Compression stockings for preventing deep vein thrombosis in airline passengers (Review)

Fifty of 2637 participants with follow-up data available in the trials of wearing compression stockings on both legs had a symptomless DVT; three wore stockings, 47 did not (odds ratio (OR) 0.10, 95% confidence interval (CI) 0.04 to 0.25, P < 0.001; high-certainty evidence). There were no symptomless DVTs in three trials. Sixteen of 1804 people developed superficial vein thrombosis, four wore stockings, 12 did not (OR 0.45, 95% CI 0.18 to 1.13, P = 0.09; moderate-certainty evidence). No deaths, pulmonary emboli or symptomatic DVTs were reported. Wearing stockings had a significant impact in reducing oedema (mean difference (MD) −4.72, 95% CI −4.91 to −4.52; based on six trials; low-certainty evidence). A further three trials showed reduced oedema in the stockings group but could not be included in the meta-analysis as they used different methods to measure oedema. No significant adverse effects were reported. girth; blood samples (markers of coagulation and fibrinolysis: tissue factor, tissue factor pathway inhibitor, thrombin-antithrombin complex, and D-Dimer).


A B S T R A C T Background
Air travel might increase the risk of deep vein thrombosis (DVT). It has been suggested that wearing compression stockings might reduce this risk. This is an update of the review first published in 2006.

Objectives
To assess the e ects of wearing compression stockings versus not wearing them for preventing DVT in people travelling on flights lasting at least four hours.

Search methods
The Cochrane Vascular Information Specialist searched the Cochrane Vascular Specialised Register, CENTRAL, MEDLINE, Embase, CINAHL and AMED databases and World Health Organization International Clinical Trials Registry Platform and ClinicalTrials.gov trials registers to 1 April 2020. We also checked the bibliographies of relevant studies and reviews identified by the search to check for any additional trials.

Selection criteria
Randomised trials of compression stockings versus no stockings in passengers on flights lasting at least four hours. Trials in which passengers wore a stocking on one leg but not the other, or those comparing stockings and another intervention were also eligible.

Data collection and analysis
Two review authors independently selected trials for inclusion and extracted data. We sought additional information from trialists where necessary.

Main results
One new study that fulfilled the inclusion criteria was identified for this update. Twelve randomised trials (n = 2918) were included in this review: ten (n = 2833) compared wearing graduated compression stockings on both legs versus not wearing them; one trial (n = 50) compared wearing graduated compression tights versus not wearing them; and one trial (n = 35) compared wearing a graduated compression stocking on one leg for the outbound flight and on the other leg on the return flight. Eight trials included people judged to be at low or medium risk of developing DVT (n = 1598) and two included high-risk participants (n = 1273). All flights had a duration of more than five hours.

Library
Trusted evidence. Informed decisions. Better health.

Cochrane Database of Systematic Reviews
Fi y of 2637 participants with follow-up data available in the trials of wearing compression stockings on both legs had a symptomless DVT; three wore stockings, 47 did not (odds ratio (OR) 0.10, 95% confidence interval (CI) 0.04 to 0.25, P < 0.001; high-certainty evidence). There were no symptomless DVTs in three trials. Sixteen of 1804 people developed superficial vein thrombosis, four wore stockings, 12 did not (OR 0.45, 95% CI 0.18 to 1.13, P = 0.09; moderate-certainty evidence). No deaths, pulmonary emboli or symptomatic DVTs were reported. Wearing stockings had a significant impact in reducing oedema (mean di erence (MD) −4.72, 95% CI −4.91 to −4.52; based on six trials; lowcertainty evidence). A further three trials showed reduced oedema in the stockings group but could not be included in the meta-analysis as they used di erent methods to measure oedema. No significant adverse e ects were reported.

Authors' conclusions
There is high-certainty evidence that airline passengers similar to those in this review can expect a substantial reduction in the incidence of symptomless DVT and low-certainty evidence that leg oedema is reduced if they wear compression stockings. The certainty of the evidence was limited by the way that oedema was measured. There is moderate-certainty evidence that superficial vein thrombosis may be reduced if passengers wear compression stockings. We cannot assess the e ect of wearing stockings on death, pulmonary embolism or symptomatic DVT because no such events occurred in these trials. Randomised trials to assess these outcomes would need to include a very large number of people.

Background
In the last few years, there has been increasing interest in whether compression stockings (or 'flight socks') reduce the risk of deep vein thrombosis (DVT; blood clots in the legs) and other circulatory problems in airline passengers. The stockings are worn throughout the flight and are similar to those known to be e ective in patients lying in bed a er an operation. By applying a gentle pressure, to the ankle in particular, compression stockings help blood to flow. Pressure combined with leg movement helps blood in superficial (surface) veins to move to the deep veins and back to the heart. The blood is then less likely to clot in the deep veins, which could be fatal if the clot moves to the lungs.

Study characteristics and key results
This review included 12 trials (2918 participants) and we were able to combine the data from nine trials with a total of 2637 participants (current to April 2020). Almost half of the participants were randomly assigned to wearing stockings for a flight lasting at least five hours while the other half did not wear stockings.
None of the passengers developed a DVT with symptoms (slowly developing leg pain, swelling and increased temperature) and no serious events (a blood clot in their lungs (pulmonary embolism) or dying) were reported. Passengers were carefully assessed a er the flight to detect any problems with the circulation of blood in their legs, even if they had not noticed any problems themselves. Wearing compression stockings resulted in a large reduction in symptomless DVT among airline passengers who were allocated to wear compression stockings compared to those allocated not to wear compression stockings. This di erence in symptomless DVT between the two groups is equivalent to a reduction in the risk from a few tens per thousand passengers to two or three per thousand. People who wore stockings had less swelling in their legs (oedema) than those who did not wear them. Fewer passengers developed superficial vein thrombosis when wearing compression stockings than those not wearing stockings. Not all the trials reported on possible problems with wearing stockings but in those that did, the researchers said that the stockings were well-tolerated, without any problems.

Reliability of the evidence
High-certainty evidence shows that airline passengers wearing compression stockings develop less symptomless DVT and low-certainty evidence shows that leg swelling is reduced when compared to not wearing compression stockings. Our certainty in the evidence was limited by the way that swelling was measured. There is moderate-certainty evidence that superficial vein thrombosis may be reduced in passengers who wear compression stockings. We cannot assess the e ect of wearing stockings on death, pulmonary embolism or symptomatic DVT because no such events occurred in these trials. Randomised trials to assess these outcomes would need to include a very large number of people.

Oedema
Follow-up period immediately post flight Post flight values measured on a scale from 0 (no oedema) to 10 (maximum oedema) The mean oedema score ranged across control groups from 6 to 9 The mean oedema score in the intervention groups was on average 4.7 lower (4.9 lower to 4.5 lower) Not estimable None of the trials reported adverse effects, apart from 4 cases of superficial vein thrombosis in varicose veins in the knee region that were compressed by the upper edge of the stocking in one trial. However, the metaanalysis of the data on this outcome from this trial and 7 others found a non-statistically significant difference (see above) *The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI Kuipers and colleagues also reported that the "absolute risk of a symptomatic event within 4 weeks of flights longer than 4 h is 1/4600 flights" and "the risk of severe pulmonary embolism (PE) occurring immediately a er air travel increases with duration of travel, up to 4.8 per million in flights longer than 12 h" (Kuipers 2007).
A number of factors have been suggested as possible causes of any increase in the risk of developing DVT when flying (Adi 2003). These can be classified as: • travel-related: for example, prolonged immobilisation in narrow economy class seats with limited leg room, insu icient fluid intake, dehydration and low humidity (

Description of the intervention
It has been suggested that the use of compression stockings during long-haul flights may help to reduce the risk of developing DVT. It has also been suggested that standing up or walking around occasionally in flight, drinking plenty of water and performing leg-stretching exercises may also help to reduce a person's risk (Geroulakos 2001

How the intervention might work
Compression stockings are thought to reduce the risk of DVT by exerting graduated pressure on the leg, with the pressure being greatest at the ankle. This, when combined with muscular activity in the limb, is thought to displace blood from the superficial venous system to the deep venous system. This, in turn, reduces blood stasis that can lead to clotting and increases the velocity and volume of blood flow in the deep venous system, thereby potentially preventing thrombosis (Sachdeva 2014).

Why it is important to do this review
There has been increased research interest in the issue of DVT in airline passengers in recent years. For example, as well as the review by Adi 2003 and Adi 2004 and another by Ansari 2005, the World Health Organization announced the launch of a research programme to investigate the relationship between air travel and venous thrombosis in May 2002 (WHO 2002). The findings of this report show that the increased risk of VTE observed in passengers on long-haul flights is due to extended periods of immobility. As the number of people taking long-haul flights is increasing, and as these passengers will have known or unknown thrombosis risk factors, they concluded that "air travel-related VTE is an important public health issue" (WRIGHT project 2007).

O B J E C T I V E S
To assess the e ects of wearing compression stockings versus not wearing them for preventing DVT in people travelling on flights lasting at least four hours.

Types of studies
Randomised trials that compare people wearing compression stockings versus people not wearing them during flights lasting at least four hours. Randomised trials in which people wore a stocking on one leg and not the other are included for any leg-specific outcomes but are not eligible for any meta-analyses of outcomes (such as pulmonary embolism (PE) and death). This is because it would not be possible to know which leg had contributed to the particular outcome.
In future updates, we will only include studies which measure DVT (symptomatic or symptomless); studies which do not measure DVT will be excluded.

Library
Trusted evidence. Informed decisions. Better health.

Types of participants
Any passenger on a flight of more than four hours' continuous duration. This includes people of both sexes, all ages, all risk factors, irrespective of any other interventions they may have used for the prevention of deep vein thrombosis (DVT). Flights in any direction are eligible, as are both daytime and night-time flights.

Types of interventions
The primary analyses are of unconfounded randomised trials in which the only di erence between the groups is the allocation to wear, or not wear, compression stockings. This includes trials in which no other form of prevention was used, and trials in which other forms of prevention were available equally to both groups. If confounded randomised trials, in which some people were allocated to wear stockings and other people were allocated to an alternative form of prevention, had been found (none were), subsidiary analyses would have been performed for these.

Primary outcomes
• Diagnosis of symptomatic or symptomless DVT (by ultrasound, venogram or isotope)

Secondary outcomes
• Diagnosis of pulmonary embolism (by ventilation perfusion lung scan, pulmonary angiogram, spiral computed tomography (CT) scanning, or postmortem examination) • Death • Superficial vein thrombosis • Oedema • Adverse e ects arising from the use of compression stockings

Electronic searches
The Cochrane Vascular Information Specialist conducted systematic searches of the following databases for randomised controlled trials and controlled clinical trials without language, publication year or publication status restrictions: The Information Specialist modelled search strategies for other databases on the search strategy designed for CENTRAL. Where appropriate, they were combined with adaptations of the highly sensitive search strategy designed by the Cochrane Collaboration for identifying randomised controlled trials and controlled clinical trials (as described in the Cochrane Handbook for Systematic Reviews of Interventions Chapter 6, Lefebvre 2011). Search strategies for major databases are provided in Appendix 1.

Searching other resources
We checked the bibliographies of relevant studies and reviews identified by the search to check for any additional trials. We also checked the website of the World Health Organization (WHO) (www.who.int/), for information on the ongoing WHO Research into Global Hazards of Travel project (WRIGHT project 2007).

Selection of studies
For this update, two review authors (CB and MC) screened the titles and abstracts of all retrieved records to identify obvious exclusions (i.e. records that were found by our electronic searches but were clearly irrelevant to this review). We obtained full copies for reports that might relate to eligible studies and two review authors (CB and MC) assessed these to determine if they met the inclusion criteria for the review. CB and MC resolved any disagreements through discussion.

Data extraction and management
Using a pre-specified data extraction form, MC performed data extraction and CB confirmed this independently. Information extracted included: • descriptive data for the people in the trial (including age, sex and all reported risk factors); • details of the interventions (including type of compression stocking; the duration, direction and time of day of the flight; and any additional interventions for the prevention of deep vein thrombosis or confounding interventions); • outcomes, as described in Types of outcome measures above.

Assessment of risk of bias in included studies
Two review authors (MC, CB) independently evaluated the eligible study for quality, using Cochrane's tool for assessing risk of bias (Higgins 2011). This tool provides judgements made on six domains, which include randomisation sequence generation, allocation concealment methods, blinding (participants, personnel and outcome assessors), incomplete outcome data, selective outcome reporting, and any other relevant biases. The two investigators performed evaluations of low risk, unclear risk, or high risk for each domain for every included study and resolved any disagreements through discussion.

Measures of treatment e ect
For dichotomous outcomes we calculated odds ratios (ORs) with 95% confidence intervals (CIs). For continuous data, we planned a meta-analysis using mean di erences (MD) with standard deviations (SDs) and 95% CIs.

Library
Trusted evidence. Informed decisions. Better health.

Unit of analysis issues
The unit of analysis for this review was the individual participant. One trial randomly allocated passengers to wear a stocking on one leg during an outward flight and on the other leg on the return journey. Only leg-specific outcome data were used in this case.

Dealing with missing data
We based analysis on an intention-to-treat basis and therefore all randomised participants of interest from the included studies were to be included in the analysis. Where data were missing, we investigated if these were described within the reports and if evenly distributed between the groups.

Assessment of heterogeneity
Statistical heterogeneity was assessed both visually within the forest plots and by using the I statistic as described in Chapter 9 of Higgins 2011. Heterogeneity was taken to be substantial with an I value of 50% to 75% and considerable with an I value of 75% or more.

Assessment of reporting biases
If su icient trials had been identified we intended to assess reporting bias using a funnel plot. As fewer than 10 studies reported on any one outcome this was not possible for this 2021 update.

Data synthesis
The decision about whether or not to combine the results of individual studies depended on an assessment of heterogeneity. The studies were assessed for homogeneity of study design and when they were judged to be su iciently homogeneous in their design, a meta-analysis was carried out and the statistical heterogeneity was assessed. The preferred statistical analysis was the OR and the fixed-e ect model, using the statistical so ware in Cochrane's Review Manager so ware (Review Manager 2020), but other analyses were considered in the light of the very low event rates in the included studies. If considerable heterogeneity was detected we planned to use the random-e ects model.
As noted above, unconfounded trials (i.e. allocation to wearing stockings versus not wearing them with no other di erences between the groups) would have been analysed separately from confounded trials (i.e. allocation to wearing stockings versus another intervention), if any of the latter had been found (none were found).

Subgroup analysis and investigation of heterogeneity
If su icient data had been available, subgroup analyses would have been performed separating participants into di erent groups on the basis of risk of DVT and into di erent time periods following the flight, with the main focus of the analyses expected to be the period within one week of the flight. However, no data were available from the trials for outcomes measured beyond this period of follow-up.

Sensitivity analysis
We carried out sensitivity analysis using both RevMan and Stata, to determine if the results of the meta-analysis were robust due to the small or low overall event rate in some of the included trials.

Summary of findings and assessment of the certainty of the evidence
A table summarising the best evidence of relevant outcomes was constructed for comparison of compression stockings versus not wearing compression stockings. Study populations consisting of passengers at low or medium risk and high risk of developing DVTs were considered. The most important and clinically relevant outcomes (both desirable and undesirable) that were thought to be essential for decision-making were selected for the Summary of findings 1. These are described in the Types of outcome measures and include symptomatic and symptomless DVT; PE; death; superficial vein thrombosis; oedema and adverse e ects from the use of compression stockings. Assumed control intervention risks were calculated by the mean number of events in the control groups of the selected studies for each outcome. The system developed by the GRADE working group was used for grading the quality of evidence as high, moderate, low and very low, based on within-study risk of bias, directness of evidence, heterogeneity, precision of e ects estimates, and risk of population bias (GRADE 2004). We used GRADEpro so ware to create the 'Summary of findings' table (Gradepro GDT).

Library
Trusted evidence. Informed decisions. Better health.
Cochrane Database of Systematic Reviews

Included studies
For this update we included one additional study: Broatch 2019. Therefore in total we identified 12 randomised controlled trials with a combined total of 2918 participants that were eligible for this review ( A total of 2883 participants were randomised in the 11 unconfounded trials identified. Eight of the trials recruited a total of 1598 participants who were judged to be at low or medium risk of a DVT (Hagan 2008; LONFLIT 4 -Kendall 1; LONFLIT 4 -Kendall 2; LONFLIT 4 -Scholl 1; LONFLIT 4 -Scholl 2; LONFLIT 4 -Traveno 1; LONFLIT 4 -Traveno 2; Scurr 2001). These studies excluded passengers with previous episodes of DVT, coagulation disorders, limited mobility due to bone or joint problems, neoplastic disease, varicose veins or participants taller than 190 cm and heavier than 90 kg, those advised to wear graduated compression tights in flight, or on medication for cardiovascular disease, diabetes or hypertension. Broatch 2019 did not specify if the participants in their study were low risk, or describe any exclusion criteria, but participants were 'elite athletes' and the study authors discuss the increased risk of VTE 'as a result of the large volume and high-intensity exercise typically performed'. The other two trials recruited a total of 1273 high-risk participants (LONFLIT 2; LONFLIT 5). LONFLIT 5 excluded participants taller than 190 cm, or weighing more than 90 kg, or having recent/presence of thrombosis, severe bone, joint, or mobility problems, severe hypertension, or clinical disease requiring treatment. The exclusion criteria of LONFLIT 2 were not clear. The LONFLIT trials were reported to have been conducted during 2001 to 2003. The Scurr trial was described as ongoing in a report published in early 2001 (Scurr 2001a); but the actual start and finish dates for recruitment were not reported (Scurr 2001b). Hagan 2008 was conducted between April and October 2006. In all the trials the flight duration was at least five hours and passengers allocated to wear stockings were told to wear these for the duration of the flight. In the LONFLIT 2 trial, participants were advised to put the stockings on 6 to 10 hours before the flight. In the other trials they were advised to put them on within a few hours before the flight.
All unconfounded trials, except Broatch 2019 and Hagan 2008, assessed incidence of symptomless DVT within a few days of the flight but information on the assessment of symptomatic DVT, pulmonary embolism and death is not reported for all of them. In addition to the 11 unconfounded trials, we found one trial (n = 35) in which participants were randomly allocated to wear a stocking on one leg during an outward flight and on the other leg on the return journey. This trial used class II compression stockings, with the flights lasting approximately 14.5 hours. Participants were also randomised to receive dried vine leaves (Antistax) versus diuretics versus no drugs. However, owing to the strong e ect of the diuretics on the outward flight, these were not used on the return journey and results of the e ects of the stocking on the return flight were reported for the nine patients this a ected (Loew 1998).
We also identified two conference abstracts for studies from the LONFLIT group which reported on research involving 420 high-risk participants comparing stockings versus low molecular weight heparin versus control (Belcaro 2002a), and 400 highrisk participants comparing aspirin versus low molecular weight heparin versus low molecular weight heparin and stockings versus control (Belcaro 2002b). However, correspondence with the first author of these abstracts, Gianni Belcaro, in 2005 suggests that the relevant data from these comparisons have been used within our analyses for the LONFLIT trials described above.

Excluded studies
No additional studies were excluded for the 2021 update. One trial was excluded previously as it was not randomised (Iwama 2002).

Risk of bias in included studies
See Figure 2 and Cochrane Database of Systematic Reviews

Random sequence generation (selection bias)
Allocation concealment (selection bias) Blinding of participants and personnel (performance bias): All outcomes Blinding of outcome assessment (detection bias): All outcomes Incomplete outcome data (attrition bias): All outcomes Selective reporting ( Cochrane Database of Systematic Reviews Cochrane Database of Systematic Reviews All 12 trials were described as randomised but only Hagan 2008 provided su icient information to be judged as being at low risk of selection bias (Broatch 2019; Hagan 2008; Loew 1998; LONFLIT 2; LONFLIT 4 -Kendall 1; LONFLIT 4 -Kendall 2; LONFLIT 4 -Scholl 1; LONFLIT 4 -Scholl 2; LONFLIT 4 -Traveno 1; LONFLIT 4 -Traveno 2; LONFLIT 5; Scurr 2001). All 12 trials were at high risk of bias as they did not blind the participants to which group they had been randomised. All trials, except Broatch 2019, reported some losses to follow-up mostly due to poor compliance or flight connection problems.

Allocation
Only Hagan 2008 was judged as being at low risk of bias for random sequence generation and allocation concealment. Scurr 2001 was judged as being at low risk of allocation concealment but did not provide adequate information on the randomisation method used. All the remaining trials provided insu icient information on these domains and were therefore assessed as being at unclear risk of selection bias (Broatch 2019; Loew 1998; LONFLIT 2; LONFLIT 4 -Kendall 1; LONFLIT 4 -Kendall 2; LONFLIT 4 -Scholl 1; LONFLIT 4 -Scholl 2; LONFLIT 4 -Traveno 1; LONFLIT 4 -Traveno 2; LONFLIT 5).

Blinding
Given the intervention was to wear or not wear a compression stocking or stockings it was not possible to blind the passengers. All studies were assessed as being at high risk of performance bias given the subjective nature of some of the outcomes. Blinding of outcome assessment was not described in nine studies. Scurr 2001 described adequate blinding techniques for DVT detection only and so was assessed as being at low risk of detection bias for this outcome, but unclear overall. Hagan 2008 was at high risk of detection bias as outcomes were self-reported. The remaining ten trials did not describe outcome assessment in su icient detail and so were at an unclear risk of detection bias (Broatch 2019; Loew 1998; LONFLIT 2; LONFLIT 4 -Kendall 1; LONFLIT 4 -Kendall 2; LONFLIT 4 -Scholl 1; LONFLIT 4 -Scholl 2; LONFLIT 4 -Traveno 1; LONFLIT 4 -Traveno 2; LONFLIT 5).

Incomplete outcome data
See the table 'Characteristics of included studies' for details on losses to follow-up in each trial. In summary, all trials except Broatch 2019, reported losses to follow-up, largely due to poor compliance or flight connection problems. Outcome data were typically unavailable for less than 10% of participants, with only LONFLIT 4 -Traveno 2 (19 of 165, 12%) and Scurr 2001 (31 of 231, 13%) reporting higher losses than this in trials of wearing stockings on both legs versus neither but numbers were similar between the treatment and control groups. In LONFLIT 2, 52 of 885 (6%) of participants were lost to follow-up but it is not clear if these were evenly distributed between the groups, so LONFLIT 2 was given a high risk of bias judgement.

Selective reporting
Eleven included studies reported on all the expected outcomes and were judged as being at low risk of reporting bias (

Other potential sources of bias
A gender imbalance between the compression stockings and no compression stockings group (70% female versus 53% respectively) was reported in Scurr 2001. It is not clear if this could a ect the outcomes. No other potential sources of bias were identified.

E ects of interventions
See: Summary of findings 1 Compression stockings compared with no compression stockings for people taking long haul flights

Symptomatic deep vein thrombosis
None of the 2821 participants in the nine trials of wearing compression stockings on both legs versus neither were reported to have developed a symptomatic DVT. Broatch 2019 and Hagan 2008 did not report on this outcome.

Symptomless deep vein thrombosis
Of the 2821 participants randomised into the nine trials of wearing compression stockings on both legs versus not wearing them, follow-up data were available for 2637 (LONFLIT 2; LONFLIT 4 -Kendall 1; LONFLIT 4 -Kendall 2; LONFLIT 4 -Scholl 1; LONFLIT 4 -Scholl 2; LONFLIT 4 -Traveno 1; LONFLIT 4 -Traveno 2; LONFLIT 5; Scurr 2001). Among these, 50 people were reported to have developed a symptomless DVT, which was detected by the investigations done within the trials, either using ultrasound or Ddimer testing and fibrinogen tests. Three of these people had been allocated to wear stockings and the remaining 47 people were not wearing stockings.
In three of the nine trials, no symptomless DVTs were found in any of the participants, regardless of whether they wore compression stockings or not (LONFLIT 4 -Kendall 1; LONFLIT 4 -Traveno 1; LONFLIT 4 -Traveno 2). The overall incidence of symptomless DVT was 2.43% in the two trials that recruited high-risk participants (29 among the 1191 participants with follow-up, distributed as follows: three in the compression stockings group and 26 in the no compression stockings group) and 1.45% in the seven trials that recruited people judged to be at low or medium risk (21 among the 1446 participants with follow-up: two in the compression stockings group and 19 in the no compression stockings group).
There was no evidence of statistical heterogeneity among the results of the trials and the combined estimate of the e ect of wearing compression stockings versus not wearing them is an OR of 0.10 (95% CI 0.04 to 0.25, P < 0.001, high-quality evidence). However, because of the very low overall event rate, the fact that some trials had zero events and the fact that some trials had a small number of events in one group and none in the other, we explored the stability of this result depending upon the assumptions made when calculating estimate of e ect.
The default method used to calculate an OR in Review Manager 2020 is the Mantel-Haenszel method, which adds a continuity correction of 0.5 to groups in which there were no events. This may be too high for the present circumstances. Statistical work on continuity corrections in meta-analyses of sparse data has concluded that the "Mantel-Haenszel summary estimates using the alternative continuity correction factors gave the least Cochrane Database of Systematic Reviews biased results for all group size imbalances. Logistic regression was virtually unbiased for all scenarios and gave good coverage properties. The Peto method provided unbiased results for balanced treatment groups but bias increased with the ratio of the study arm sizes. The Bayesian fixed-e ect model provided good coverage for all group size imbalances. The two alternative continuity corrections outperformed the constant correction factor in nearly all situations. The inverse variance method performed consistently badly, irrespective of continuity correction used." (Sweeting 2004).
It is not possible to explore this further using the statistical tools available in Review Manager 2020 so we performed sensitivity analyses using Stata, using the Mantel-Haenszel method with various small continuity corrections, logistic regression and the Peto method for comparison. These special analyses were done using data from an earlier version of our meta-analysis that had slightly fewer events and a smaller number of participants from the LONFLIT 5 trial, and an overall estimate of e ect of 0.07 (0.02 to 0.22). However, the general finding that the choice of analysis technique makes little important di erence to the overall conclusions would still hold. The recalculated ORs using the Mantel-Haenszel method converged to a steady level of 0.04 (95% confidence interval 0.01 to 0.18) as the continuity correction was diminished to zero. This was identical to the one obtained using logistic regression. The result using the Peto method was 0.15 (95% confidence interval 0.09 to 0.28, P < 0.001). Broatch 2019 and Hagan 2008 did not report on this outcome.

Pulmonary embolism
None of the 2821 participants in the nine trials of wearing stockings on both legs versus neither were reported to have developed a pulmonary embolism. Broatch 2019 and Hagan 2008 did not report on this outcome.

Death
None of the 2821 participants in the nine trials of wearing stockings on both legs versus neither were reported to have died. Broatch 2019 and Hagan 2008 did not report on this outcome.

Superficial vein thrombosis
Eight trials ( . The six separate randomised LONFLIT comparisons measured oedema using a score based on oedema tests, ankle circumference and volume, and swelling and discomfort as assessed by the participant. The score had a maximum (worst) value of 10 and was assessed before and a er the flight. Within each comparison, the randomised groups had similar oedema scores before the flight (mean values of approximately 1), and the final values (rather than a change score) are used in our analyses. These final values were approximately 2 or 3 for people in the compression stockings group, compared to 6 to 9 in the group allocated not to wear compression stockings. As reported by the trialists, each of the randomised comparisons showed a significant reduction in oedema in their own right, both on objective measures and on subjective measures reported by the participants. In each comparison, there was a small increase in oedema for participants in the compression stockings group but a much larger increase for the no compression stockings group. The longer flights showed the greatest di erences between the randomised groups. When combined, although there is significant heterogeneity among the results of the individual comparisons, the overall result clearly indicates a large and significant benefit for passengers who were allocated to wear compression stockings, compared to those allocated not to wear them (MD −4.72, 95% CI −4.91 to −4.52; 1246 participants, 6 studies; I = 92%; P < 0.001; low-quality evidence).
In the Hagan 2008 trial, oedema was measured by calculating the change of ankle circumference (before and a er landing) between those wearing compression stockings versus those not wearing compression stockings. Hagan 2008 reported that there was a decrease in ankle swelling compared with not wearing compression stockings (MD −0.19 cm, 95% CI −0.33 to −0.065 cm; P = 0.012).
The Loew trial also studied oedema, within its randomised between-individual comparison of wearing a compression stocking on one leg only (Loew 1998). These trialists assessed clinical oedema using a three point scale: 1 = none; 2 = slight; 3 = definite. They found that the leg on which a compression stocking was worn had less oedema than the other leg, and reported "oedema was most significant in the non-stockinged leg". Before flying, the oedema ratings were none: 56, slight: 5 and definite: 0 for the leg on which a stocking would be worn compared to 57, 4 and 0 respectively for the other leg. A er the flight, the scores for the leg on which a stocking had been worn had worsened slightly to none: 48, slight: 10, definite: 3; but the scores for the other leg were much worse at none: 30, slight: 22 and definite: 9.
In Broatch 2019, oedema was reported as calf girth. Measurements (cm) were obtained during baseline testing and a er travel. They found a small reduction (1.7 ± 1.9%; e ect size = 0.44 ± 0.51) in right calf girth for the participants who wore compression socks during the 9.5 hours of air travel.
It was not possible to pool the data from Broatch 2019, Hagan 2008 and Loew 1998 with data from the LONGFLIT 4 trials due to di erent methods used to measure oedema.

Adverse e ects arising from the use of compression stockings
Some of the reports of the LONFLIT trials commented on possible adverse e ects resulting from wearing compression stockings. In these reports, the tolerability of the stockings was described as very good with no complaints of side e ects (LONFLIT 4 -Scholl 1; LONFLIT 4 -Scholl 2; LONFLIT 4 -Traveno 1; LONFLIT 5). None of the other trials reported adverse e ects of wearing the stockings, apart

Summary of main results
This review provides relatively precise estimates for a very large reduction in symptomless DVT among airline passengers who were allocated to wear compression stockings compared to those allocated not to wear such stockings.
The choice of statistical analysis -in the circumstances we encountered of having rare events that are very unevenly distributed between the treatment groups -is potentially controversial. However, as shown in the results section, whichever method is used there is a clear e ect of wearing stockings compared to not wearing them, equivalent to the odds of a symptomless DVT being decreased by approximately 90% (high-certainty evidence). This might relate to, for example, a reduction in the risk of a symptomless deep vein thrombosis from about 10 to 30 per 1000 to 1 to 3 per 1000 long-haul passengers. There is also a large reduction in leg oedema associated with the wearing of stockings although the certainty of the evidence for this was deemed to be low. There is moderate-certainty evidence that superficial vein thrombosis may be reduced if passengers wear compression stockings. The studies reported no cases of symptomatic DVT, PE or deaths.
There is no robust evidence to indicate that the di erent types of stockings assessed in the trials included in this review vary in their e ects, nor that particular subgroups of people similar to those in these trials would not experience this benefit from wearing these stockings. A reliable investigation of these issues would require larger randomised trials, direct randomisation of di erent types of stockings and trials in which a wider range of participants are recruited. There also does not appear to be any significant increase in adverse e ects associated with wearing the stockings in the types of people assessed.

Overall completeness and applicability of evidence
The relevance of the substantial reduction in symptomless DVT for outcomes such as death, pulmonary embolism and symptomatic deep vein thrombosis cannot be assessed from this review because there were no such events in any of the included trials. This may be because the trials involved special additional tests on all participants which, by identifying symptomless DVT, may have led to e ective management and thereby prevented more serious consequences. It is also possible that death, pulmonary embolism and symptomatic DVT would have been so rare among the people in these trials that, even without the special diagnostic tests and subsequent treatments, no such events would have been recorded. Randomised trials to assess these outcomes would likely need to include a very large number of people. Therefore, this review provides a clear guide to the large e ects of compression stockings on reducing symptomless DVT and oedema, but is unable to assess the impact this has on outcomes that might be judged of more relevance to airline passengers and the people who care for them.

Quality of the evidence
This review provides high-certainty evidence that wearing graduated compression stockings reduces the risk of developing a symptomless DVT when travelling on a long-haul flight (over four hours). There is moderate-certainty evidence that wearing compression stockings may reduce risk of developing superficial vein thrombosis. The certainty of evidence was downgraded because the confidence interval crosses no di erence and does not rule out a small increase. Post-flight oedema was reduced in passengers who wore compression stockings but we have graded the certainty of this evidence as low. This is because the measurement technique used in the pooled data was not blinded or validated, and was carried out by the same investigators. As the methods from a further three studies used were di erent, we were unable to pool the data. However these additional studies also reported reduced oedema in passengers (or a leg) wearing compression stockings. As discussed above, we were unable to assess the certainty of the evidence for the outcomes of PE, death and symptomatic DVT as no events occurred.

Potential biases in the review process
The search used was comprehensive and we have included all relevant studies. However, the possibility remains that some relevant trials may have been missed. Two review authors independently performed study selection and data extraction in order to minimise bias in the review process. The inclusion and exclusion criteria set out in the protocol were strictly adhered to in order to limit subjectivity (Clarke 2003). We followed Cochrane processes as described by Higgins 2011 for assessing the risk of bias. Our analyses are based mainly on assessments of symptomless DVT which were identified through special tests and, if found, led to additional interventions for the participants. The reliability of the diagnosis of symptomless DVT is dependent on the quality of the test used. This can lead to di erent rates of false positives and false negatives for di erent tests. However, because we used randomised trials in which the assessment of participants in both groups of each trial involved the same diagnostic technique, the possibility of mis-diagnosis will have been the same for both groups and will not have introduced bias within the trials.

Agreements and disagreements with other studies or reviews
We are aware of other systematic reviews on this topic. Hsieh 2005 appears to have used similar methods to us and did not find any eligible studies that we had not already identified. Philbrick 2007 included case-control studies, cohort studies and randomised controlled trials which reported on travel as a risk factor for VTE; or tested preventive measures (including pharmacological agents) for travel-related VTE. They concluded that compression stockings "prevented travel-related VTE (P < 0.05 in 4 of 6 studies)". One cross-over trial compared intermittent pneumatic compression devices, compression stockings and periodic exercise in participants who were immobilised for eight hours (to mimic long-haul travel, Coppens 2007). Similar to our findings, this study reported that compression stockings decreased the amount of oedema experienced by wearers and suggested that this may play a role in preventing VTE.

Library
Trusted evidence. Informed decisions. Better health.

Implications for practice
High-certainty evidence shows that airline passengers similar to those in the trials in this review can expect a substantial reduction in their risk of a symptomless DVT if they wear compression stockings. Wearing stockings might reduce the incidence of this outcome from a few tens per thousand passengers, to two or three per thousand. There is moderate-certainty evidence that superficial vein thrombosis may be reduced if passengers wear compression stockings. Low-certainty evidence shows that passengers who wear stockings will also experience less oedema in their legs. However, this review is unable to identify whether these e ects of wearing stockings translate into e ects on outcomes such as death, pulmonary embolism and symptomatic DVT.

Implications for research
This review shows that the question of the e ects on symptomless DVT of wearing versus not wearing compression stockings in the types of people studied in these trials should now be regarded as answered. Further research may be justified to investigate the relative e ects of di erent strengths of stockings or of stockings compared to other preventative strategies. Further randomised trials to address the remaining uncertainty about the e ects of wearing versus not wearing compression stockings on outcomes such as death, pulmonary embolism and symptomatic DVT would need to be large. As suggested by Adi 2004, a study to assess whether airline travel itself is associated with an increased risk of symptomatic DVT might require several tens of thousands of participants and so a randomised trial to investigate a preventative strategy would probably require a sample size at least this large.

A C K N O W L E D G E M E N T S
We dedicate this update to our good friend and colleague, Monica Kjeldstrom, who was an author on the protocol and first version of this review but died too young in 2014. Additional interventions: participants were asked to refrain from nicotine, alcohol, and caffeine consumption for 24 hours before each testing session, as monitored by 24-hour food diaries. During the flight participants were asked to refrain from alcohol and caffeine consumption, and to consume 500 mL of water every 2 hours to ensure adequate hydration; and all participants instructed to "move (their) legs and feet for 3 -4 minutes per hour while seated but to avoid excessive movement".

Outcomes
Outcomes assessed: countermovement jumps; subjective ratings of jet lag (alertness, mood, fatigue, muscle soreness, motivation, and overall health) using the jet lag questionnaire; cardiovascular function (systolic blood pressure, heart rate, and oxygen saturation); calf girth; blood samples (markers of coagulation and fibrinolysis: tissue factor, tissue factor pathway inhibitor, thrombin-antithrombin complex, and D-Dimer).

Funding
This study was supported using research funds from compression garment manufacturer 2XU (Australia).
Declarations of interest Not reported.

Notes
Study author contacted to check if VTE was investigated. Answer received and confirmed that it was not.

Bias Authors' judgement Support for judgement
Random sequence generation (selection bias) Unclear risk States 'participants were assigned (in a randomized and counterbalanced fashion) ...' but no details given Allocation concealment (selection bias)

Unclear risk No details provided
Blinding of participants and personnel (performance bias) All outcomes

High risk No blinding
Blinding of outcome assessment (detection bias) All outcomes

Unclear risk Not reported
Broatch 2019 (Continued) Compression stockings for preventing deep vein thrombosis in airline passengers (Review) Copyright

Interventions
Stocking was worn on both legs (tights) on either the outward flight or on the return. Type of stocking: SKINS travel and recovery garment, listed as class I medical device (SKINS Compression Garments Pty Ltd, Sydney, NSW; ID: 880116). 5 mmHg at ankle, 17 to 20 mmHg at calf, 10 mmHg above knee and 4 mmHg at buttocks. Length of flight: 9.6 hours (GCT), 9.7 hours (no GCT). Type of seat: not reported. Route and time of flight: not reported. Additional interventions: suggestions for in flight exercises given to both groups.

Outcomes
Primary outcomes assessed: difference in change of ankle circumference (before flight and after landing) between control (no GCT) and treatment (GCT). Additional interventions: participants were also randomised to receive either dried vine leaves (Antistax) for 7 days before the flight and on the day (13 people), diuretics on the day of the flight (9), or no drugs (13). Participants in the diuretics group were told not to take these drugs on the return flight.

Outcomes
Outcomes assessed: limb volume, oedema status, subjective symptoms, phlebological status and clinical and Doppler findings. Measurements were made soon after landing on both flights.

Funding
Not reported.
Declarations of interest Not reported.