms. Based on the currently available evidence, we conclude that the absolute risk of symptomatic VT in the general travelling population is not high enough to justify the widespread use of prophylaxis, in particular of prophylaxis that may cause side effects. There may be a rationale for preventive measures in individuals at high risk, but it is currently not known which prophylactic measures have a positive balance of effect and risk.

Conflict of interest statement

No conflict of interest was declared.

Ancillary