Objectives Exposure to traffic-related air pollution (TRAP) has been associated with adverse respiratory and systemic outcomes. Physical activity (PA) in polluted air may increase pollutant uptake and thereby health effects. The authors aimed to determine the short-term health effects of TRAP in healthy participants and any possible modifying effect of PA.
Methods Crossover real-world exposure study comparing in 28 healthy participants pulmonary and inflammatory responses to four different exposure scenarios: 2 h exposure in a high and low TRAP environment, each at rest and in combination with intermittent moderate PA, consisting of four 15 min rest and cycling intervals. Data were analysed using mixed effect models for repeated measures.
Results Intermittent PA compared to rest, irrespective of the TRAP exposure status, increased statistically significant (p≤0.05) pulmonary function (forced expiratory volume in 1 s (34 mL), forced vital capacity (29 mL), forced expiratory flow (FEF25–75%) (91 mL)), lung inflammation (fraction of exhaled nitric oxide, FeNO, (0.89 ppb)), and systemic inflammation markers interleukin-6 (52.3%), leucocytes (9.7%) and neutrophils count (18.8%). Interquartile increases in coarse particulate matter were statistically significantly associated with increased FeNO (0.80 ppb) and neutrophil count (5.7%), while PM2.5 and PM10 (particulate matter smaller than 2.5 and 10 µm in diameter, respectively) increased leucocytes (5.1% and 4.0%, respectively). We found no consistent evidence for an interaction between TRAP and PA for any of the outcomes of interest.
Conclusions In a healthy population, intermittent moderate PA has beneficial effects on pulmonary function even when performed in a highly polluted environment. This study also suggests that particulate air pollution is inducing pulmonary and systemic inflammatory responses.
Statistics from Altmetric.com
What this paper adds
Previous studies have shown short-term effects of traffic-related air pollution (TRAP) and physical activity. However, they were generally not designed to disentangle the effects of air pollution from those caused by physical activity, nor their possible interaction, while this one was.
Our study finds that air pollution and physical activity both have independent effects, and shows that intermittent moderate physical activity increases pulmonary function irrespective of high levels of TRAP, and suggests that particulate air pollution induces pulmonary and systemic inflammatory responses.
Our study provides a possible explanation for some of the unexpected results of previous studies and stresses the importance of assessing air pollution as well as physical activity levels when assessing health effects.
Exposure to traffic air pollution is associated with adverse health effects such as respiratory symptoms, and also increased morbidity and mortality.1–4 Traffic-related air pollutants are thought to contribute to these adverse health outcomes through mechanisms of oxidative stress, and local and systemic inflammation.5 ,6
Under real world conditions in susceptible, but not in healthy participants, small lung function reductions were observed after short-term exposure to traffic-related air pollution (TRAP).7 Various studies also observed airway inflammation in susceptible7 ,8 and also in healthy participants,9–12 and systemic effects such as an increase in neutrophil granulocytes13 ,14 and other inflammatory blood markers.14
Regular physical activity (PA) reduces the risk of many adverse health outcomes such as cancer, cardiovascular disease and diabetes partly because of its effects on the inflammatory processes.15–18 Short-term exercise on non-regular basis produces a short-term inflammatory response, whereas long-term habitual exercise might lead to an anti-inflammatory effect19 such as reductions in serum C reactive protein (CRP) levels.20 PA also increases the volume of inhaled air.21 An increased minute ventilation may result in substantial increases in the inhaled dose of traffic-related air pollutants22–24 and potentially lead to adverse health effects.
Although studies have shown effects of PA and air pollution, they were generally not designed to disentangle the effects of air pollution from those caused by PA, nor their possible interaction. We hypothesised that short-term exposure to TRAP may cause respiratory and systemic health impairments in healthy adults, and that moderate PA in polluted air, as occurring in active transportation, may modify such health effects. Our objective was to determine, in a healthy population in a real-world situation, the short-term health effects of roadside air pollution in these participants and any possible modifying effect of PA.
The study was conducted in Barcelona, Spain, between February and November 2011 and followed a well-controlled crossover study design. The Ethics Review Committee of the ‘Institut Municipal d'Investigació Mèdica (IMIM)’ approved the study and all participants gave written informed consent prior to participation.
We recruited 31 volunteers for this study. Eligible participants were non-smoking adults in the age range 18–60 years, healthy and without history or findings of pulmonary or cardiovascular disease, or other acute or chronic conditions (including infections, fever, cold) except sporadic nasal allergies, but without medication-based treatment. Volunteers abstained from taking over-the-counter medications (eg, pain relievers), vitamins and herbal supplements before the experiment days. Alcohol consumption in the evening before the experiment was limited to one glass of wine or beer. At the time of recruitment, a first lung function test was performed to train the volunteers in spirometry testing for the first experiment day to assess their eligibility and for later reference. Two hours were found to be the average time that residents in Barcelona spent in transit throughout the day.25 Therefore, all volunteers were exposed for 2 h to either heavily polluted air (located on a pedestrian bridge approximately 5 m above a main transit roadway for motorised traffic (often diesel powered)—Ronda Litoral) or to low TRAP (pedestrian friendly market square—Barceloneta market square) between 8:00 and 10:00 during morning rush hour (see online supplementary appendix figure S1). Two participants were studied simultaneously on each occasion. During the 2 h exposure time, one participant performed intermittent exercise consisting of 15 min cycling on a cycle ergometer alternating with 15 min of rest while the second volunteer rested throughout the 2 h exposure. Each volunteer participated in each of the four exposure scenarios in a narrow time period to exclude confounding by seasonal variation (see online supplementary appendix figure S2). The exercise interval was repeated four times summing up a total of 1 h exercise during the 2 h exposure period. Each volunteer was required to reach his/her individual range of moderate PA intensity, which was controlled through his or her heart rate, monitored continuously by a fingertip pulse oximeter (Konica Minolta, Japan). For moderate-intensity PA, the volunteers’ target heart rate was 50–70% of their maximum heart rate, which was estimated for each volunteer on the basis of age and sex (males: HRmax=220 (age); females: HRmax=206–0.88(age)).26 During each of the PA intervals, all participants were instructed and supervised constantly by the same technician, to pedal at a speed and resistance level that brought them close to the upper level of estimated moderate PA pulse range (70%) of their individual maximum heart rate. Volunteers completed the four exposure scenarios in random order. Each volunteer was to participate in all four scenarios: low-level and high-level air pollution exposure, each in combination with and without PA. To avoid a diurnal effect, all exposures and measurements were conducted at the same hours during the day as well as during the same days of the week to account for variations in traffic characteristics.
Volunteers filled out a half-hourly activity diary noting time, activity, location, travel mode and perceived exposure to environmental tobacco smoke (ETS) or other pollutants during the 3 days before each experiment day. That allowed us to estimate the volunteers’ air pollution pre-exposure and energy expenditure due to PA for the preceding 3 days. All participants received identical meals during all exposure days.
All baseline and post-exposure health measurements took place in the clinical research facilities within the study centre, a 5 min drive away from the study sites. To keep the volunteers’ exposure to TRAP before baseline measurements minimal, volunteers were required to arrive at the study centre before 6:45 in the morning. Before then the TRAP is still low in Barcelona. No restrictions were made on the volunteers’ travel mode choice to the study centre. On volunteers’ arrival they were given time to rest before baseline health values were taken. After returning from the exposure site all participants remained in a quiet and temperature-controlled clinical research facility. All health measurements were taken by trained technicians according to standard operating procedures (SOP) in order to assure accurate sample collection.
The fraction of exhaled nitric oxide (FeNO) was measured pre-exposure and 3 times post-exposure (ca 30 min, 3 h and 6 h after exposure) by chemiluminescence using a hand-held nitric oxide (NO) analyser, NIOX MINO (Aerocrine, Sweden). The instrument assesses airway NO with a single breath method. Measurements took place in accordance with the American Thoracic Society and the European Respiratory Society (ERS) guidelines for offline measurements of FeNO.27 Participants were asked to exhale to residual volume, insert the mouthpiece, inhale to total lung capacity through an NO scrubber and subsequently exhale into the device for 10 s at a constant flow rate at approximately 50 mL/s, resulting in approximately 500 mL exhaled breath, which was used for analysis.
After the FeNO measurements, the lung function parameters forced expiratory volume in 1 s (FEV1), forced vital capacity (FVC), peak expiratory flow, and maximum expiratory flow at 25% and 75% of the vital capacity (forced expiratory flow, FEF25–75%) were measured using a portable EasyOne spirometer (ndd, Switzerland). Each participant performed at least three and a maximum of eight manoeuvres per testing time point. The best values of FEV1 and FVC meeting the reproducibility criteria (difference ≤150 mL) according to the guidelines of the ERS and American Thoracic Society were selected.28 ,29 Spirometry was performed at baseline and ca 30 min, 2 h, 4 h and 6 h after exposure end with volunteers in sitting position and wearing a nose clip.
A blood sample was taken before and around 30 min postexposure from all participants. All venepunctures for blood withdrawal were undertaken by a registered nurse according to SOP. Blood samples collected in EDTA vacutainers were analysed for complete blood cell count (haemogram) closely after withdrawal by the local hospital laboratory. Serum samples for measurement of interleukin (IL) 1b, IL-6, IL-8, IL-10 and tumour necrosis factor (TNF) α were processed shortly after blood withdrawal and instantly after aliquotation stored at −80°C and later analysed in duplicates using high-sensitivity LUMINEX technology (Merck Millipore, UK) following standard procedures. For a timeline of all the health measurements please refer to the online supplementary appendix.
The on-site exposure monitoring included continuous reading of UFP counts (ultrafine particle matter in the size range 0.01–1.0 µm), using an optical particle counter CPC 3007 (TSI, Minnesota, USA). Readings above 100 000 particles per cubic cm (#/cm3) from the TSI model 3007 CPC required a correction for coincidence, which, if left unadjusted, would result in under-estimation of particle counts at high concentrations.30 The following equation was applied: cpc_corr=38 457×(exp(cpc×0.00001)).30 Nitrogen oxides (NOx) were measured via an NOx analyser (2B Technologies, Boulder, Colorado, USA). During the 2 h exposure period PM2.5 and PM10 (particulate matter mass with aerodynamic diameters of less than 2.5 µm (PM2.5) and less than 10 µm (PM10)) were collected using a Harvard Impactor (HI) (Air Diagnostics and Engineering) at a flow rate of 10 L/min. The gravimetric analysis of air quality filter samples was conducted in a specialised laboratory according to SOPs in temperature and relative humidity-controlled conditions. We collected for a total of 112 PM2.5 and PM10 mass filter samples 18 field blanks that were equally distributed over the whole field work period in order to validate the collected data. The average blank values were subtracted from the final particulate matter mass. Black carbon (BC) concentrations were measured using a portable aethalometer (Magee Scientific, Berkeley, California, USA) and corrected for filter attenuation.31 ,32
Statistical analyses were performed using STATA V.12. Exposure data were summarised as means, tested for correlations and transformed to IQRs. The a priori α level was set at p<0.05 for all planned comparisons. Normality was evaluated using simple graphical methods and Kolmogorov-Smirnov test.
We applied mixed effect models for the analysis of repeated measurement data with baseline values and individuals both as random effects to account for intra-individual variability in all health outcomes. Blood biomarkers were not normally distributed. To account for that we calculated per cent changes from baseline for use in mixed model regression. Spirometry markers and FeNO were used as crude data, and means from post-exposure measurements were regressed together in each model, whereas the time of measurement was included as a covariate to test for non-linear responses among the different measurement time points. In addition, all models were adjusted for on-site ambient temperature and relative humidity, sex, age and body mass index. To account for volunteers’ exposure for the day before the experiments took place we included the estimated time that each volunteer was exposed to ETS, the energy expenditure for their physical activities expressed in metabolic equivalent of task (MET) based on Ainsworth et al33 compendium of physical activities and the volunteers’ NO2 exposure using time-adjusted land use regression models.34
We tested in separate models for PA versus rest, irrespective of the TRAP site, and high TRAP versus low TRAP site irrespective of the PA status. Another model included the four exposure scenarios (low and high air pollution with and without PA). Single pollutant models included continuous pollutant levels with additional adjustment for PA (yes/no).
Furthermore, we tested for interaction between PA and the TRAP site, and PA and each pollutant separately.
Analysis of variance (ANOVA) for repeated measures was applied to test for differences between the measurement times and between the exposure conditions. We compared differences within one time point of measurement (baseline, t1, t2, t3, t4) between the exposure conditions, as well as between the pre-exposure to postexposure change using ANOVA or t test (in blood markers) within each exposure condition separately.
Overall 31 volunteers were recruited for study. All participants except three completed the four different exposure scenarios. Two of these three volunteers had to be excluded due to predefined exclusion criteria diagnosed after entry into the study, and one volunteer left the study after participating in one trial day due to personal reasons. The remaining 28 volunteers completed all four trial days and are therefore included in the analysis (table 1).
All measured pollutants were statistically significantly higher in the high air pollution site (traffic site) compared to the low-level air pollution site (market square). Considerably contrasting levels could be found for NOx (which was 10 times higher in the traffic site compared to the market square site), BC (7× higher) and UFP (5× higher; table 2).
There were high correlations (>0.9) between BC, NOx and UFP concentrations, NOx and PM2.5 with BC, as well as PM2.5 with NOx and PM10. Moderate correlations were observed with PMcoarse (see online supplementary appendix table S2).
We compared means between the four exposure scenarios and found no statistically significant differences between the baseline values (see online supplementary appendix tables S3 and S4).
Changes over time in FEV1, FVC, FEV1/FVC ratio, FEF25–75% and FeNO means combined for high TRAP, low TRAP, PA and rest are presented in figure 1.
Associations between air pollutants, PA and health endpoints
Effects on lung function and lung inflammation parameters
In mixed effect analysis we observed statistically significant increases in FEV1 (34 mL, p=0.001), FVC (29 mL, p=0.017) and FEF25–75% (91 mL, p=0.001) for PA when compared to rest, irrespective of the TRAP exposure status. PA irrespective of the TRAP exposure status also increased statistically significant FeNO (0.803 ppb, p=0.040; table 3).
The exposure scenario ‘PA in low air pollution’ increased statistically significant FEV1 (31 mL, p=0.025) and FEF25–75% (101 mL, p=0.009).
PMcoarse increased statistically significant FeNO (0.803 ppb, p=0.040; table 3).
Our data only suggested a statistically significant interaction between PA, and PM10 and PM2.5 for FEV1/FVC ratio (p=0.048 and 0.047, respectively), and for PA and UFP for FeNO (p=0.046; table 3).
Effects on circulating inflammation markers
Pooled analysis for PA compared to rest irrespective of the TRAP exposure status was associated with a statistically significant per cent increase from baseline in leucocytes (9.7%, p≤0.001), neutrophils (18.8%, p≤0.001) and IL-6 (52.3%, p=0.034; table 4).
Analysis by exposure scenarios showed that PA in low TRAP exposure led to an increase of leucocytes (9.8%, p=0.007), neutrophils (24.6%, p≤0.001) and IL-6 (74.2%, p=0.030), and PA in high TRAP exposure increased more leucocytes (14.1%, p≤0.001) and less neutrophils (21.1%, p≤0.001; table 4).
Single pollutant models showed a positive association between interquartile increases in PM10 and PM2.5, and leucocytes (4.0%, p=0.046 and 5.1%, p=0.047, respectively). Neutrophils increased after the exposure to PMcoarse (5.7%, p=0.049) and IL-8 decreased from baseline with interquartile increases in UFP (−1.9%, p=0.041; table 4).
We found only a statistically significant interaction between PA, and PM10 and PM2.5 for neutrophil counts (p=0.022 and 0.025, respectively; table 4).
In this well-controlled case crossover design examining lung function, lung inflammation and systemic inflammation, in response to a 2 h TRAP exposure and PA in real-world situations, we found that PA was associated with a small, but statistically significant increase in FEV1, FVC, FEF25–75%, exhaled NO, leucocytes, neutrophil counts and IL-6. PM2.5 and PM10 increased leucocytes and neutrophils. Ultrafine particulate matter decreased IL-8. We found limited and inconsistent evidence of an interaction between PA and UFP, PM2.5 and PM10 for pulmonary and systemic inflammation.
Our study adds to the limited body of evidence of small studies with inconclusive and often mixed results of the effects of air pollution while physically active. Our study provides a possible explanation of some of the unexpected results of previous studies and stresses the importance of assessing air pollution and also PA levels when assessing health effects.
The average concentrations of ultrafine particles, BC and NOx levels at the high-exposure site were 5, 7 and 10 times higher than those measured at the low-exposure site, respectively. However, the high correlations among the pollutants made it difficult to separate the effects on the health endpoints. The average ultrafine particle levels found at our high traffic site were considerably higher than those found in previous real world exposure studies in the UK, the Netherlands or Belgium.7 ,9 ,35 Mean particle concentrations in our high air pollution site were comparable with exposures measured in chamber studies,36 ,37 while in our ‘low-exposure site’, air pollution levels were rather comparable with those categorised in other experimental studies as ‘high air pollution’ or ‘traffic site’.11 ,38
We found PA-associated increases in FEV1, FVC and FEF25–75%. This finding is consistent with previous studies investigating the health effects of air pollution combined with short bursts of PA. The authors also observed an improved lung function in healthy participants shortly after exposure. Giles et al observed an FEV1 increase in endurance-trained males after 20 km cycling following an exposure to filtered air. This increase was attenuated when participants were exposed to diesel exhaust prior to the exercise.39 Strak et al9 examined the effects of air pollution on healthy cyclists and also showed weak lung function increases immediately after cycling, and only 6 h after negative associations between lung function and air pollution. Jarjour et al found statistically non-significant changes in FEV1 when comparing pulmonary function after cycling on low-traffic and high-traffic routes. FEV1 increased post-ride on both routes and 4 h after cycling on the low-traffic routes, whereas on high traffic the FEV1 decreased slightly.38 In a Canadian study, 42 healthy adults cycled for 1 h on high-traffic and low-traffic routes as well as indoors. The authors found an increase in FEV1 2 and 3 h after exposure onset.11 In healthy participants, exercise usually has bronchodilatory effects, most likely due to an activation of β2-receptors by endogenous catecholamines.40
Local and systemic inflammation
We found that PA was associated with a statistically significant increase in FeNO post-exposure suggesting eosinophilic airway inflammation. The literature about the impact of exercise on FeNO is inconsistent; however, there is evidence for exercise increasing the amount of exhaled NO.27 ,41 ,42 In our study, a statistically significant (0.803 ppb, p=0.040) increase in FeNO was observed with the coarse fraction of PM. Bos et al35 found that FeNO levels increased after training in the urban setting, whereas FeNO did not change after aerobic training in a rural environment. The role of the PA itself in that study remains unclear due to potential variations between the two training programmes and, furthermore, the study design was not a cross-over design in subject allocation. In another study, ultrafine particles and soot during cycling were weakly and non-statistically significantly associated with increased FeNO 6 h after cycling.9 A Dutch study showed a TRAP-associated FeNO increase immediately and also 2 h after a 5 h exposure with intermittent cycling.10 The formation of NO is likely to be a normal physiological response to exercise leading to smooth muscle relaxation.43
The systemic inflammatory response in our study was predominantly neutrophilic and associated with PA irrespective of the TRAP exposure status. Besides, we found that PM10 and PM2.5 increased leucocytes, and PMcoarse increased neutrophils. In previous studies an effect of air pollution exposure has been observed on systemic inflammation with a mainly neutrophilic response.44 ,45 Unlike our results, air pollution appeared to increase neutrophil counts in the Bos et al35 study after training in the urban area but not in those participants training in a rural area. PA intensity consisted of running in this study, which was possibly more vigorous compared to our moderate intensity cycling protocol33 and, in addition, effects were based on longer term exposure. In another study an increase in blood neutrophils was observed after cycling for 20 min on a road while not after cycling in a clean room.13 The UFP levels on that road were comparable with those in our low air pollution site. McCreanor et al7 showed an increase in sputum neutrophils in asthmatics 24 h after walking for 2 h next to a street with mainly diesel traffic.
We observed statistically significant increased levels of IL-6 after PA. IL-6 is known as both a pro-inflammatory and anti-inflammatory cytokine as well as a myokine synthesised in muscles in response to exercise,46 a fact that is in keeping with our observation of increased IL-6 levels after PA. IL-6 in return stimulates the release of anti-inflammatory cytokines such as IL-10.47 We did not observe any increases in IL-10, possibly due to the too short time period between the PA and the blood withdrawal to detect such changes. Nwokoro et al48 studied commuting cyclists in London and did not find IL-6 or other cytokine increases. In a Belgian study IL-6 did not change statistically significantly from baseline after cycling either near a major bypass road or in a clean room.13 However, participants cycled only for 20 min, which may not be long enough to induce a detectable IL-6 release. The local and systemic inflammatory responses in our study were mainly related to particles with size diameter above 2.5 µm. The elemental composition of the PMcoarse fraction is a bit different compared to the smaller PM fraction and tends to contain more resuspended dust, mineral elements, tire-derived particles and some metals.49
We found no consistent evidence for an interaction between air pollution and PA for lung function, lung inflammation and systemic inflammation markers, suggesting that PA does not modify the effect of air pollution.
The strength of this study is its unique design: an experimental setting with volunteers exposed to real-world conditions. To the best of our knowledge this is the first study to use a controlled crossover study design, which allows disentangling the effects of PA from those attributed to TRAP. Unlike in chamber studies, the exposure in our real-world setting brings a mix of airborne pollutants actually present in real TRAP, which has been shown to lead to different health effects than those observed in chamber studies.50 Moreover, our crossover design facilitates the exclusion of confounding by factors that are stable within an individual over time but vary between participants, since every volunteer serves as his or her own control.
A limitation of the study was the relatively small sample size leading to restricted power for the selected biomarkers, and the relative high exposure in the ‘low exposure’ site. Moreover, we cannot exclude that the effects had been different in women compared to men due to the different hormonal status potentially leading to an altered vulnerability. The evidence of the effects of acute air pollution exposures specifically in females is limited and the authors of this study aimed for a one-to-one females/males ratio in our sample in order to ensure the representability of the normal population. However, due to our limited sample size we considered it as disadvantageous to perform a sex-specific analysis. Furthermore, performing the experiments in a real-world environment implied less controllable study conditions. Some unknown factors including day-to-day variations, with a range of unclear modifiers, may have influenced our study results. Also, we did not estimate the increase in the inhaled doses due to cycling, which somewhat limits the analyses. Another major limitation of this study is that volunteers could not be blinded to the exposure conditions. Volunteers, however, were not informed about pollutant levels. Furthermore, multiple comparisons remain a limiting factor in our data analysis. However, we reported all p values and pertaining CIs. Moreover, we focused more on the consistency of the results rather than on individual statistically significant p values.
To conclude, this study suggests that in healthy participants, intermittent moderate PA has beneficial effects on pulmonary function even when performed in a highly traffic-related polluted environment. This study also suggests that particulate air pollution is inducing airway and systemic inflammatory processes. We recommend that future studies assessing the health effects of air pollution in active travel modes should account for the effect of PA. All changes in selected biomarkers were small and observed in a healthy population without clinical symptoms. The clinical relevance as well as the reversibility of these effects remain unclear. We propose assessing both in similar designs in multiple repeated short-term exposures.
The work reported in this paper was part of the TAPAS project (Transportation, Air Pollution and Physical Activities: An Integrated Health Risk Assessment Programme of Climate Change and Urban Policies), which was financed by Coca-Cola Foundation. The authors thank the volunteers who took part in this study, and Magí Farré and Esther Menoyer (IMIM Hospital del Mar Medical Research Institute, Human Pharmacology and Clinical Neurosciences Group) for providing us with the facilities for clinical measurements and their support. They are also pleased to acknowledge the addition of laboratory expertise by Oscar Dias (IMIM Hospital del Mar Medical Research Institute—URLEC), and Parc de Salut Mar Biobank (MARBiobanc supported by Instituto de Salud Carlos III FEDER (RD09/0076/00036)) for samples storage. The ESCAPE project (European Study of Cohorts for Air Pollution Effects) and Gerard Hoek (IRAS), Rob Beelen (IRAS), and Marta Cirach (CREAL) for their help and for providing us with land-use-regression estimations, Zorana Jovanovic Andersen (University of Copenhagen) for statistical advice, Tania Martinez and Anna Sillero (CREAL) for helpful technical support, Grisa Mocniek (Aerosol, Slovenia) for the provision of a microaethalometer (Magee Scientific, USA), and the Institute for Risk Assessment Sciences (IRAS) for the provision of a Harvard Impactor (HI).
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Files in this Data Supplement:
- Data supplement 1 - Online supplement
Contributors NK was involved in the conception, hypotheses delineation and design of the study, the acquisition of the data, data analysis and interpretation, and writing of the article. AdN was involved in the conception, hypotheses delineation and design of the study, and was substantially involved in article revision prior to submission. DW was involved in the acquisition of the data and data analysis. DM was involved in the data analysis. GC-T and LB were involved in the acquisition of the data. SG was involved in the conception, hypotheses delineation and design of the study, data analysis and interpretation, and was substantially involved in article revision prior to submission. MJN obtained the funding, was involved in the conception, hypotheses delineation and design of the study, data analysis and interpretation, and was substantially involved in article revision prior to submission.
Competing interests None.
Patient consent Obtained.
Ethics approval Comité etico de Investigacïon Clïnica Parc de Salut Mar, Barcelona, Spain.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.