High Prevalence of Exercise-Induced Laryngeal Obstruction in Athletes


Medicine & Science in Sports & Exercise: November 2013 - Volume 45 - Issue 11 - p 2030–2035
doi: 10.1249/MSS.0b013e318298b19a
Clinical Sciences

Introduction: Unexplained respiratory symptoms reported by athletes are often incorrectly considered secondary to exercise-induced asthma. We hypothesized that this may be related to exercise-induced laryngeal obstruction (EILO). This study evaluates the prevalence of EILO in an unselected cohort of athletes.

Methods: We retrospectively reviewed the prevalence of EILO in a cohort of athletes (n = 91) referred consecutively during a 2-yr period for asthma workup including continuous laryngoscopy during exercise (CLE) testing. We compared clinical characteristics and bronchial hyperreactivity between athletes with and without EILO.

Results: Of 88 athletes who completed a full workup, 31 (35.2%) had EILO and 38 (43.2%) had a positive bronchoprovocation or bronchodilator reversibility test. The presence of inspiratory symptoms did not differentiate athletes with and without EILO. Sixty-one percent of athletes with EILO and negative bronchoprovocation and bronchodilator reversibility tests used regular asthma medication at referral.

Conclusions: In athletes with unexplained respiratory symptoms, EILO is an important differential diagnosis not discerned from other etiologies by clinical features. These findings have important implications for the assessment and management of athletes presenting with persistent respiratory symptoms despite asthma therapy.

1Respiratory Research Unit, Department of Respiratory Medicine, Copenhagen University Hospital Bispebjerg, Copenhagen, DENMARK; and 2Department of Respiratory Medicine, Royal Brompton Hospital, London, England, UNITED KINGDOM

Address for correspondence: Emil Walsted Nielsen, M.B., Respiratory Research Unit, Department of Respiratory Medicine, Copenhagen University Hospital Bispebjerg, Bispebjerg Bakke 23, DK-2400 Copenhagen NV, Denmark; E-mail: emilwnielsen@gmail.com.

Submitted for publication February 2013.

Accepted for publication April 2013.

Article Outline

Respiratory symptoms are commonly reported by athletes of all abilities and impact on their health and performance (19). Indeed, in some series, up to 70% of athletes report troublesome dyspnea (21), and asthma is the most commonly diagnosed chronic medical condition in Olympians (13).

The differential diagnosis for dyspnea in an athlete is broad and includes exercise-induced bronchoconstriction (EIB), hyperventilation, exercise-induced laryngeal dysfunction, and cardiac conditions (29). In athletes, EIB is often considered the primary and most important differential diagnosis and yet is often diagnosed and treated after a symptom-based appraisal alone (20). This is despite studies consistently highlighting a poor correlation between the presence of asthma-like symptoms and the objective evidence of EIB, that is, athletes clinically diagnosed with EIB often have little evidence to support the diagnosis (2,9,21,27). The reason for the poor predictive value of symptoms in this setting is uncertain but may relate to a high prevalence of conditions that mimic asthma (30), for example, wheeze and dyspnea being generated by a temporary obstruction at the level of the larynx (25,28).

The term exercise-induced laryngeal obstruction (EILO) has been used to encompass several entities that can result in exercise-associated obstruction of the larynx, including exercise-induced laryngomalacia, laryngochalacia, and vocal cord dysfunction (8,23). In its mildest form, EILO is seen as a minor decrease in the cross-sectional area of the laryngeal inlet, whereas in its most severe form, the laryngeal inlet may be completely obstructed (16,23). Obstruction of the laryngeal inlet during exercise can arise at either the supraglottic (i.e., arytenoids and supporting structures) or glottic level (i.e., vocal cords) (7,22).

The gold standard for diagnosing EILO is continuous laryngoscopy during exercise (CLE) (16). This technique uses flexible nasendoscopy to provide a continuous image recording of the larynx throughout exercise, thereby enabling a dynamic recording of the movement of the laryngeal structures. The uninterrupted nature of this recording is important, given the fact that symptoms may rapidly abate on exercise cessation (i.e., as ventilation falls rapidly) and thus be undetected if only assessed before and after a challenge. It also permits characterization of the anatomical location of the obstruction, which is recognized to have treatment implications (22,24).

At the Bispebjerg University Hospital in Copenhagen, Denmark, athletes referred with unexplained respiratory symptoms undergo a comprehensive workup that includes CLE testing in addition to evaluation of airway function and inflammation. The aim of this study was therefore to report our experience with this approach and specifically the prevalence and characteristics of EILO in a cohort of athletes referred for a 2-yr period. We hypothesized that EILO would be highly prevalent and would explain the differential diagnosis of exercise-associated respiratory symptoms in a significant proportion of the athletes referred. We also hypothesized that the assessment performed would detect several athletes on inappropriate treatment.

Back to Top | Article Outline



All athletes referred to the asthma service at Bispebjerg University Hospital in Copenhagen, Denmark, for the investigation of exercise-induced respiratory symptoms between 2010 and 2011 were evaluated. Clinical files were assessed using a questionnaire-based data sheet. Files were reviewed for clinical symptoms at presentation, pulmonary function tests, bronchoprovocation testing outcome, and airway inflammatory status.

Back to Top | Article Outline


After clinical assessment and baseline pulmonary function evaluation, 91 athletes were referred for CLE testing during the study period. An athlete was defined as an individual who undertook >10 h of exercise every week (12). Athletes with incomplete records were excluded from review. The present study was assessed by the Danish Ethics Committee (H-3-2012-FSP59) and was found not to require ethics approval, being a retrospective study of clinical routine investigations. Informed consent was obtained from all subjects.

Back to Top | Article Outline

Test Procedures

Clinical assessment

Respiratory symptoms (dyspnea and wheezing) at rest and during exercise were assessed by a senior respiratory physician during the initial clinical interview. Dyspnea and wheezing were classified according to phase in the respiratory cycle as inspiratory, expiratory, or both.

Back to Top | Article Outline

Anthropometric data

Age, sex, height, and weight were recorded, and body mass index (kg·m−2) was calculated.

Back to Top | Article Outline

Atopic status

A skin prick test was performed according to European standards with a panel of allergen extracts (11). When no skin prick test was performed, allergen-specific IgE was measured in serum using a radioallergosorbent test (14).

Back to Top | Article Outline

Pulmonary Function and Bronchodilator Reversibility Testing

Lung function and bronchodilator reversibility testing were measured using maximum expiratory flow volume in accordance with the standards specified by the European Respiratory Society and the American Thoracic Society (26).

Fractional exhaled nitric oxide was analyzed using equipment from Eco Medics (Duernten, Switzerland) or Aerocrine (Solna, Sweden) following the recommendations of the European Respiratory Society and the American Thoracic Society (3). A positive response was defined as >25 ppb (3).

Back to Top | Article Outline

Bronchoprovocation Testing

Athletes were diagnosed with asthma on the basis of the presence of symptoms and a positive bronchoprovocation test (BPT) or bronchodilator reversibility test.

Subjects withheld asthma medications in line with recommendations (4) before BPT. However, to ensure that a BPT was not falsely negative secondary to the influence of inhaled corticosteroid (ICS) treatment, those athletes taking ICS with a negative BPT had a repeated BPT after 8 wk of ICS treatment cessation. The ICS withdrawal was performed out of season for elite athletes.

If bronchodilator reversibility (BDR) testing was negative, one or more BPT were performed to confirm or rule out asthma. The BPT were done sequentially on different days in the following order: mannitol, methacholine, eucapnic voluntary hyperventilation until achieving a positive test, or until all three tests had been performed. If the referring physician had performed a recent BDR or BPT, this was not repeated. The BPT results in the present study are the tests performed after ICS withdrawal.

Methacholine challenge testing was performed on all athletes with forced expiratory volume in 1 s (FEV1) >70% predicted in accordance with the method of Yan et al. (31). Using a breath actuated dosimeter APS (Jaeger GmbH, Germany), the athlete inhaled incremental doses of nebulizer Methacholine (19.63 mg·mL−1) until reaching a drop in FEV1 of 20% or more compared with the pretest FEV1, which was considered a positive test (provocative dose of methacholine causing a fall in FEV1 of 20% or more ≤ 1.548 mg), or until reaching a maximum of 1.548 mg inhaled methacholine.

Mannitol bronchial provocation was performed as described by Brannan et al. (5). Airway hyperresponsiveness as a dichotomy variable with a cutoff value of 635 mg (PD15 ≤ 635 mg) was used. A reduction in FEV1 of 15% or more from baseline FEV1 was considered a positive test.

Eucapnic voluntary hyperventilation testing was performed as described by Anderson et al. (1). The athlete hyperventilated dry air (5% carbon dioxide) for 6 min until reaching a required ventilation volume of 30 times the prechallenge FEV1. A reduction in FEV1 of 10% or more in at least two consecutive postchallenge measurements was considered a positive test.

Back to Top | Article Outline

Continuous laryngoscopy during exercise

CLE testing was performed using the protocol previously described by Heimdal et al. (16), excluding ergospirometry. In brief, the larynx was visualized during exercise using a flexible video laryngoscope and Visera video recording equipment (Olympus, Tokyo, Japan). Topical xylometazoline hydrochloride and lidocaine gel and spray were used to prepare the nares and were not applied to the oropharynx. Before warm-up exercise, athletes were requested to inhale beta-2 agonist (1 mg of terbutaline or 0.4 mg of salbutamol) to minimize any bronchoconstriction that would potentially prevent the patient from exercising at maximum effort.

Back to Top | Article Outline

CLE scoring

The video recordings were graded by an asthma specialist, with experience in diagnosing EILO (V.B.), using the Norwegian standardized 0–3 scoring system (23) (Fig. 1). A diagnosis of EILO was based on the presence of symptoms and a Norwegian grade score of 2 or greater (i.e., at least moderate). To determine interobserver agreement, a second observer (E.W.N.) rated the recordings. The observers were blinded to other measures.

Back to Top | Article Outline

Statistical analysis

Mean ± SD values are reported for normally distributed data, and median and interquartile ranges otherwise. Athletes with and without EILO were compared using an unpaired t-test or Mann–Whitney U test as appropriate. Fischer’s exact test was used to compare categorical, unpaired data. A P value of <0.05 was considered significant. Observer agreement was calculated using weighted kappa. All data analysis was performed using the Statistical Package for the Social Sciences (version 20.0; SPSS Inc., Chicago, IL).

Back to Top | Article Outline


Back to Top | Article Outline

Subject characteristics and pulmonary function results

Of the 91 athletes referred for assessment, 88 underwent a complete assessment including CLE (n = 3 excluded secondary due to incomplete data). Thirty-one athletes (35.2%) had EILO, whereas 57 (64.8%) were classified as having normal laryngeal function during exercise. EILO was more prevalent in women than that in men (OR = 4.09, 95% CI = 1.52–11.04, P < 0.01) (Table 1).

Of the 88 athletes, 38 had evidence of asthma diagnosed by BPT in 94% and/or BDR in 10%. In athletes with EILO, 12 (13.6% of all athletes) also had a positive BPT and/or BDR (Fig. 2). There was no difference in the frequency of positive BDR/BPT between the EILO (38.7%) and the non-EILO groups (45.6%) (P = 0.65, Table 1). There was also no difference in age, body mass index, FEV1, fractional exhaled nitric oxide, atopic status, or reversibility to β2-agonists between athletes with and without EILO.

Supraglottic EILO was the predominant type and was seen in 22 athletes (25% of all athletes), whereas glottic EILO was seen in 3 athletes (3.4%), and a combination of supraglottic and glottic EILO was seen in 6 athletes (6.8%). There were no differences in anatomic location in athletes with positive BPT compared with those with negative BPT.

Back to Top | Article Outline


When subjects were evaluated according to the presence or absence of EILO, the character of symptoms was similar between the groups (P = 0.67). Importantly, a clinical report of inspiratory wheeze or dyspnea did not discriminate athletes with or without EILO, 48.4% and 47.4%, respectively. These results were similar in athletes with (P = 0.69) and without (P = 0.61) evidence of a positive BPT (Table 2).

Back to Top | Article Outline

Asthma treatment

At the time of referral, 53 athletes (60.2%) were taking regular asthma treatment, including the following: ICS, 43.2%; short-acting β2-agonists, 52.3%; and leukotriene receptor antagonists (LTRA), 14.8%. Of athletes treated for asthma, 33 (62.3%) did not have pulmonary function or bronchoprovocation evidence to support a diagnosis of asthma. Importantly, 62% of athletes with no evidence of asthma or EILO as well as 60% of those classified as nonasthmatic athletes with EILO were taking regular asthma medication (Fig. 3).

Back to Top | Article Outline


In a cohort of athletes referred with unexplained exercise-induced respiratory symptoms, one-third had evidence of EILO. This abnormality was commensurate with symptoms and in all cases was graded as at least moderate in severity. A further third of athletes were found to have bronchial hyperreactivity, and in 12.5% of athletes, EILO was found to coexist with evidence of bronchial hyperreactivity. These findings underpin the fact that EILO is a key differential diagnosis in athletes presenting with unexplained exercise-induced respiratory symptoms and has implications for the way in which athletes are appraised and treated.

Making a correct assessment and diagnosis of respiratory symptoms in athletes can be difficult (30). Many athletes presenting with wheeze and dyspnea receive asthma treatment without objective tests to confirm the underlying diagnosis (20,29). Our finding that the clinical assessment of the timing of symptoms and wheeze (i.e., inspiratory vs expiratory) was not of discriminatory value may explain the numerous prior reports of the poor predictive value of symptoms in predicting EIB in athletes (2,9,21,27). Moreover, other characteristics (e.g., atopic status and resting lung function) apart from sex were similar between athletes with and without EILO.

The limitations of clinical assessment alone are further emphasized by the finding that half of those reporting predominantly inspiratory features did not have EILO when tested with CLE. The high prevalence of EILO (35.2%), including those with both EILO and asthma (12.5%) in this population, might explain why some athletes still have exercise-induced respiratory symptoms despite receiving appropriate asthma treatment. Rundell et al. (28) found a similar overlap 9% of between EILO and asthma. These findings should highlight to clinicians that a diagnosis of EIB should not simply be discarded on the basis of a high clinical suspicion of EILO as the cause of symptoms.

The current study is the first to accurately describe the anatomical location of EILO in a cohort of athletes. Accordingly, 10% of the EILO occurred at the glottic level and 71% at the supraglottic level, whereas 19% had combined closure. This is consistent with other studies using CLE, where 17%–21% of EILO-positive subjects had glottic obstruction (6,7). A characterization of the nature and location of EILO is important and may have implications for treatment. Specifically, Maat et al. (22,24) have reported successful surgical intervention for those with supraglottic EILO.

Our findings of a high prevalence of EILO are supported by recent reports in other athletic groups (15) but are at variance with some previous studies (27). Specifically, Hanks et al. (15) reported vocal cord dysfunction in up to 70% of 148 collegiate athletes referred with exertional dyspnea, whereas Rundell et al. (28) reported a prevalence of 5.1%. This variance likely relates to the methods and diagnostic criteria used. Studies to date have predominantly evaluated laryngeal function before and after an exercise challenge and used crude estimates of upper airway obstruction (e.g., auscultation for stridor), whereas CLE permits a continuous assessment of symptoms as they occur. Furthermore, in some studies, it is not clear what is considered an abnormal constriction, and no graduation of the constriction is made. A slight constriction of the vocal cords during or after exercise might not be pathologic, and this might at least in part explain the prevalence variance from our work. Assessing an individual when symptoms are present remains the gold standard, and thus we suspect that our findings present a true representation of the impact of EILO in this clinical scenario. Further, interobserver variation exists when grading the EILO using the Norwegian method. Maat et al. (23) previously reported an interobserver agreement of 0.70–0.86 using weighted kappa. This is consistent with the present study, which presents a weighted kappa of 0.67 for supraglottic obstruction and 0.73 for glottic obstruction. Nevertheless, it is important that studies are replicated in other centers and in other athletic populations.

It is alarming to find that a significant number of athletes with EILO, but without asthma (N = 12, 13.6% of all athletes), were treated with asthma medication. In total, 33 athletes (62.3% of athletes taking asthma medication) received medication without having objective evidence of airway hyperreactivity. Although airway hyperresponsiveness might fluctuate over time, and thereby potentially have hidden cases of mild asthma in the present study, this emphasizes the importance of verifying the true underlying diagnosis with objective tests. However, even when verified objectively, asthma remains difficult to treat in some athletes, and our current findings suggest that EILO should be considered as a coexistent diagnosis.

The mechanism underlying EILO in athletes is currently unknown, and several questions remain. More specifically, the natural clinical course of EILO in athletes remains unclear, and it is not apparent why there exists a higher prevalence of EILO in female athletes (OR = 4.09 in this study). The ineffectiveness of existing asthma medication, such as inhaled β2-agonists, corticosteroids, or leukotriene antagonists, suggests that mechanisms are different from those in asthma, and it may be that processes relating to airway dimension, inhibition of laryngeal reflexes, or impairment of the power and innervation of the laryngeal muscles (17,18) are relevant. Larynx dimensions are known to cause expiratory flow limitation in women, particularly in aerobically fit individuals (10). It is possible that smaller upper airway dimensions could contribute to EILO, but this is yet to be investigated.

Back to Top | Article Outline

Methodological considerations

The high prevalence of EILO in this cohort may be influenced by a selection bias based on the fact that athletes were referred to our tertiary referral clinic, which specializes in asthma and sports. The present study is also limited by the retrospective nature of the analysis. Further prospective work would therefore be informative to provide prevalence in different centers using the CLE test.

A further limitation relates to our lack of current knowledge with respect to the repeatability of CLE testing or day-to-day variability of EILO. This is important because symptoms and the presence of EILO may vary, thus limiting estimates of prevalence based on a one-off assessment. Furthermore, it is possible that athletes without evidence of EILO or asthma when assessed in the laboratory setting might have different results if it was possible to assess symptoms “in the field” or under intense competition scenarios.

Back to Top | Article Outline


This study has established that EILO is a prevalent and important differential diagnosis in athletes presenting with unexplained exercise-induced respiratory symptoms. In this population, symptoms arising from EILO can mimic asthma and are not discernible by clinical appraisal alone.

These findings have significant implications for the diagnosis and ultimately treatment of respiratory symptoms in this group of individuals, and further work is needed to explore the mechanisms underlying EILO and optimum treatment.

The authors did not receive any funds from external sources for the present study. The results of the present study do not constitute endorsement by the American College of Sports Medicine.

The authors wish to confirm that there are no known conflicts of interest associated with this publication, and there has been no significant financial support for this work that could have influenced its outcome.

Back to Top | Article Outline


1. Anderson SD, Argyros GJ, Magnussen H, Holzer K. Provocation by eucapnic voluntary hyperpnoea to identify exercise induced bronchoconstriction. Br J Sports Med. 2001; 35 (5): 344–7.
2. Ansley L, Kippelen P, Dickinson J, Hull JHK. Misdiagnosis of exercise-induced bronchoconstriction in professional soccer players. Allergy. 2012; 67 (3): 390–5.
3. Barnes PJ, Dweik RA, Gelb AF, et al. Exhaled nitric oxide in pulmonary diseases: a comprehensive review. Chest. 2010; 138 (3): 682–92.
4. Bateman ED, Hurd SS, Barnes PJ, et al. Global strategy for asthma management and prevention: GINA executive summary. Eur Respir J. 2008; 31 (1): 143–78.
5. Brannan JD, Porsbjerg C, Anderson SD. Inhaled mannitol as a test for bronchial hyper-responsiveness. Expert Rev Respir Med. 2009; 3 (5): 457–68.
6. Christensen P, Thomsen SF, Rasmussen N, Backer V. Exercise-induced laryngeal obstructions objectively assessed using EILOMEA. Eur Arch Otorhinolaryngol. 2010; 267 (3): 401–7.
7. Christensen PM, Thomsen SF, Rasmussen N, Backer V. Exercise-induced laryngeal obstructions: prevalence and symptoms in the general public. Eur Arch Otorhinolaryngol. 2011; 268 (9): 1313–9.
8. Christopher KL, Morris MJ. Vocal cord dysfunction, paradoxic vocal fold motion, or laryngomalacia? Our understanding requires an interdisciplinary approach. Otolaryngol Clin North Am. 2010; 43 (1): 43–66, viii.
9. Dickinson JW, Whyte GP, McConnell AK, Nevill AM, Harries MG. Mid-expiratory flow versus FEV1 measurements in the diagnosis of exercise induced asthma in elite athletes. Thorax. 2006; 61 (2): 111–4.
10. Dominelli PB, Guenette JA, Wilkie SS, Foster GE, Sheel AW. Determinants of expiratory flow limitation in healthy women during exercise. Med Sci Sports Exerc. 2011; 43 (9): 1666–74.
11. Dreborg S, Belin L, Eriksson NE, et al. Results of biological standardization with standardized allergen preparations. Allergy. 1987; 42 (2): 109–16.
12. Elers J, Pedersen L, Backer V. Asthma in elite athletes. Expert Rev Respir Med. 2011; 5 (3): 343–51.
13. Fitch KD, Sue-Chu M, Anderson SD, et al. Asthma and the elite athlete: summary of the International Olympic Committee’s consensus conference, Lausanne, Switzerland, January 22–24, 2008. J Allergy Clin Immunol. 2008; 122 (2): 254–60, 260.e1–7.
14. Haahtela T, Jaakonmäki I. Relationship of allergen-specific IgE antibodies, skin prick tests and allergic disorders in unselected adolescents. Allergy. 1981; 36 (4): 251–6.
15. Hanks CD, Parsons J, Benninger C, et al. Etiology of dyspnea in elite and recreational athletes. Phys Sportsmed. 2012; 40 (2): 28–33.
16. Heimdal JH, Roksund OD, Halvorsen T, Skadberg BT, Olofsson J. Continuous laryngoscopy exercise test: a method for visualizing laryngeal dysfunction during exercise. Laryngoscope. 2006; 116 (1): 52–7.
17. Ho AM, Chung DC, Karmakar MK, Gomersall CD, Peng Z, Tay BA. Dynamic airflow limitation after topical anaesthesia of the upper airway. Anaesth Intensive Care. 2006; 34 (2): 211–5.
18. Ho AM, Chung DC, To EW, Karmakar MK. Total airway obstruction during local anesthesia in a non-sedated patient with a compromised airway. Can J Anaesth. 2004; 51 (8): 838–41.
19. Hull JH, Ansley L, Robson-Ansley P, Parsons JP. Managing respiratory problems in athletes. Clin Med. 2012; 12 (4): 351–6.
20. Hull JH, Hull PJ, Parsons JP, Dickinson JW, Ansley L. Approach to the diagnosis and management of suspected exercise-induced bronchoconstriction by primary care physicians. BMC Pulm Med. 2009; 9: 29.
21. Lund T, Pedersen L, Larsson B, Backer V. Prevalence of asthma-like symptoms, asthma and its treatment in elite athletes. Scand J Med Sci Sports. 2009; 19 (2): 174–8.
22. Maat RC, Hilland M, Røksund OD, et al. Exercise-induced laryngeal obstruction: natural history and effect of surgical treatment. Eur Arch Otorhinolaryngol. 2011; 268 (10): 1485–92.
23. Maat RC, Roksund OD, Halvorsen T, et al. Audiovisual assessment of exercise-induced laryngeal obstruction: reliability and validity of observations. Eur Arch Otorhinolaryngol. 2009; 266 (12): 1929–36.
24. Maat RC, Roksund OD, Olofsson J, Halvorsen T, Skadberg BT, Heimdal JH. Surgical treatment of exercise-induced laryngeal dysfunction. Eur Arch Otorhinolaryngol. 2007; 264 (4): 401–7.
25. Morris MJ, Christopher KL. Diagnostic criteria for the classification of vocal cord dysfunction. Chest. 2010; 138 (5): 1213–23.
26. Pellegrino R, Viegi G, Brusasco V, et al. Interpretative strategies for lung function tests. Eur Respir J. 2005; 26 (5): 948–68.
27. Rundell KW, Im J, Mayers LB, Wilber RL, Szmedra L, Schmitz HR. Self-reported symptoms and exercise-induced asthma in the elite athlete. Med Sci Sports Exerc. 2001; 33 (2): 208–13.
28. Rundell KW, Spiering BA. Inspiratory stridor in elite athletes. Chest. 2003; 123 (2): 468–74.
29. Weiler JM, Bonini S, Coifman R, et al. American Academy of Allergy, Asthma & Immunology Work Group report: exercise-induced asthma. J Allergy Clin Immunol. 2007; 119 (6): 1349–58.
30. Weiss P, Rundell KW. Imitators of exercise-induced bronchoconstriction. Allergy Asthma Clin Immunol. 2009; 5 (1): 7.
31. Yan K, Salome C, Woolcock AJ. Rapid method for measurement of bronchial responsiveness. Thorax. 1983; 38 (10): 760–5.


© 2013 American College of Sports Medicine