Effect of long term treatment with salmeterol on asthma control: a double blind, randomised crossover study

BMJ 1997; 314 doi: http://dx.doi.org/10.1136/bmj.314.7092.1441 (Published 17 May 1997)
Cite this as: BMJ 1997;314:1441
  1. Paul Wilding, research fellowa,
  2. Miranda Clark, research assistanta,
  3. Joanna Thompson Coon, research assistanta,
  4. Sarah Lewis, statisticiana,
  5. Lesley Rushton, statisticianb,
  6. Jon Bennett, research fellowa,
  7. Janet Oborne, research assistanta,
  8. Susan Cooper, research assistanta,
  9. Anne E Tattersfield, professora
  1. a Division of Respiratory Medicine, City Hospital, Nottingham NG5 1PB
  2. b Department of Public Health Medicine, University Hospital, Nottingham NG7 2UH
  1. Correspondence to: Professor Tattersfield
  • Accepted 3 February 1997


Objectives: To determine the effect of adding salmeterol 50 μg twice daily for six months to current treatment in subjects with asthma who control their inhaled corticosteroid dose according to a management plan.

Design: A double blind, randomised crossover study.

Setting: Nottingham.

Subjects: 101 subjects with mild or moderate asthma taking at least 200 μg twice daily of beclomethasone dipropionate or budesonide.

Interventions: Salmeterol 50 μg twice daily and placebo for six months each, with a one month washout. Subjects adjusted inhaled steroid dose according to guidelines.

Main outcome measure: Reduction in inhaled steroid use, exacerbations of asthma, and use of oral steroids.

Results: Data were available for 87 subjects. When compared with placebo salmeterol treatment was associated with a 17% reduction in inhaled steroid use (95% confidence interval 12% to 22%) with no significant difference in the number of subjects who had an exacerbation (placebo 25%, salmeterol 16%) or use of oral steroids. For secondary end points salmeterol treatment was associated with higher morning and evening peak expiratory flow and forced expiratory volume in one second; a reduction in symptoms, bronchodilator use, and airway responsiveness to methacholine; and no effect on serum potassium concentration, 24 hour heart rate, or the final forced expiratory volume in one second achieved during a salbutamol dose-response study.

Conclusions: In subjects who adjusted their inhaled steroid treatment according to guidelines the addition of salmeterol 50 μg twice daily was associated with a reduction in inhaled steroid use and improved lung function and symptom control.

Key messages

  • One hundred and one subjects with mild or moderate asthma took salmeterol 50 μg and placebo for six months each in a crossover study

  • Subjects adjusted their inhaled steroid dose according to a management plan based on peak flow recordings and symptoms

  • The dose of inhaled steroids was reduced by 17% with salmeterol, with no change in exacerbations or use of oral steroids

  • Despite the reduction in inhaled steroid dose, salmeterol was associated with bronchodilatation and a reduction in symptoms

  • The efficacy of salmeterol was maintained over the six months


Salmeterol, a long acting β2 agonist, when inhaled twice daily, causes bronchodilatation that is maintained over 24 hours.1 2 The findings by Sears et al that asthma control was worse when subjects took the short acting β2 agonist fenoterol regularly rather than a β2 agonist as required3 led to concerns that regular treatment with long acting β2 agonists might have similar adverse effects. Subsequent studies comparing salmeterol 50 μg twice daily with placebo have shown that salmeterol causes bronchodilatation that is maintained for at least three months4 5 and an improvement in quality of life6 and symptom control.4 5 6 7 8 9 When compared with placebo or salbutamol, however, salmeterol has usually not reduced exacerbations,4 5 8 9 10 11 nor does it reduce inflammation in asthmatic airways.12 13

Current recommendations suggest that treatment for asthma should be modified according to symptoms and peak expiratory flow measurements.14 15 The introduction of a long acting β2 agonist may therefore lead to a reduction in inhaled corticosteroid use. Previous studies have looked at the effect of adding salmeterol when inhaled steroid dose is kept constant. We determined the effect of adding salmeterol to other treatment on the basis of current practice, with subjects changing their inhaled steroid dose according to predetermined criteria based on symptoms and peak expiratory flow. We studied the long term efficacy and safety of salmeterol in subjects with mild or moderate asthma who took salmeterol and placebo for six months each.



We recruited 101 subjects (50 female) aged 19-60 years from our register of asthma volunteers and from outpatient clinics in Nottingham and Mansfield. The subjects had to have a forced expiratory volume in one second of at least 50% predicted and either (a) a 15% increase in forced expiratory volume in one second after inhaling salbutamol 400 μg at entry or within a year or (b) 15% diurnal variability in peak expiratory flow recordings. All were receiving an inhaled short acting β2 agonist as needed and at least 400 μg/day beclomethasone dipropionate or budesonide. Two subjects were taking ipratropium bromide and two theophylline, both in constant dose throughout the study. All had stable asthma at entry, with no exacerbations or respiratory tract infection in the previous six weeks. Subjects gave written informed consent; the study was approved by Nottingham City Hospital's ethics committee.


We measured forced expiratory volume in one second and forced vital capacity with a dry bellows spirometer, the Vitalograph (Vitalograph, Buckingham), as the best of three readings within 100 ml for forced expiratory volume in one second. We measured peak expiratory flow with a mini-Wright peak flow meter (Clement Clarke, Harlow) as the best of three readings. A salbutamol dose-response curve was obtained by the patient inhaling salbutamol at 15 minute intervals to provide cumulative doses of 100 μg, 200 μg, 400 μg, and 800 μg. We measured bronchial reactivity using a modified method of Yan et al,16 with the subjects inhaling three puffs of saline followed by doubling doses of methacholine from 0.048 μmol to 196 μmol. We measured forced expiratory volume in one second one minute after each dose, and the test was stopped when it had fallen by at least 20%. The provocative dose of methacholine causing a 20% fall in forced expiratory volume in one second (PD20) was calculated by linear interpolation of the last two readings on the log dose-response plot. The subjects took no study drugs within eight hours, no short acting β2 agonists within four hours, and no ipratropium bromide within 24 hours of all visits.

The subjects recorded peak expiratory flow before taking the study drug, rescue bronchodilator use, and day and night-time symptom scores (0 = no symptoms, 4 or 5 = severe symptoms) twice daily throughout the study.


This was a double blind, crossover study in which the subjects inhaled placebo and salmeterol 50 μg twice daily from identical dry powder inhalers (Diskhaler with Rotadisk; GlaxoWellcome, Greenford) for 24 weeks followed by a four week washout, the order of treatment being randomised by computer program. We assessed asthma control after a two week run-in when the subjects recorded peak expiratory flow and symptom scores. If control was considered reasonable by both the doctor and the subject an individualised “target” peak expiratory flow representing good control of asthma was determined and the subject issued with a personalised management plan (see below). Otherwise the steroid dose was increased or reduced by one puff twice daily with further review after four weeks, or for longer if necessary, until their asthma was stable. One puff could be 100 μg, 200 μg, 250 μg, or 400 μg of beclomethasone dipropionate or budesonide according to the inhaler used (which was constant throughout the study). The subjects requiring less than 200 μg beclomethasone dipropionate or budesonide twice daily were excluded from entry.

We measured forced expiratory volume in one second and forced vital capacity every four weeks during the 56 week study and determined the number of used and unused Rotadisks to assess compliance. We measured PD20 methacholine and serum potassium concentration after 12 and 24 weeks of each treatment, carried out a salbutamol dose-response study at 20 weeks (both at least 8 hours after study treatment), and performed 24 hour electrocardiography by Holter monitor during weeks 1 and 24. A 24 hour contact telephone number and physician with a radiopager were available throughout the study.

Asthma management plan

We adjusted treatment at clinic visits and between visits with written and oral instruction, which was reinforced at each visit. When the subjects had had no symptoms and a morning peak expiratory flow greater than or equal to their target peak expiratory flow for 14 consecutive days they reduced their inhaled steroid dose by one puff twice daily; they were not allowed, however, to reduce the dose by more than one puff twice daily below the dose at entry. If morning or evening peak expiratory flow fell by 10% on two consecutive days or by 15% on one day subjects increased their inhaled steroid by two puffs twice daily for at least one week, adding 30 mg prednisolone daily if peak expiratory flow fell by 30%. The additional treatment was continued until symptom scores and peak expiratory flow had returned to previous levels, when it was reduced by one puff twice daily per week to the previous dose or by 5 mg prednisolone every two days; if a subject's condition deteriorated the higher dose was maintained.

Exacerbations were defined by the presence of any two of the following: a 30% fall in morning peak expiratory flow from target peak expiratory flow; a fall in forced expiratory volume in one second of either 0.7 litres or 30% from baseline; increased use of β2 agonist (by more than four puffs a day); the need for oral prednisolone; more treatment on two consecutive nights or increased symptom scores on two successive days.


Primary end points were mean daily inhaled steroid dose, number of exacerbations, and courses of oral prednisolone over the last five months of treatment. Parametric analyses were performed on mean values of forced expiratory volume in one second; morning and evening peak expiratory flow; minimum, mean, and maximum heart rate; and log transformed values for daily inhaled steroid dose, serum potassium concentrations, and PD20 methacholine. PD20 values were included only if measured within 8-14 hours of the study drug (66% of subjects). Reversibility to salbutamol was characterised by mean initial and maximum forced expiratory volume in one second and mean increase in forced expiratory volume in one second. For end points measured over the last five months of treatment, all the subjects with data for at least one day of treatment were included in the analyses. Initially, an analysis of variance model was fitted containing terms for the main effects of baseline (mean over last two weeks of the run-in when appropriate), treatment, period, carry over, subject within sequence (order of drug randomisation), and a baseline by treatment interaction.17 The effects of baseline, carry over, and interactions were removed if non-significant at the 10% level; the main model then fitted included terms for subject, period, and treatment only.

All remaining outcomes including the percentage of days (24 hours) without use of a bronchodilator and symptom free days and nights were analysed non-parametrically by Wilcoxon's rank sum test to establish the significance of treatment and carry over effects. Differences in numbers of exacerbations and courses of oral steroids were compared with Prescott's test.17


Of the 101 subjects who entered the study, 14 subjects failed to complete the study owing to protocol violations (three), pregnancy (three), personal reasons (five), and adverse events (three; see results). Data for the efficacy analysis were available for 87 subjects, including one subject who moved house during month 10. Mean compliance was 92%, with 67% and 64% of subjects achieving more than 90% compliance with placebo and salmeterol respectively.

Subjects consisted of 51 men and 50 women with a mean age of 39 (SD 10) years; 50 had never smoked, 39 were former smokers, and 12 were current smokers. Most of the 101 subjects had a positive skin test to house dust mite (71%), cat (79%), or grass pollen (56%). Table 1) shows the subjects' lung function.

View this table:
Table 1

Baseline characteristics of subjects receiving salmeterol first (sequence 1) and placebo first (sequence 2) and of all subjects combined

Baseline measurements of subjects were similar regardless of which drug they had been randomised to receive first–salmeterol (n = 51) or placebo (n = 50) (table table 1). There were no carry over effects (all P>0.15) apart from minimum heart rate (P = 0.09), which was excluded from further analysis. The only significant interaction between treatment and baseline measurements was a weak interaction with baseline inhaled steroid dose (P = 0.10).

Primary end points

Geometric mean inhaled steroid use was stable over the last five months of each treatment (fig 1) and significantly lower with salmeterol (561 μg) than with placebo (674 μg); the average difference was 17% (95% confidence interval 12% to 22%, P<0.001). Analysis according to baseline inhaled steroid dose showed that the 45 subjects initially taking 600 μg or less had a 19% lower inhaled steroid dose with salmeterol (329 μg) than with placebo (404 μg) (12% to 25%), whereas the 42 subjects taking more than 600 μg inhaled steroid initially showed a 14% difference (6% to 21%; 998 μg v 1162 μg inhaled steroid).

Fig 1

Geometric mean inhaled steroid dose during treatment with salmeterol and placebo, with standard error bars

Most subjects had no exacerbation (n = 52) and needed no oral steroids (n=66) during the study, and most of the remaining subjects had one exacerbation or one course of oral steroids only (table table 2). Oral steroids were used more often in the first six months than the second six months with both placebo (8 v 5 subjects) and salmeterol (7 v 2).

View this table:
Table 2

Oral steroid use and asthma exacerbations during last five months of treatment with placebo and salmeterol

With salmeterol and placebo 16 and 25 subjects respectively had an exacerbation, and 9 and 13 subjects had a course of oral steroids; neither difference was significant. Most (27/34) of the episodes in which subjects took oral steroids fulfilled the criteria for an exacerbation; some exacerbations were not treated with oral steroids.

Secondary endpoints

Lung function and bronchial responsiveness

Table table 3) and figure 2) show that salmeterol was associated with higher peak expiratory flow values compared with placebo in both the morning and evening and that forced expiratory volume in one second and forced vital capacity were also significantly higher with salmeterol.

View this table:
Table 3

Mean values for secondary end points after placebo and salmeterol

Fig 2

Mean morning and evening peak flow and forced expiratory volume in one second during treatment with salmeterol and placebo, according to whether salmeterol was given first (sequence 1) or second (sequence 2)

Geometric mean values for PD20 methacholine were significantly higher after salmeterol than placebo (table table 3). The salbutamol dose-response study showed a higher initial forced expiratory volume in one second (2.82 v 2.71 litres, P<0.001) and a smaller increase in forced expiratory volume in one second (11% v 16%) during the salmeterol treatment period compared with placebo, but the maximum forced expiratory volume in one second achieved was almost identical (3.11 v 3.10 litres, P=0.68) (fig 3).

Fig 3

Mean forced expiratory volume in one second in response to increasing doses of salbutamol following salmeterol and placebo. The dose-response studies were carried out at least eight hours after the last dose of salmeterol

Symptom scores and use of bronchodilator

The subjects had more symptom free days and nights and used less relief bronchodilation with salmeterol. The median percentages for salmeterol and placebo were 96% and 86% for symptom free nights, 82% and 64% for symptom free days, and 73% and 58% for bronchodilator free days (all differences significant at P<0.001).

Safety aspects

Serum potassium concentrations with salmeterol and placebo did not differ significantly (table table 3). The quality of the 24 hour electrocardiography on all four occasions was sufficient to be able to assess maximum heart rate in 60 subjects and mean heart rate in 47. There were no significant differences in mean or maximum 24 hour heart rate between placebo and salmeterol (table table 3) and no difference in the frequency distribution of ventricular and supraventricular ectopic beats.

Three subjects withdrew from the study because of adverse events, two during or after salmeterol treatment (chest pain and hair loss), and one during placebo (headache).

Salmeterol was associated with more palpitations (4 v 1); other expected adverse effects occurred infrequently with placebo (cramp 2, hypokalaemia 1, tremor 1) but not with salmeterol treatment.


Long term regular treatment with the short acting β2 agonists has provided no perceptible clinical benefit in most studies.3 4 5 6 18 19 Concern that they may have deleterious effects emphasised the need for long term studies of the long acting β2 agonists.3 Recent studies suggest that when added to other treatment the long acting β2 agonists improve asthma control compared with placebo 4 5 6 7 8 9 10 11 and salbutamol.4 5 6 Our study design differed from previous studies in that the subjects were asked to adjust their inhaled steroid intake within certain limits according to their peak flow rate and symptoms. This design was chosen to ensure that subjects were not undertreated or overtreated with steroids for a long period and to mimic what is likely to happen in clinical practice when salmeterol is introduced.14 15 The subjects were taking different types and different strengths of steroid inhaler, but these remained constant for an individual throughout the study.

The subjects followed their management plan well, and the relatively low incidence of exacerbations suggests that it was effective.

By basing the study on a single centre we aimed to ensure high compliance with treatment and a low withdrawal rate as this can be an important source of bias. Data were available for 87 of the 101 subjects, and mean compliance with treatment was 92%. Subjects who failed to complete did so largely for personal reasons, with no difference between salmeterol and placebo in the number or reasons for withdrawal.

The addition of salmeterol for six months was associated with a reduction in inhaled steroid use and symptoms and an increase in forced expiratory volume in one second and morning and evening peak expiratory flow. We have presented the findings over the last five months of treatment to reduce any crossover effects and to allow time for subjects to adjust their inhaled steroid dose in response to change in treatment. We found no carry over effects apart from minimum heart rate, which was omitted from further analysis, and no treatment baseline interactions except for a weak relation with baseline log inhaled steroid dose, which we explored further.

The mean inhaled steroid dose was 17% lower with salmeterol than with placebo during the last five months of treatment. The subjects on lower doses of inhaled steroid initially had a slightly higher percentage fall in inhaled steroid dose with salmeterol, though there was considerable overlap with those initially on a higher dose. The 17% reduction should not be interpreted as the steroid sparing dose of salmeterol as subjects had limited freedom to reduce their inhaled steroid dose (by two puffs a day maximum from the dose at randomisation so that the minimum dose possible was 100 μg twice daily). If all subjects had reduced their dose to the minimum allowed the geometric mean dose would have been 387 μg (rather than 561 μg). The decision to limit the reduction in dose was based on current consensus guidelines14 15 and an appreciation that peak expiratory flow is only one measure of asthma control. Salmeterol was associated with fewer symptoms and no difference in exacerbations or use of rescue oral steroids compared with placebo, despite the reduction in inhaled steroid dose.

Relatively few subjects had an exacerbation during the study, probably because the management plan ensured early detection of deterioration and encouraged prompt treatment. Most studies comparing salmeterol with placebo or a short acting β2 agonist have found no difference in exacerbations, although most had insufficient power to detect this.4 5 8 9 10 11 A large surveillance study found a smaller reduction in withdrawals owing to asthma with salmeterol compared with salbutamol, although the trend for deaths was in the opposite direction.20 Airway inflammation is not reduced by salmeterol or the shorter acting β2 agonists compared with inhaled steroids12 13; whether this is relevant to the frequency or severity of exacerbations is uncertain.

This is the first controlled study to compare salmeterol 50 μg twice daily with placebo for more than three months. The effect of salmeterol was maintained over six months, with no evidence of tolerance for any clinical end point. When we looked at bronchodilator responsiveness to salbutamol baseline forced expiratory volume in one second was higher with salmeterol, but the final forced expiratory volume in one second achieved was identical in the two treatment periods, as in previous studies.21 22 We interpret this as showing that bronchodilator responsiveness is maintained after regular salmeterol treatment.

The PD20 methacholine was higher after salmeterol than placebo, although the differences were small. The magnitude of the effect will be affected by the timing of the measurement, 8-14 hours after treatment, when some residual effect of salmeterol would be expected, and possibly by the development of tolerance, as seen with other β2 agonists23 24 as well as salmeterol.25 26

Salmeterol, like other β2 agonists, causes systemic effects such as hypokalaemia and tremor when given in high doses.27 28 Such effects are rare with a 50 μg twice daily dose,8 28 and there was no difference in adverse effects, serum potassium concentrations, or 24 hour heart rate between treatments.

The design of this study is relevant to how salmeterol is likely to be used in clinical practice, and the findings provide some reassurance that if salmeterol is introduced at an earlier point in treatment14 (at step 3 rather than step 4 of the guidelines for treatment of asthma) a subsequent reduction in inhaled steroid dose is unlikely to have a detrimental effect in patients who followed an appropriate management plan.


We thank the β agonist working party of the National Asthma Campaign for their support; Loretta Jacques, Gillian Brooks, and Namrata Taak from GlaxoWellcome for their help at all stages of the study; Chris Knowles for help with Holter monitoring; Drs I D A Johnston and M Ward for help with patient recruitment; Dan Duthie for help with the figures; Rachel Holt and Phil Dennis in Nottingham and Lynn Knowles and the staff in the cardiorespiratory unit at King's Mill Hospital, Mansfield, for technical support; and Sarah Pacey for help with drug storage and administration.

Funding: The study was funded by GlaxoWellcome and the National Asthma Campaign.

Conflict of interest: None. (The study was designed and analysed by the authors, and the results were checked against a similar analysis carried out by GlaxoWellcome.)


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 19.
  20. 20.
  21. 21.
  22. 22.
  23. 23.
  24. 24.
  25. 25.
  26. 26.
  27. 27.
  28. 28.