Systematic review

We found 1 systematic review of good quality that compared the combination of an ICS plus a LABA with another ICS/LABA combination for controller therapy.⁹⁴ The review included only randomized, controlled, parallel-design trials and required that only single inhaler devices were used to administer study drugs. Studies lasting fewer than 12 weeks or administering “adjustable maintenance dosing” or “single inhaler therapy” rather than fixed doses were excluded. The review included five studies, all of which compared BUD/FM with FP/SM and included a total of 5,537 adult and adolescent subjects. Three of the five are included in the RCT section of this report;^{95, 97, 98} one was excluded from this report due to the study design, with a second randomization at one month (only allowing a valid comparison of FP/SM with BUD/FM for one month; our duration criteria was at least 6 weeks).¹⁰² The fifth was a study whose results were not published. Doses of BUD and FM in the included trials ranged from 400–800mcg/day and 12–24mcg/day, respectively. All of the studies administered 500mcg and 100mcg of FP and SM per day. Included studies ranged from 12 weeks to 30 weeks and took place in the United States and Europe.

All included studies enrolled adolescents and adults, and neither restricted asthma severity or current treatment, although participants had to have a history of chronic asthma, treated with moderate to high maintenance doses of ICS prior to entry. All trials required patients to be stable for one month before the run-in period and to continue to demonstrate the need for frequent reliever use during the run-in. Demographics of the included studies indicated that treatment and comparison groups were well-balanced. All included studies were funded by pharmaceutical manufacturers.

Four of the trials measured symptom scores, rescue medication use and exacerbations.^{95, 97, 98, 102} Two trials used a double-blind, double-dummy design; ^{97, 98} the other two were open-label. There were no statistically significant differences between FP/SM and BUD/FM in mean change in daytime symptom scores (three studies; treatment difference = −0.02; 95% CI −0.6 to 0.03; N = 3,464) or percent of symptom-free days (two studies; treatment difference = 1.25; 95% CI −1.18 to 3.67; N = 3,027). Exacerbations were reported as participants experiencing an exacerbation requiring oral steroid treatment and as participants experiencing exacerbations resulting in hospital admission. For exacerbations requiring oral steroid treatment, there was no statistically significant difference between FP/SM and BUD/FM (four studies; OR = 0.89; 95% CI 0.74 to 1.07; N=4,515). Similarly, no statistically significant difference was found between FP/SM and BUD/FM groups for exacerbations resulting in hospital admissions (four studies; OR = 1.29; 95% CI 0.68 to 2.47; N = 4,053). In addition, a composite measure was created in order to measure exacerbations resulting in a hospital admission or an emergency department visit. This comparison also failed to yield a statistically significant difference between treatments (four studies; OR 1.3; 95% CI 0.94, 1.8; N = 4,861). There was also no significant difference between FP/SM and BUF/FM in rescue medication use (three studies; treatment difference = −0.06 puffs/day; 95% CI −0.13 to 0.02; N = 3,469).

Randomized controlled trials

Of the four RCTs we included (seven articles) (Table 12), all four compared the same medications (BUD/FM compared with FP/SM). All but one study administered both of the ICS+LABA combinations in a single inhaler; one trial administered BUD+FM in separate inhalers.¹⁰¹ Study duration ranged from 12 weeks¹⁰¹ to seven months.⁹⁵ All four trials administered BUD and FM via DPI; three did so in a single DPI; one trial administered BUD+FM in separate inhalers.¹⁰¹

Within-trial equipotency of daily ICS dose varied. All four trials administered the same total daily dose of FP/SM (500/100), which is considered a medium daily dose of ICS when delivered via DPI and a high daily dose when delivered via pMDI (Table 3). In two trials, 500mcg of FP was compared with an equipotent daily dose of BUD.^95–97 In one of these, there was a third arm that contained an adjustable-dose BUD/FM arm, although this is not a comparison of interest for the current report. Of the non-equipotent dosage studies, one study compared low (but adjustable) and medium (but fixed) daily doses of BUD with a high dose of FP,^98–100 and another compared a high daily dose of BUD with a medium dose of FP.¹⁰¹

Study Populations

The four head-to-head RCTs included a total of 5,818 subjects. All studies were conducted in adolescent and/or adult populations. None included children < 12 years of age. All trials were multinational. All enrolled subjects that were not adequately controlled on current therapy. Three were conducted in subjects with moderate to severe persistent asthma; one did not report the severity classification.^{98, 99} Three trials (75%) excluded smokers with at least a 10 pack-year history; one (25%) allowed some smokers and reported that 5% to 7% of subjects in each group were current smokers.

Sponsorship

Of the four head-to-head trials, 3 (75%) were funded by pharmaceutical companies; 1 trial (25%) did not report the source of funding but at least one author had a primary affiliation with a pharmaceutical company. No trials were funded primarily by a source other than a pharmaceutical company.

Head-to-head comparisons

1. BUD/FM MART compared with ICS/LABA for maintenance and SABA for relief

The results of the four RCTs contributing five comparisons (one study compared BUD/FM MART with BUD/FM maintenance and SABA relief and with FP/SM maintenance and SABA relief) are described below under the appropriate drug comparisons. Overall, all five comparisons reported statistically significantly lower rates of exacerbations for those treated with BUD/FM MART, but no differences in symptoms.

We conducted meta-analyses for seven outcomes that were reported with sufficient data in multiple trials (Appendix I). These included symptom-free days, symptom scores, nocturnal awakenings, exacerbations requiring medical intervention, exacerbations resulting in emergency visit or hospital admission, rescue-free days, and rescue medicine use (puffs/day).

Our meta-analysis for exacerbations requiring medical intervention shows an odds ratio of 0.75 (95% CI: 0.66, 0.85; 5 comparisons) favoring MART. A separate meta-analysis of exacerbations resulting in emergency department visits or hospital admissions revealed similar findings; the odds ratio for MART was 0.73 (95% CI: 0.60, 0.90; 4 comparisons). MART was also associated with fewer nocturnal awakenings, compared with ICS/LABA + SABA (SMD = −0.08; 95% CI = −0.12, −0.03; 4 comparisons). I² values for each of these analyses was < 25%.

We found no statistically significant differences in symptom-free days (SMD = 0.02, 95% CI: −0.02, 0.06, 3 studies contributing 4 comparisons), symptom scores (SMD = −0.02, 95% CI: −0.07, 0.03, P = 0.48; 4 studies contributing 5 comparisons), rescue-free days (SMD = −0.04, 95% CI: −0.09, 0.01, 3 studies contributing 4 comparisons), or rescue medicine puffs per day (SMD = −0.06, 95% CI: −0.14, 0.02, P = 0.14; 4 studies contributing 5 comparisons). The I² value for rescue medication use was 76.6, indicating high statistical heterogeneity.

Of note, the comparisons that administered scheduled maintenance ICS doses that were lower in the BUD/FM MART group all found statistically significantly lower exacerbation rates for those treated with BUD/FM MART.^{98–100, 104} In addition, the BUD/FM MART group had a lower mean daily steroid dose (maintenance plus relief) than the ICS/LABA for maintenance with SABA relief in three of the five trials.^{98–100, 104, 106} Thus, it does not appear that delivering a higher total ICS dose explains the better exacerbations outcomes in the BUD/FM MART group.

2. BUD/FM MART compared with BUD/FM for maintenance and SABA for relief

We found one good-^98–100 and one fair-quality ^{103, 105} RCT for this comparison. Both trials reported asthma symptoms, nocturnal awakenings, exacerbations, and rescue medicine use (Table 13). One trial also reported missed work, hospitalizations, and emergency visits^98–100 (Evidence Tables A and B). The results are mixed but show a trend favoring the BUD/FM MART for several outcomes. Both reported statistically significant differences in exacerbations favoring BUD/FM MART, but reported no difference in symptoms. One trial reported fewer nocturnal awakenings in both children and adults treated with BUD/FM MART.^{103, 105} The single study reporting hospitalizations and emergency visits found no difference between groups in the full population analysis^{98, 99} but a small but significant decrease in hospitalizations/emergency visits favoring BUD/FM MART among those age 16 and older.¹⁰⁰ The trial reporting missed work found a numerical difference favoring BUD/FM MART, but the statistical significance was not reported.^98–100

None of the trials reported any outcomes favoring the BUD/FM for maintenance and SABA for relief.

3. BUD/FM MART compared with FP/SM for maintenance and SABA for relief

We found two good-^{98–100, 104} and one fair-quality RCTs¹⁰⁶ comparing these treatments. All three trials reported asthma symptoms, exacerbations, and rescue medicine use (Evidence Tables A and B). Two trials reported nocturnal awakenings and hospitalizations or emergency visits.^{98–100, 104} One trial also reported missed work ^98–100 and two reported quality of life.^{98–100, 106}

The results are mixed but show a trend favoring BUD/FM MART for some outcomes. All three trials reported no difference in symptoms or nocturnal awakenings, but statistically significantly lower exacerbation rates in those treated with BUD/FM MART. Outcomes related to rescue medications use were mixed. One trial reported no difference in rescue medicine use or rescue-free days;¹⁰⁴ one reported no difference in rescue medicine use but a greater percentage of rescue-free days for those treated with FP/SM plus SABA for relief (56% compared with 59.1%, P < 0.05);^98–100 one reported less rescue medicine use for those treated with BUD/FM MA RT (0.58 puffs/day compared with 0.93, P < 0.001).¹⁰⁶ The trials reporting quality of life, and hospitalizations or emergency visits found no difference between treatment groups. The single trial reporting missed work found the lowest mean number of sick days in the FP/SM arm (2.36 per 6 months), the highest in the BUD/FM fixed-dose arm (3.11 per 6 months), and 2.48 days per 6 months in the MART arm, but the statistical significance was not reported.^98–100

Of note, the fair-quality trial¹⁰⁶ reduced the starting doses to levels that could be considered inadequate compared to the subjects’ previous doses. If randomized to FP/SM, subjects were stepping down in their level of control and did not have the possibility to adjust the dose for 4 weeks. The BUD/FM MART group could increase their dose with as needed BUD/FM. This initial possible under-treatment may have biased the study in favor of the BUD/FM MART group.

1. Inhaled Corticosteroids (ICSs) compared with Leukotriene Receptor Antagonists (LTRAs)

We conducted meta-analyses for six outcomes that were reported with sufficient data in multiple trials (Appendix I). Those treated with ICSs had a greater increase in the proportion of days free from rescue medication (SMD −0.25, 95% CI: −0.31, −0.19, 12 studies), greater reduction in rescue medicine use per day (SMD −0.23, 95% CI: −0.29, −0.17, 13 studies), greater increase in percent of symptom free days (SMD −0.21, 95% CI: −0.28, −0.15, 13 studies), greater improvement in symptom score (SMD −0.28, 95% CI: −0.34, −0.22, 10 studies), less frequent exacerbations (SMD −0.17, 95% CI: −0.22, −0.12, 13 studies), and a greater increase in quality of life (AQLQ scores; SMD −0.19, 95% CI: −0.27, −0.12, 7 studies) than those treated with leukotriene modifiers. For all six meta-analyses, sensitivity analyses indicate no difference in overall meta-analysis conclusions with any single study removed. In addition, there was no significant heterogeneity between studies (Appendix I).

When looking at montelukast alone compared with ICSs, our meta-analysis again shows that patients treated with ICSs had a greater increase in the proportion of days free from rescue medication use, greater reduction in rescue medicine use per day, greater increase in the proportion of symptom free days, greater improvement in symptom score, fewer exacerbations, and greater improvement in quality of life than those treated with montelukast (Appendix I).

When looking at zafirlukast alone compared with ICSs, our meta-analysis again shows that patients treated with ICSs had a greater increase of the proportion of days free from rescue medication use, greater increase of the proportion of symptom free days, greater change in symptom score, and fewer exacerbations than those treated with zafirlukast (Appendix I).

A previously published good quality systematic review included 18 RCTs (N = 3,757), 13 of which compared ICS therapy to ML therapy in children and adolescents 18 years and younger diagnosed with asthma at least 6 months prior to enrollment.¹⁰⁹ Six of the included trials also met our inclusion criteria^{125, 126, 129–132}; seven did not. Duration of studies varied but ranged from 4–12 weeks, 24–28 weeks, and 48–56 weeks, with one study being 112 weeks long. While most of the studies included patients age 6–18, one study included children younger than 6 (2–8 years) for which a nebulizer was used for ICS administration. Intervention drugs included oral montelukast (4 to 10 mg) compared to either inhaled BDP 200–400 mcg/day (0.5 mg nebulized), FP 200 mcg/day, BUD 200–800 mcg/day or TAA 400 mcg/day.

Seven trials (N = 2,429) contributed to the primary outcome, with ICS-treated patients showing a significantly lower risk of developing an exacerbation requiring systemic corticosteroids (RR 0.83, 95% CI: 0.72 – 0.96; NNT 24). However, no statistically significant difference was noted between groups with respect to withdrawals due to exacerbations (N = 680, RR 0.73, 95% CI: 0.36 – 1.48) and hospitalizations due to exacerbations (N = 533, RR 0.33, 95% CI: 0.03 – 3.15). Additional data were pooled based on secondary outcomes of interest and found ICS significantly improved mean change from baseline of symptom score (N = 575, SMD 0.18, 95% CI 0.01 – 0.34]), rescue inhaler use (puffs/24 hours: N = 1823, SMD 0.34 puffs/day, 95% CI 0.16 – 0.53]), and rescue-free days (N = 1904, SMD 0.16, 95% CI 0.07 – 0.25).

Another good quality systematic review with meta-analysis compared licensed doses of LTRAs with ICSs.¹⁰⁷ It included 3 trials testing a higher ICS dose; 3 trials testing a lower ICS dose; and the 21 remaining trials using equal nominal daily doses of ICS. It included 27 studies (9100 subjects); 3 of these in children and 24 in adults. Nine of these included trials also met our inclusion criteria.^{110–115, 118, 120–123} Eighteen of the included studies in this systematic review did not meet our inclusion/exclusion criteria. Duration of studies varied but ranged from 4–8 weeks, 12–16 weeks, and 24 to 37 weeks. The intervention drugs included montelukast (5 to 10 mg) and zafirlukast (20 mg twice daily). The ICS dose was uniform across 21 trials; seven of those used BDP 400 mcg/day, one used BDP 400–500 mcg/day, and 11 used FP 200 mcg/day. Three trials tested a high dose of ICS (BUD 800 mcg/day), one trial failed to report the dose used, and three trials used low dose BDP or equivalent. Eight trials enrolled patients who had mild asthma; 19 enrolled patients with moderate asthma; 3 trials did not report baseline FEV1.

Eighteen trials contributed to the primary outcome showing a 65% increased risk of exacerbations requiring systemic steroids for any LTRA (10 trials in montelukast and 5 trials in zafirlukast) compared to any ICS dosing regimen. The pediatric trials (3) could not be pooled due to a lack of exacerbations. However, 5 trials were pooled for exacerbations requiring hospitalization and there was no significant difference. Data at 12 weeks was pooled according to outcome and found ICS significantly improved change in symptom score (6 trials, SMD 0.29, 95% CI: 0.21 to 0.37), nocturnal awakenings (6 trials, SMD 0.21, 95% CI: 0.13 to 0.30), daily use of B2-agonists (6 trials, WMD 0.28 puffs/day, 95% CI: 0.20 to 0.36), symptom-free days (3 trials, WMD −12, 95% CI: −16 to −7), rescue-free days (3 trials, WMD −14%, 95% CI: −18, −10), and quality of life (2 trials, WMD −0.3, 95% CI: −0.4, −0.2). Similarly, ICS significantly improved asthma control days (3 trials, WMD −8 %, 95% CI: −15, −1]) and rescue-free days (2 trials, WMD −9%, 95% CI: −14, −03). LTRAs significantly increased the risk of withdrawal (19 trials, RR 1.3, 95% CI: 1.1, 1.6) which was attributable to poor asthma control (17 trials, RR 2.6, 95% CI: 2.0, 3.4).

A third and final fair-rated meta-analysis compared LTRAs to ICSs.¹⁰⁸ It included 6 studies (5278 subjects); 5 retrospective cohort studies and 1 prospective trial. None of these 6 studies met our inclusion criteria. The analysis included trials of subjects with a diagnosis of asthma, without restriction to severe asthma patients or children. Duration of trials was at least 6 months. The pooling of the 6 trials showed a significantly higher annual rate of emergency department visits in the LTRA group (P < 0.005). The rate of hospitalizations was shown to decrease significantly with the use of ICSs compared to LTRAs (2.23% compared with 4.3%; P < 0.05).

2. Fluticasone (FP) compared with Montelukast (ML)

We found 9 fair quality RCTs(10 articles) that compared ML with FP^{114–117, 125–130, 133} that met our inclusion criteria. Our meta-analyses of outcomes from these trials show that patients treated with FP had a greater increase in the proportion of days free from rescue medication use (SMD −0.25, 95% CI: −0.34, −0.16, 7 studies), greater reduction in rescue medicine use per day (SMD −0.25, 95% CI: −0.33, −0.16, 5 studies), greater increase in the proportion of symptom-free days (SMD −0.24, 95% CI: −0.32, −0.16, 6 studies), greater improvement in symptom score (SMD −0.24, 95% CI: −0.33, −0.14, 4 studies), fewer exacerbations (SMD −0.17, 95% CI: −0.26, −0.09, 6 studies), and greater improvement in quality of life (AQLQ scores: SMD −0.15, 95% CI: −0.25, −0.06, 4 studies) than those treated with ML (Appendix I).

Details of the characteristics of the 9 individual RCTs^{114–117, 125–130, 133} are summarized in Tables 14 and 15.

3. Beclomethasone (BDP) compared with Montelukast (ML)

Six fair quality RCTs^{110–113, 118, 124, 134} meeting our inclusion criteria compared montelukast with beclomethasone (Tables 14 and 15). Most of the outcomes reported favored BDP over ML or found no difference between groups. In general, the results comparing BDP with ML appear to be consistent with the overall results comparing ICSs with LTRAs. Our meta-analyses of outcomes using sufficient data from multiple trials shows that compared to ML-treated patients, those treated with BDP had fewer exacerbations (SMD −0.15, 95% CI: −0.30, −0.002), and trends toward a greater proportion of rescue free days (SMD −0.08, 95% CI: −0.19, −0.04) and a greater proportion of symptom-free days (SMD −0.11, 95% CI: −0.25, 0.02), neither of which reached statistical significance (Appendix I).

Details of the individual RCTs are summarized in Tables 14 and 15. The only trial enrolling children < 12 years of age was a fair-rated multinational, multi-center RCT in children (N = 360) comparing ML 5 mg/day (N = 120) compared with medium dose BDP 400 mcg/day (N = 119) compared with placebo (N = 121) for 56 weeks.¹²⁴ Subjects with mild persistent asthma, age 6.4 – 9.4 for boys and 6.4 – 8.4 for girls were enrolled worldwide (from most continents). The primary objective of the trial was to assess the effects of ML and BDP on linear growth, however some of our primary outcomes of interest were also reported. Fewer subjects treated with ML or BDP had asthma reported as an adverse experience compared to those treated with placebo, but the difference between groups was not statistically significant (36.7% compared with 42.9% compared with 50.4%, P = NS for ML compared with BDP). There were no statistically significant differences in the percentage of patients requiring oral steroids (25% compared with 23.5%), the percentage requiring more than one course of oral steroids (5.8% compared with 5.9%), or the percentage of days of b-agonist use (10.55% compared with 6.65%) between those treated with ML and those treated with BDP.

4. Budesonide (BUD) compared with Montelukast (ML)

We found three fair quality RCTs comparing BUD with ML^{119, 131, 132} that met our inclusion criteria (Tables 14 and 15). Too few studies reported sufficient data for meta-analysis of our included outcomes. Of the three RCTs, one enrolled adult populations, one¹³¹ enrolled children and adolescents ages 6–18, and one¹³² enrolled children ages 2–8. Most subjects in these trials had mild persistent asthma. Study duration ranged from 12 weeks to 52 weeks. The reported outcomes of interest were either not statistically significantly different between the two groups or favored BUD. For symptoms, two trials^{119, 131} reported no statistically significant difference between groups. Two trials reporting exacerbations found more favorable results for those treated with BUD than those treated with ML.^{119, 132} The single trial reporting quality of life found no difference between the treatments for overall quality of life measures.¹³²

5. Fluticasone (FP) compared with Zafirlukast

We found four fair quality RCTs comparing FP with zafirlukast^120–123 that met our inclusion criteria. All four trials show similar results favoring FP over zafirlukast for symptoms, rescue medicine use, and quality of life. Our meta-analyses again show that subjects treated with FP had a greater increase in days free from rescue medication use (SMD −0.30, 95% CI: −0.40, −0.20, 4 studies), greater increase of the proportion of symptom free days (SMD −0.29, 95% CI: −0.39, −0.19, 4 studies), greater improvement in symptom score (SMD −0.31, 95% CI: −0.41, −0.21, 4 studies), and fewer exacerbations (SMD 0.21, 95% CI: −0.31, −0.11, 4 studies) than those treated with zafirlukast (Appendix I).

1. ICS (any) compared with LABA (any) for monotherapy

We conducted meta-analyses for five outcomes that were reported with sufficient data in multiple similar trials (Appendix I). These included percentage improvement in symptom-free days, change in symptom scores, exacerbations, percentage improvement in rescue-free days, and change in rescue medicine use. We found no statistically significant differences in the percentage improvement in symptom-free days (SMD = 0.05; 95% CI = −0.10, 0.21; 7 studies), change in symptom scores (SMD = 0.14; 95% CI = −0.05, 0.34;6 studies), percentage improvement in rescue-free days (SMD = −0.14; 95% CI = −0.35, 0.07; P = 0.186; 5 studies), and change in rescue medicine use (as number of puffs per day) (SMD = 0.14; 95% CI = −0.11, 0.40; 7 studies). We found that those treated with LABAs had a significantly higher odds of experiencing an exacerbation than those treated with ICSs (OR = 2.8; 95% CI = 1.7, 4.9; 6 studies). The measure of statistical heterogeneity was high in the analysis of rescue puffs per day (I² 78.4). For all analyses except percentage of rescue free days, sensitivity analyses indicate no difference in overall meta-analysis conclusions with single studies removed. For the percent rescue free days analysis, removal of Lundback et al caused the difference between ICS and LABA to reach statistical significance (favoring LABA) (point estimate = −0251; 95% CI: −0.390, −0.113; P < 0.001).

2. Fluticasone (FP) compared with Salmeterol (SM)

Seven fair-quality RCTs compared FP with SM for monotherapy.^{135, 137–141, 143, 144, 150} None included children ≤ 12 years of age. All seven also included comparisons with an FP/SM combination product. Study duration was 12-weeks for six trials and 12 months for one.¹³⁷ Four compared SM with low-dose FP and three compared SM with medium-dose FP. Six of the seven were conducted in the United States; one was conducted in Sweden.¹³⁷

The majority of trials assessed asthma symptoms, nocturnal awakenings, exacerbations, and rescue medicine use. Two trials^{140, 143} reported quality of life. The majority of trials found no difference or a trend toward better outcomes in those treated with FP than those treated with SM (Evidence Tables A and B).

3. Beclomethasone (BDP) compared with Salmeterol (SM)

Three fair-quality RCTs compared BDP with SM.^147–149 One¹⁴⁷ enrolled adolescents and adults ≥ 12 years of age; the other two studies enrolled children and adolescents aged 6–14¹⁴⁸ or 6 –16.¹⁴⁹ Study duration ranged from 26 weeks to 12 months. All three compared SM with medium-dose BDP.

All three trials reported exacerbations and rescue medicine use; two reported symptoms^{147, 149} and nocturnal awakenings;^{147, 148} one reported missed school.¹⁴⁸ With the exception of one trial that reported greater improvement in the percentage of rescue-free days for those treated with SM (36% compared with 28%, P = 0.016),¹⁴⁷ all three trials reported no differences or better outcomes for those treated with BDP than for those treated with SM (Evidence Tables A).

4. Triamcinolone (TAA) compared with Salmeterol (SM)

One good-rated 16-week multicenter RCT^{145, 146} (SOCS Trial) compared TAA with SM in 164 adolescents and adults aged 12–65. The trial reported fewer exacerbations and a lower treatment failure rate for those treated with TAA, but no statistically significant difference in symptoms, rescue medicine use, or quality of life (Evidence Tables A).

5. Budesonide (BUD) compared with Formoterol (FM)

Two fair-rated 12-week multicenter RCTs^{136, 142} compared BUD with FM in adolescents and adults aged ≥ 12. The results showed trends toward fewer exacerbations and greater improvements in symptoms, nocturnal awakenings, and rescue medicine use for those treated with BUD (Evidence Tables A). Whether these trends were statistically significantly different was not reported (the studies focused on comparing BUD/FM with the other treatments).

1. Montelukast compared with Salmeterol

One fair-rated RCT (N = 191) compared ML 10 mg/day (N = 97) compared with SM 100 mcg/day (N = 94) as monotherapy for 8 weeks.¹⁵¹ Subjects with chronic asthma and evidence of exercise-induced bronchoconstriction age 15 to 45 were enrolled from multiple centers in the United States. The trial was designed to evaluate exercise-induced bronchoconstriction and most of the outcomes reported were intermediate outcomes that are not included in our report. The trial also reported mortality as an outcome, with no deaths in the ML group and one in the SM group (P = NR).

2. Montelukast compared with Eformoterol

One fair-rated cross-over RCT (N = 58) compared eformoterol 24 mcg/day with ML 10 mg/day (six weeks of treatment, one-week washout, six weeks of treatment with the other medication, one-week washout, then all subjects received fluticasone 500 mcg/day for six weeks).¹⁵² Subjects age 16 to 75 with mild to moderate persistent asthma previously treated with or without ICS were enrolled from multiple research centers in Australia. We only report results of the ML and eFM comparison because the fluticasone portion of the study does not have a comparison. Over the 12 weeks of treatment, subjects treated with eFM had fewer symptoms (percentage of symptom-free days: 23 compared with 0; P = 0.01; symptom scores: 1.2 compared with 1.6; P = 0.02), less rescue medicine use (percentage of rescue-free days: 40 compared with 30; P = 0.008), and better quality of life (QOL score: 0.4 compared with 0.6; P = 0.001) compared to those treated with ML.

1. ICS+LABA compared with ICS

The results of the 9 individual trials are described below under the appropriate drug comparisons. We conducted meta-analyses for outcomes that were reported with sufficient data in multiple trials (Appendix I). These included symptom-free days, symptom scores, rescue medicine-free days, and rescue medicine use (puffs/day). We found statistically significant differences favoring those treated with ICS+LABA for all four outcomes. Those treated with ICS+LABA had greater improvement in the percentage of symptom-free days (SMD = 0.24, 95% CI: 0.14, 0.33; 6 studies), symptom scores (SMD = 0.28, 95% CI: 0.15, 0.41; 4 studies), percentage of rescue medicine-free days (SMD 0.32, 95% CI 0.20, 0.43; 4 studies), and rescue medicine use (puffs per day) (SMD 0.25, 95% CI 0.12, 0.38; 7 studies) (Appendix I)

2. Fluticasone (FP)+Salmeterol (SM) compared with Fluticasone (FP)

Seven fair-quality RCTs^{138, 141, 154, 155, 158–160} (2896 subjects) compared FP+SM with FP alone (Table 18). All 7 compared the combination of FP and SM administered in a single inhaler with FP alone. Six of the studies used low dose FP; one used medium dose FP.¹⁵⁴ Five were 12-week trials and 2 were 24-week trials.^{155, 158} All were conducted in populations of ≥ 12 or 18 years of age.

All 7 trials reported outcome measures for symptoms and rescue medicine use, two trials reported nocturnal awakenings,^{138, 141} and 3 reported exacerbations. ^{155, 158, 160} Six trials reported greater improvements in symptoms for those treated with FP/SM combination products than for those treated with FP alone. Just one trial found no difference in symptoms.¹⁴¹ All 7 trials reported statistically significantly better outcomes for most measures of rescue medicine use (puffs/day, % of rescue-free days, % of rescue-free nights, episodes of use) for those treated with FP/SM. Just one trial reported no statistically significant difference for one of its measures of rescue medicine use, but there was a trend toward greater improvement for those treated with FP/SM (mean improvement in puffs/24 hours: −2.4 compared with −1.8).¹⁴¹ The trials reporting nocturnal awakenings and exacerbations found no difference between groups (Evidence Tables A and B).

3. Budesonide (BUD)+Formoterol (FM) compared with Budesonide (BUD)

Two fair-quality RCTs (1,036 subjects) compared BUD+FM with BUD alone.^{156, 157} Both compared BUD+FM administered in separate inhalers with low-dose BUD alone. One was a 12-week Russian trial that enrolled 338 adults.¹⁵⁶ The other was a 1 -year multinational trial that enrolled 1970 adolescents and adults ≥ 12 years of age.¹⁵⁷ The two trials reported some conflicting results. The 12-week trial reported better improvement in symptoms and rescue medicine use for subjects treated with BUD+FM, but no difference in quality of life. The 1-year trial reported no statistically significant differences between the two groups for symptoms, nocturnal awakenings, exacerbations, or rescue medicine use.

2. Fluticasone (FP) + Salmeterol (SM) compared with Fluticasone (FP)

Fourteen fair-quality RCTs (7,091 subjects) compared FP+SM with a higher dose of FP^{53, 127, 169–176, 195–197, 200}(Table 19). Eleven administered FP+SM in a single inhaler device^{127, 169–171, 173–175, 195–197, 200} and 3 tested the combination delivered by separate inhalers. Three studies ¹²⁷ included children ≤ 12 years of age. Study duration was 8 weeks for 1 trial, 12 weeks for 6 trials, 16 weeks for 2 trials, 24 weeks for 4 trials, and 52 weeks for 1 trial.

The majority of trials assessed asthma symptoms and rescue medicine use. Eight trials also reported exacerbations and two reported quality of life. For these outcomes, most of the trials either reported no difference or outcomes favoring FP+SM combination therapy over the increased dose of FP. One trial, comparing FP twice daily with FP/SM once daily, reported a statistically significant difference in favor of FP alone for mean daily asthma symptom score.¹⁹⁶ For subjects treated with FP+SM compared to those treated with FP alone, 7 trials reported fewer symptoms or better improvement in symptoms,^{169, 170, 172, 173, 175, 176, 200}9 trials reported a greater decrease or less frequent use of rescue medicine,^{53, 169–173, 176, 195, 200} one trial reported a trend toward fewer exacerbations,¹⁷⁰ and one trial reported greater improvement in nocturnal awakenings.¹⁷² The two trials reporting quality of life found no statistically significant difference in overall quality of life measures^{127, 175}(Evidence Tables A and B).

Meta-analyses of 8 trials show no statistically significant difference in exacerbations, but the pooled odds ratio favors those treated with FP+SM (OR = 0.86, 95% CI: 0.67, 1.1; 8 studies). Sensitivity analyses indicate that the removal of any one study does not change the overall conclusion. There was no significant heterogeneity between studies (I² = 0). Additional meta-analyses for symptom-free days, symptom scores, rescue-free days, and rescue medicine use show a trend toward results similar to those in the overall meta-analysis for ICS+LABA compared with higher dose ICS.

3. Budesonide (BUD) + Formoterol (FM) compared with Budesonide (BUD)

Seven fair quality RCTs (6,460 patients) compared BUD+FM with a higher dose of BUD^{103, 105, 157, 177–180, 198}(Table 19). Five administered BUD+FM in a single inhaler device^{103, 105, 177, 178} and two tested the combination delivered by separate inhalers. Two of the trials^{103, 105} included children ≤ 12 years of age. One enrolled children with mild to moderate persistent asthma between the ages of four and 11.¹⁰³ The other enrolled subjects with moderate persistent asthma between the ages of 4 and 80.¹⁰⁵ Study duration was 12 months for 6 trials and 12 weeks for one trial.¹⁷⁸

All trials assessed asthma symptoms, exacerbations, and rescue medicine use. Four trials also reported nocturnal awakenings. For these outcomes, the majority of trials reported no difference or outcomes favoring BUD+FM combination therapy. For subjects treated with BUD+FM compared to those treated with BUD alone, 5 of 6 trials reported fewer symptoms or better improvement in symptoms,^{103, 105, 178–180, 198} 1 trial (of five reporting) found greater reduction in nocturnal awakenings,¹⁷⁸ and 4 trials reported a greater decrease or less frequent use of rescue medicine.^{105, 178–180, 198} Four trials found no difference in exacerbations.^{103, 105, 177, 178} One study found that the number of asthma exacerbations per patient-treatment year was significantly lower with BUD+FM (0.185) compared with a higher dose of BUD alone (0.315) (P = 0.049). ¹⁹⁸ The remainder of trials reported no difference for these outcomes except for one trial reporting a trend toward fewer exacerbations in subjects treated with the increased dose of BUD than those treated with BUD+FM^{179, 180}.

Meta-analyses of 7 trials found trends consistent with the overall ICS+LABA compared with higher dose ICS meta-analyses. Subjects treated with BUD+FM had greater improvement in the percentage of symptom-free days (SMD = −0.19, 95% CI: −0.27, −0.11, 6 studies), greater improvement in symptom scores (SMD = −0.18, 95% CI: −0.28, −0.07; 2 studies), greater improvement in the percentage of rescue-free days (SMD = −0.21, 95% CI: −0.36, −0.05, 3 studies), and greater reduction in rescue medicine use (SMD = −0.16, 95% CI: −0.26, −0.06, 5 studies) than those treated with a higher dose BUD alone. There was no statistically significant difference in exacerbations (OR = 0.98, 95% CI: 0.72, 1.34, 5 studies).

4. Beclomethasone (BDP) + Salmeterol (SM) compared with Beclomethasone (BDP)

Six fair quality RCTs (2,574 subjects) compared BDP+SM with a higher dose of BDP^181–187 (Table 19). All six administered BDP+SM in separate inhalers. One trial¹⁸⁵ enrolled children and adolescents between the ages of four and 18. The remainder were conducted in populations ≥ 12 years of age. Study duration was 12 weeks for one trial,¹⁸⁶21 –24 weeks for four,^{181–184, 187} and one year for one.¹⁸⁵

All trials assessed asthma symptoms, exacerbations, and rescue medicine use. Four trials also reported nocturnal awakenings and two reported quality of life outcomes. For each of these outcomes, the majority of trials reported no difference or outcomes favoring BDP+SM combination therapy; none reported a statistically significantly greater improvment for those treated with BDP alone. For symptoms, three trials reported no difference^{181, 182, 185, 186} and three found results favoring BDP+SM.^{183, 184, 187} For nocturnal awakenings, one trial reported no difference¹⁸⁴ and three found results favoring BDP+SM.^{181–183, 187} For exacerbations, five trials reported no difference^{181–184, 186, 187} and one reported a trend toward fewer exacerbations requiring steroids for those treated with BDP alone.¹⁸⁵ All but one trial ^{181, 182} reported a greater decrease or less frequent use of rescue medicine for those treated with BDP+SM than for those treated with BDP alone. The two trials reporting quality of life found no significant difference between the groups^{181, 182, 186}.

Meta-analyses of these six trials showed trends consistent with the overall ICS+LABA compared with higher dose ICS meta-analyses. Subjects treated with BDP+SM had statistically significantly greater reduction in rescue medicine use (SMD = 0.18, 95% CI: 0.05, 0.31; 4 studies) and trended toward greater improvement in the percentage of symptom-free days (SMD = 0.14, 95% CI: −0.01, 0.28; 2 studies) than those treated with a higher dose BDP alone. There was no statistically significant difference in the percentage of subjects with exacerbations (SMD = −0.019, 95% CI: −0.095, 0.058; 5 studies contributing 6 comparisons).

5. Beclomethasone (BDP) + Formoterol (FM) compared with Beclomethasone (BDP)

Three fair RCTs (982 subjects) meeting our inclusion/exclusion criteria compared BDP+FM with a higher dose of BDP alone.^{188, 189, 199} All 3 enrolled adults ≥ 18 that were not controlled on ICSs. Two compared BDP+FM in a single inhaler device¹⁸⁸ and one tested the combination delivered by separate inhalers.¹⁸⁹ Two studies^{188, 189} reported statistically significantly better symptom and rescue medicine use outcomes for subjects treated with BDP+FM than those treated with FM alone (Evidence Tables A and B). Huchon et al.¹⁹⁹ reported that a reduction in rescue medication use was statistically significant from baseline for the BDP+FM group and did not change for the BDP alone group, but did not report whether the difference between the groups was significant. Two studies found a trend toward fewer exacerbations in those treated with BDP+FM.^{189, 199}

6. Fluticasone (FP) + Salmeterol (SM) compared with Budesonide (BUD)

One good 12-week RCT (N = 349)¹⁹² and one fair 24-week RCT (N = 353)^{190, 191} meeting our inclusion/exclusion criteria compared FP+SM with a higher relative dose of BUD alone. The 12-week trial compared FP/SM (200/100) with BUD (800) and the 24-week trial compared FP/SM (500/100) with BUD (1600). Both were multinational trials that enrolled subjects ≥ 12 years of age. Both administered FP/SM in a single inhaler device. The two trials reported some conflicting results. The 12-week trial found no statistically significant difference between treatment groups in symptoms, exacerbations, or rescue medicine use. The 24-week trial reported fewer symptoms, less rescue medicine use, and greater improvement in quality of life for those treated with FP+SM than those treated with BUD alone, but no significant difference in exacerbations.

7. Budesonide (BUD) + Formoterol (FM) compared with Fluticasone (FP)

One 12-week fair RCT meeting our inclusion/exclusion criteria compared BUD+FM in a single inhaler with a higher relative dose of FP alone in 344 adults with moderate persistent asthma.¹⁹³ The trial reported no statistically significant difference in symptoms or nocturnal awakenings. But, those treated with BUD+FM had fewer exacerbations and required less rescue medicine compared to those treated with FP alone.

8. Fluticasone (FP) + Salmeterol (SM) compared with Triamcinolone (TAA)

We found one fair RCT meeting our inclusion/exclusion criteria that compared FP+SM (in separate inhalers) with a higher relative dose of TAA alone.⁵³ This trial is also included above in this section for the FP+SM compared with FP comparison because there was an FP-only arm as well. It enrolled 680 adults and adolescents ≥ 12 years of age with persistent asthma not adequately controlled on ICS. They reported greater improvement in symptoms, nocturnal awakenings, and rescue medicine use for those treated with FP+SM than for those treated with TAA alone.

1. ICS+LABA compared with ICS (same dose)

We conducted meta-analyses for five outcomes that were reported with sufficient data using similar measures in multiple trials (Appendix I). Those treated with ICS+LABA had a greater increase in the proportion of days free from rescue medication (SMD 0.31, 95% CI: 0.25, 0.37, 20 comparisons), greater reduction in rescue medicine use per day (SMD −0.29, 95% CI: −0.36, −0.23, 21 comparisons), greater increase in percentage of symptom free days (SMD 0.27, 95% CI: 0.22, 0.32, 25 comparisons), greater improvement in symptom score (SMD −0.27, 95% CI: −0.33, −0.21, 17 comparisons), and a greater increase in quality of life (AQLQ scores; SMD 0.26, 95% CI: 0.14, 0.37, 7 comparisons) than those treated with ICS alone.

One previously published good systematic review²⁰³ compared the addition of any LABA to any ICS (ICS+LABA) with continuing the same dose of ICS. The review included 77 trials (N = 21,248 with 16,623 adults and 4,625 children) that contributed information. Trial duration ranged from 4 to 54 weeks. Most studies (N = 43) were 12 to 16 weeks. Twenty-seven trials examined ICSs+LABAs delivered via a single device. The systematic review reported that the addition of a LABA to an ICS reduced the risk of exacerbations requiring systemic steroids by 23% (RR 0.77, 95% CI: 0.68 to 0.87) compared to ICS alone. In addition, the addition of LABA resulted in greater improvement in symptoms, rescue medicine use, and quality of life. They found no difference in nocturnal awakenings.

2. Budesonide (BUD) + Formoterol (FM) compared with Budesonide (BUD)

Two good^{207, 217} and 14 fair RCTs^{136, 142, 157, 179, 198, 206, 210–213, 215, 219, 221, 222} (9,298 subjects total) compared the addition of FM to BUD with continuing the same dose of BUD (Table 20). One of these trials reported using eformoterol (eFM).²¹³ Eight trials administered BUD+FM in a single inhaler device,^{136, 198, 206, 211, 215, 219, 221, 222} three tested the combination delivered by separate inhalers,^{157, 179, 213} and five administered them both as a single inhaler and in separate inhalers to different study groups.^{142, 207, 210, 212, 217}

Five trials included children ≤ 12 years of age.^{212, 215, 219, 221, 222} Study duration was 12 weeks for 11 trials, 26 weeks for 1 trial,²²² 32 weeks for one trial,²¹³ and one year for three trials.^{157, 179, 198}

The majority of trials assessed asthma symptoms, nocturnal awakenings, exacerbations, and rescue medicine use. Six trials also assessed quality of life and one assessed missed work or school. For these outcomes, all trials either reported no difference or outcomes favoring BUD+FM combination therapy over the same dose of BUD. No trial reported a statistically significant difference in favor of BUD alone for any of these outcomes. For subjects treated with BUD+FM compared to those treated with BUD alone, 10 trials (71%) reported fewer symptoms or better improvement in symptoms,^{135, 137, 139, 142–144, 157, 173, 179, 180, 185, 198, 204–211, 213, 214, 216–218} six trials (of seven reporting the outcome) reported fewer exacerbations or a lower risk exacerbations,^{136, 157, 179, 206, 213, 215} and 10 trials (71%) reported a greater decrease or less frequent use of rescue medicine.^{135, 137, 139, 143, 144, 157, 173, 179, 180, 185, 204–211, 213–218, 221} For three of the eleven trials reporting nocturnal awakenings, results favored the BUD+FM group.^{206, 207, 211} The other eight reported no difference.^{136, 142, 157, 210, 212, 215, 217, 219} Three ^{212, 213, 219} of the four trials reporting quality of life found no statistically significant difference in overall quality of life measures and one²¹¹ reported greater improvement in those treated with BUD+FM. The single trial reporting missed work or school found no significant difference between groups.²¹³

3. Fluticasone (FP)+Salmeterol (SM) compared with Fluticasone (FP)

Nine fair quality RCTs (3,029 subjects) compared the addition of SM to FP with continuing the same dose of FP^{135, 137, 139, 143, 173, 204, 209, 220, 223} (Table 20). All 9 administered FP+SM in a single inhaler device.^{135, 137, 139, 143, 173, 204, 209, 220, 223} None tested the combination delivered by separate inhalers. One trial included children ≤ 12 years of age.²²⁰ Study duration was 12 weeks for 5 trials,^{135, 139, 143, 204, 220} 24 weeks for one trial,¹⁷³ and 12 months for 3 trials.^{137, 209, 223}

The majority of trials assessed asthma symptoms, exacerbations, and rescue medicine use. Three trials also reported nocturnal awakenings and one reported quality of life. For these outcomes, all trials either reported no difference or outcomes favoring FP+SM combination therapy over the same dose of FP. No trial reported a statistically significant difference in favor of FP alone for any of these outcomes. For subjects treated with FP+SM compared to those treated with FP alone, five trials (71%) reported fewer symptoms or better improvement in symptoms,^{135, 143, 173, 204, 209} three trials (of five reporting) reported fewer patients having exacerbations or withdrawn due to exacerbations,^{135, 137, 143} and 6 trials (86%) reported a greater decrease or less frequent use of rescue medicine.^{135, 139, 143, 173, 204, 209} Two of the three trials reporting nocturnal awakenings found no difference between groups,^{135, 139} one reported a higher percentage of awakening-free nights for the FP+SM group.¹⁴³ The single trial reporting quality of life measures reported a trend toward better scores on the activities limitation domain of the AQLQ, but no difference in other domains (activities limitation: 1.0 compared with 0.62, P = NR).¹⁴³

4. ICS+Salmeterol (SM) compared with ICS

Three fair quality RCTs (835 subjects) compared the addition of SM to any ICS with continuing the same dose of ICS (plus placebo)^{205, 208, 214} (Table 20). All three administered ICS+SM by separate inhalers. One trial included children, enrolling 210 subjects between the ages of 4 and 16.²¹⁴ Study duration was 12 weeks for two trials^{205, 214} and 14 weeks for one.²⁰⁸

All three trials reported symptoms and rescue medicine use, one reported exacerbations,²⁰⁵ and one reported quality of life measures.²⁰⁸ In all three trials, those treated with ICS+SM had greater improvements in symptoms (in one trial the difference was only statistically significant for nighttime symptoms)²⁰⁵ and rescue medicine use. The single trial reporting exacerbations found no statistically significant difference in the number of patients requiring a course of oral steroids (19 compared with 15, P = 0.19).²⁰⁵ The trial reporting quality of life found no statistically significant difference in overall quality of life, but there was a trend toward greater improvement in the ICS+SM group (AQLQ global score, mean change from baseline: 1.08 compared with 0.61, P = 0.47).²⁰⁸

5. ICS+Formoterol (FM) compared with ICS

Two fair quality RCTs (541 subjects) compared the addition of FM to any ICS with continuing the same dose of ICS (plus placebo)^{216, 218}(Table 20). Both administered ICS+FM by separate inhalers. One was a 6 month trial that enrolled 239 adults with mild to moderate persistent asthma that were not adequately controlled on ICSs.²¹⁶ The other was a 12 -week trial that enrolled 302 children (ages 6–11) not adequately controlled on ICSs.²¹⁸ The 6 month trial in adults found greater improvement in symptoms and rescue medicine use in those treated with ICS+FM, but no difference in exacerbations.²¹⁶ The 12-week trial in children found no statistically significant difference in symptoms, rescue medicine use, or quality of life.²¹⁸

6. Beclomethasone (BDP) + Salmeterol (SM) compared with Beclomethasone (BDP)

One 12-month fair quality RCT meeting our inclusion/exclusion criteria compared BDP+SM in a separate inhalers with the same dose of BDP alone in 177 children and adolescents (age 6–16) with mild to moderate persistent asthma.¹⁸⁵ The trial reported no statistically significant difference in symptoms, exacerbations, or rescue medicine use.

7. Beclomethasone (BDP) + Formoterol(FM) compared with Beclomethasone (BDP)

One 24-week fair quality RCT meeting our inclusion criteria compared BDP+FM in separate inhalers with same dose of BDP alone in 645 patients with moderate to severe asthma uncontrolled by regular treatment. The results did not provide between group differences for this comparison. Analyses were focused on the comparison of BDP+FM in a single inhaler with BDP+FM in separate inhalers and with a higher dose of BDP alone.

1. ICS+LTRA compared with ICS

Of the two systematic reviews meeting our inclusion criteria, one²²⁷ identified just three studies comparing ICS+LTRA with ICS that used constant doses of ICS. It did not find three others that we identified.^{197, 228, 229, 231} Thus, we do not discuss this review further in this section and we do not include it in our overall assessment of the evidence or our strength of evidence grades as it is missing about half of trials relevant to this section.

One good systematic review with meta-analysis²²⁶ compared LTRA plus ICS with the same dose of ICS, same dose of ICS with taper, or increased doses of ICS. The systematic review included 27 studies (5871 subjects); two of the studies were in children and 25 were in adults. Sixteen of the 27 trials reported data in a way that allowed meta-analysis. Three of these included trials met our inclusion criteria.^{118, 228–230} Many were excluded for wrong medication (pranlukast) or short duration (less than six weeks). Thirteen of the studies (two in children) compared an LTRA plus an ICS with the same doses of an ICS; seven studies compared an LTRA plus an ICS with increased doses of an ICS; and seven studies compared an LTRA plus an ICS with the same doses of ICS with tapering. The LTRAs included montelukast, zafirlukast, and pranlukast. Many trials used higher than licensed doses of LTRAs. Most trials used BDP with a dosing range from low (≤ 400 mcg/day BDP or equivalent) to high (> 800 mcg/day BDP or equivalent) potency, with each trial ensuring same ICS dosing for both groups.

ICS+LTRA compared with same dose ICS. For ICS plus LTRA compared with the same dose of ICS, the systematic review reported a non-significant reduction in the risk of exacerbations requiring systemic steroids (RR 0.64, 95% CI: 0.38 to 1.07), the primary outcome. Just four trials using licensed doses of LTRAs contributed data to the primary outcomes. The systematic review found no significant difference in symptom score (WMD = −0.10, 95% CI: −0.24, 0.03) or nocturnal awakenings (WMD −6.25, 95% CI: −12.72, 0.23). Higher than licensed doses of LTRA did show a significant difference in improvement from baseline in asthma symptom scores (SMD= −0.46, 95% CI: −0.25, −0.66). Those treated with both licensed and higher than licensed doses of LTRAs had a significant decrease in beta-agonists use compared to those treated with same dose ICSs (SMD −0.15, 95% CI: −0.24, −0.05 and SMD −0.43, 95% CI: −0.22, −0.63). There was no significant difference in quality of life (WMD 0.08, 95% CI: −0.03, 0.20).

ICS+LTRA compared with increased ICS. For ICS plus LTRA compared with increased doses of ICS, only 3 of the trials included in the systematic review compared licensed doses of LTRAs with increasing the dose of ICSs. The meta-analyses found no significant difference in any outcomes including the following: change from baseline in symptoms score with licensed (WMD 0.01, 95% CI: −0.09, 0.10) or higher than licensed doses of LTRA (WMD −0.06, 95% CI: −0.16, 0.03); risk of experiencing an asthma exacerbation requiring systemic steroids with licensed doses (RR 0.92, 95% CI: 0.56, 1.51) or higher than licensed doses of LTRA (RR 1.05 95% CI: 0.55, 2.00); withdrawals due to poor asthma control with licensed (RR 0.49, 95% CI: 0.15, 1.63) or higher than licensed doses of LTRA (RR 0.72 95% CI: 0.29, 1.76); and change from baseline in use of rescue beta-agonists with licensed (WMD −0.03 95% CI: −0.24, 0.18) nor higher than licensed doses of LTRA (WMD 0.00 95% CI: −0.37, 0.37).

ICS+LTRA compared with same ICS (tapering). For ICS plus LTRA compared with the same ICS dose with tapering (seven studies), the systematic review found no significant difference in final symptom scores (WMD −0.06, 95% CI: −0.17 to 0.05), number of patients with exacerbations requiring systemic steroids (RR 0.47, 95% CI: 0.20, 1.09), difference in final beta-agonist use (WMD −0.2 puffs/day, 95% CI: −0.7 to 0.3), or change from baseline in beta-agonist use (WMD −0.15 puffs/week; 95% CI: −0.91, 0.61). There was a significant reduction in rate of withdrawals due to poor asthma control for those treated with ICS plus LTRA (RR 0.63, 95% CI: 0.42 to 0.95), however this was not significant when only the trials using intention to treat analysis were considered (RR 0.63, 95% CI: 0.42, 0.95).

2. Budesonide (BUD)+ Montelukast (ML) compared with Budesonide (BUD) same dose

We found one fair RCT²³⁰ comparing the combination of BUD+ML with the same dose of BUD (Table 21). This fair-rated RCT (N = 639), the CASIOPEA study, compared low to high dose BUD (400 to 1600 mcg/day) plus placebo (N = 313) with low to high dose BUD (400 to 1600 mcg/day) + ML 10 mg/day (N = 326) for 16 weeks.²³⁰ Subjects age 18 to 70 with poorly controlled mild to severe asthma currently being treated with a stable dose of ICS for at least 8 weeks were enrolled from hospital centers in Spain. At endpoint, there were no statistically significant differences in asthma symptom scores or quality of life. However, those treated with BUD+ML had fewer nocturnal awakenings, more asthma free days, fewer days with exacerbations, and greater decrease in rescue medicine use. The differences were reportedly independent of BUD dose.

3. Beclomethasone (BDP) + Montelukast (ML) compared to Beclomethasone (BDP) same dose

We found one trial (N = 642) which compared four treatments for 16 weeks:¹¹⁸ low dose BDP (400 mcg/day) + ML (10 mg/day) (N = 193) compared with low dose BDP 400 mcg/day (N = 200) compared with ML 10mg/day (N = 201) compared with placebo (N = 48). Subjects with uncontrolled mild to moderate asthma treated with ICS who were age 15 or greater were enrolled from 18 countries and 70 different centers. At endpoint, those treated with BDP+ML had greater improvement in daytime asthma symptom scores (−0.13 compared with −0.02; P = 0.041), nights per week with awakenings (−1.04 compared with −0.45; P = 0.01), and percentage of days with an exacerbation (13.37% compared with 17.92%; P = 0.041) compared to BDP. BDP+ML showed no significant difference in % of patients with an asthma attack or difference in total puffs/day compared to BDP. Compliance was high with both inhaled and oral groups respectively.

4. Budesonide (BUD)+ Montelukast (ML) compared with Budesonide (BUD) increased dose

We found two fair RCTs^{228, 229, 231} comparing the combination of BUD+ML with an increased dose of BUD (Table 21). One fair multinational trial (N = 889) compared medium dose BUD (800 mcg/day) plus ML (10 mg/day) (N = 448) compared with high dose BUD (1600 mcg/day) (N = 441) for 16 weeks.^{228, 229} The trial enrolled subjects age 15 to 75 with uncontrolled asthma treated with medium dose ICS. At endpoint, there were no statistically significant differences between those treated with BUD+ML and those treated with BUD for percentage of asthma free days, daytime symptom score, percentage of nights with awakenings, percentage of days with an exacerbation, percentage of patients requiring oral steroids or hospitalization, rescue medicine use, or quality of life. Adherence was high for both the tablets and inhalers, with over 95% of days fully compliant.

The other trial²³¹ (N = 71) compared low dose BUD (400 mcg/day) (N = 33) compared with low dose BUD (200 mcg/day) plus ML (5 mg/day) (N = 30) for 12 weeks. Subjects with moderate persistent asthma age 6 to 14 were enrolled from a Pediatric Asthma Clinic in India. At endpoint, those treated with increased dose of BUD had fewer exacerbations compared to BUD+ML (9.1% compared with 33.3%; P < 0.01). Adherence was high in both groups with only one patient declaring non-adherence.

5. Fluticasone (FP)+Montelukast (ML) compared with Fluticasone (FP) increased dose

We found one fair RCT¹⁹⁷ (N = 182) comparing the combination of FP+ML with an increased dose of FP in children and adolescents (6 to 17 years of age). The trial used a triple cross-over design. Subjects with uncontrolled asthma while receiving FP (100 twice daily) were randomized to FP (250 twice daily), FP (100 twice daily) plus salmeterol, or FP (100 twice daily) plus montelukast. The primary outcome was a composite of exacerbations, number of asthma control days, and FEV1. One hospitalization for asthma-related symptoms occurred in each of the three treatment groups. A total of 120 prednisone bursts were prescribed for exacerbations (47 during treatment with FP compared with 43 during treatment with FP+ML, P= NR).

1. Fluticasone (FP)+Salmeterol (SM) compared with Montelukast (ML)

The 5 included studies are described below. We conducted meta-analyses for outcomes that were reported with sufficient data in multiple trials (Appendix I). These included symptom-free days, rescue medicine-free days, and exacerbations. We found statistically significant differences favoring those treated with FP/SMfor all three outcomes. Those treated with FP/SM had greater improvement in the percentage of symptom-free days SMD −0.25, 95% CI: −0.35, −0.15), greater improvement in the percentage of rescue medicine-free days (SMD −0.27, 95% CI: −0.37, −0.17), and fewer exacerbations (SMD 0.26, 95% CI: 0.16, 0.35). (Appendix I)

The 5 studies included one good quality RCT²³² and 4 fair quality RCTs (Table 22).^{127, 128, 233, 234} The good-rated RCT (N = 432) compared low dose FP/SM (200 mcg/100 mcg daily) (N = 216) compared with ML (10 mg/day) (N = 216) as monotherapy for 12 weeks.²³² Subjects with uncontrolled asthma treated with oral or inhaled short-acting beta-agonist age 15 and older were enrolled from 51 different centers in the United States. At endpoint those treated with FP/SM showed a greater improvement in all outcomes compared to ML including a decrease in the combined asthma symptom score (−1 compared with −0.7; P ≤ 0.001), increase from baseline in % symptom free days (+40.3% compared with +27%; P ≤ 0.001), increase from baseline in % of awakening free nights (+29.8% compared with +19.6%; P = 0.011), decrease from baseline in nights/week with awakenings (−2.2 compared with −1.6; P ≤ 0.001), decrease in puffs/day (−3.6 compared with −2.2; P ≤ 0.001), increase in % of rescue free days (53.4% compared with 26.7%; P ≤ 0.001), and increase in quality of life (AQLQ overall score, increase: 1.7 compared with 1.2; P < 0.001). Exacerbations occurred less frequently in the FP/SM group (3% compared with 6%; P = NR). Compliance was approximately 99% in both groups.

The first fair-rated RCT (N = 423) also compared low dose FP/SM (200 mcg/100mcg daily) (N = 211) compared with ML (10mg/day) (N = 212) for 12 weeks.²³³ Subjects with uncontrolled asthma treated with oral or inhaled short-acting beta-agonist age 15 or older were enrolled from multiple centers in the United States. At endpoint, results were similar to those in the good quality RCT described above²³² with significant differences for all outcomes favoring FP/SM over ML: including decrease in symptoms, rescue medicine use, and exacerbations (0%, 5%; P < 0.001) (Table 22).

A third fair-rated RCTs showed mixed results, with some outcomes favoring FP/SM and others finding no difference. The first (N = 500) compared low dose FP (200 mcg/day) (N = 169) compared with low dose FP (100 mcg/day) plus SM (50 mcg/day) (delivered once daily at night) (N = 165) compared with ML (5–10 mg/day) (N = 166) for 16 weeks.¹²⁷ Subjects were age six and older, had mild to moderate asthma controlled on ICS, and were enrolled from multiple American Lung Association Asthma Clinical Research Centers in the United States. At endpoint, there were no significant differences between FP plus SM and ML in symptom-free days or rescue medicine use. But, there were significant differences in the percentage of patients with treatment failure (20.4% compared with 30.3%; P = 0.03) and asthma control (ACQ: 0.71 compared with 0.82; P = 0.004) favoring FP plus SM. Adherence was good for all groups (FP/SM 93.3% compared with ML 90.5%). The last fair-rated RCT (N = 285), the Pediatric Asthma Controller Trial (PACT), compared low dose FP 200 mcg/day via DPI (N = 96) compared with ML 5 mg/day (N = 95) compared with low dose FP 100 mcg/day plus SM 100 mcg/day via DPI (FP 100 mcg plus SM 50 mcg in the morning plus SM 50 mcg in the evening) (N = 94) for 48 weeks.¹²⁸ Of note, the dose of FP/SM used was outside of the product label recommendation. Subjects with mild to moderate asthma age 6 to 14 were enrolled from Childhood Asthma Research and Education Centers in the United States. At endpoint, the trial found no significant difference in the overall percentage of asthma control days (52.5% compared with 59.6%; P = 0.08), but found favorable results for FP/SM in the change in the percentage of asthma control days from baseline (33.3% compared with 22.3%; P = 0.011). There was no significant difference in asthma control as measured by change in ACQ score from baseline (−0.45 compared with 0.55; P = 0.42). Adherence was similar between groups (86% compared with 90%; P = NR).

A final RCT showing mixed results, known as the Pediatric Asthma Control Evaluation (PEACE) study, enrolled children age 6 to 14 with mild to moderate persistent asthma in outpatient centers at 4 sites in Turkey and 23 in Latin America.²³⁴ Using a double -blind, double-dummy design, 281 children treated with FP/SM 100mcg/50mcg twice daily were compared to 267 patients treated with ML 5mg daily. While the results showed significant improvement in the percentage of symptom free days (OR 1.74, 95% CI 1.07 – 2.82), asthma controlled weeks (16.7% more in FP/SM group, 95% CI 8.3 – 16.7), they found no difference between groups in the percentage of nights without awakenings due to nocturnal symptoms (OR 2.33, 95% CI 0.73 – 7.47). The mean exacerbation rate and time was significantly reduced with FP/SM therapy (0.12 vs. 0.3, OR 0.4, 95% CI 0.29 – 0.57) and the number of patients exacerbation free at 84 days was 89.6% in FP/SM patients compared with 74.8% in the ML group (95% CI 8 – 22). In addition, the percentage of rescue free days increased significantly with FP/SM treatment (OR 3.24, 95% CI 2.09– 5.02). Quality of life measures, however, demonstrated mixed results. While PACQLQ scores were higher in the FP/SM group (mean treatment difference 0.54, 95% CI 0.06 – 1.02), no difference was noted between groups with respect to PAQLQ score (mean treatment difference 0.09, 95% CI −0.12 – 0.30). Finally, while 7.5% of FP/SM treated patients required some form of unscheduled health care contact during the study period, substantially more patients on ML therapy required medical attention (P= NR). Adherence was similar between groups (87% compared with 84%; P = NR).

1. ICS+LABA compared with ICS+LTRA

One good quality systematic review with meta-analysis including 6,030 subjects (11 of 15 included trials contributed to the analyses) compared LABAs with LTRAs as add-on therapy to ICSs.²³⁵ The included trials compared salmeterol (100 mcg/day) or formoterol (24 mcg/day) plus ICS compared with montelukast (10 mg/day) or zafirlukast (40 mg/day) plus ICS. The ICS dose average was 400 to 560 mcg/day of beclomethasone or equivalent.²³⁵ Of the fifteen trials the met inclusion criteria, a total of 80 subjects were children. Of the 11 trials that contributed to the analyses, 10 were in adults and one was in children. Six of the included trials met our inclusion criteria.^{236–239, 241, 242} Five of the studies included in the analysis did not meet our inclusion criteria.

The systematic review included randomized controlled trials conducted in adults or children with persistent asthma where a LABA or LTRA was added to ICS for 4 to 48 weeks. Inhaled Short-Acting Beta-2 Agonists and short courses of oral steroids were permitted as rescue medications. Subjects had to be on a stable dose of ICSs throughout the trials.

The meta-analysis reported that LABA plus ICS was significantly better than LTRA plus ICS for all observed outcomes.²³⁵ Six trials contributed to the primary outcome showing a significant decrease in risk of exacerbation requiring systemic steroids for those treated with LABAs (RR 0.83; 95% CI: 0.71, 0.97). The type of LTRA used did not impact the results. The reported number of patients who must be treated with the combination of LABA and ICS instead of LTRA and ICS to prevent one exacerbation over 48 weeks was 38 (95% CI: 23, 247).

Subjects treated with LABA+ICS had greater improvement in the percentage of symptom-free days (WMD 6.75%; 95% CI: 3.11, 10.39, 5 studies), daytime symptom scores (SMD −0.18; 95% CI: −0.25, −0.12, 5 studies), nighttime awakenings (WMD −0.12; 95% CI: −0.19, −0.06, 4 studies), percentage of rescue-free days (WMD 8.96%; 95% CI: 4.39, 13.53, 4 studies), rescue medication use per day (WMD −0.49 puffs/day; 95% CI: −0.75, −0.24, 7 studies), overall asthma-related quality of life (WMD 0.11; 95% CI: 0.05, 0.17, 3 studies). There was significant heterogeneity in one of the analyses (percentage of rescue-free days; I2 = 61%; P < 0.05).

The eight RCTs meeting the inclusion/exclusion criteria for our review are summarized in Table 23. Six of the eight trials were included in the systematic review with meta-analysis²³⁵ described above. One of those not included was a fair-rated RCT,²⁴⁰ the SOLTA study. It compared low dose FP (200 mcg/day) plus SM (100 mcg/day) (N = 33) with low dose FP (200 mcg/day) plus ML 10 mg/day (N = 33) for 12 weeks in 66 adults (age 18 to 50) with uncontrolled mild to moderate asthma. The ICS/LABA combination was delivered via a single inhaler. Patients being treated with medium dose ICSs were enrolled from multiple centers in the United Kingdom. At endpoint, there were no statistically significant differences in asthma symptoms, but the trends in direction of the effect sizes favored the ICS/LABA combination (symptoms-free days: mean difference in change from baseline: 13.2%, 95% CI: −1.9%, −32.9%; P = 0.064; symptom-free nights: mean difference in change from baseline: 13.3%, 95% CI: −1.5%, −34.5%; P = 0.055). There was no significant difference in daytime rescue use (median % rescue free days at endpoint 73% compared with 70%; P = NS), but there was a difference in rescue use at night favoring FP/SM (median rescue free nights at endpoint: 93% compared with 82%; P = 0.01).

The other trial (BADGER) not included in the systematic review described above enrolled 182 children and adolescents (6 to 17 years of age).¹⁹⁷ The trial used a triple cross-over design. Subjects with uncontrolled asthma while receiving FP (100 twice daily) were randomized to FP (250 twice daily), FP (100 twice daily) plus salmeterol, or FP (100 twice daily) plus montelukast for 16 weeks of each treatment (total of 48 week treatment phase). The primary outcome was a composite of exacerbations, number of asthma control days, and FEV1. The response to LABA step-up therapy was most likely to be the best response compared with LTRA step-up (relative probability, 1.6; 95% CI: 1.1 to 2.3). One hospitalization for asthma-related symptoms occurred in each of the three treatment groups. A total of 120 prednisone bursts were prescribed for exacerbations (30 during treatment with FP+SM compared with 43 during treatment with FP+ML, P= NR).

We do not describe all of the other included RCTs in detail because they generally found results consistent with the overall conclusions of the meta-analysis. For all of our outcomes of interest, most trials reported favorable results for subjects treated with ICS+LABA; the others reported no statistically significant differences (Evidence Tables A and B).

I. Bone density/osteoporosis

We found two fair quality systematic reviews with meta-analyses that studied the effect of ICSs on markers of bone function and metabolism.^{244, 245} One included 14 studies (2,302 subjects) of patients with asthma or COPD (both RCTs and prospective cohort studies) assessing BMD.²⁴⁴ The other included six studies of asthmatic subjects with median duration of ICS use of at least three years.²⁴⁵ Pooled results from both meta-anlyses showed no statistically significant difference in BMD between patients taking ICSs and controls. The one that included patients with asthma and COPD reported that asthma patients treated with ICSs showed a slight increase in BMD (0.13%) whereas COPD patients showed a slight decrease (−0.42%); however, neither change was statistically significant.²⁴⁴

Our review includes nine studies: three of the trials^{251, 252, 259} in the systematic reviews, as well as six additional studies. ^{253, 255, 256, 260–262, 269} We excluded the remainder of studies from these two reviews because of wrong population (COPD patients), insufficient sample size, and/or poor quality. In total we include one good-rated RCT,^{255, 256} three fair -rated RCTs,^251–253 and five observational studies.^{259–262, 269}

All nine studies assessed BMD, facture risk, or both (Table 25). In total, four studies evaluated the risk of fracture^{252, 260, 261, 269} and seven measured BMD as an intermediate outcome.^{251–253, 255, 256, 259, 262, 269} Two studies compared one ICS to another,^{251, 252} three compared one ICS to placebo,^{253, 255, 256, 262} and four studies compared one ICS or any ICS to a population that did not use an ICS.^{259–261, 269} Most studies evaluated the risk of bone weakening over two to six years.

Table 25

Summary of studies on bone density or fractures.

Two of the trials were head-to-head RCTs comparing one ICS with another ICS in adult subjects.^{251, 252} One 24-month open-label trial measuring BMD and vertebral fractures randomized 374 adult patients with asthma to beclomethasone, budesonide, or placebo.²⁵² Patients were titrated to the minimal effective dose following a pre-specified management plan; subjects who required more than three courses of oral corticosteroids were withdrawn. At two years, no significant differences in BMD were reported between the three treatment groups. A smaller trial reporting BMD randomized 69 asthmatic patients to medium and high doses of beclomethasone or fluticasone.²⁵¹ At one year, no significant differences in bone mass or metabolism were noted between the two treatment groups.

Seven studies (three of them in pediatric populations) comparing an ICS-treated population to a population not treated with ICSs provided mixed evidence of an association between ICS use and loss of BMD or osteoporosis;^{253, 255, 256, 259–262, 269} three of these studies measured bone fractures.^{260, 261, 269} The studies conducted in pediatric populations reported no difference in BMD between ICS- and placebo-treated subjects and no difference in risk of osteoporosis or time to first fracture between ICS-treated subjects and those not treated with ICS.^{255, 256, 262, 269} Of the remaining studies, one reported a dose-related decline in BMD with ICS-treated subjects,²⁵⁹ one reported a dose-related increase in the risk of vertebral and nonvertebral fractures with ICS,²⁶¹ and two reported no difference in nonvertebral fracture²⁶⁰ or BMD²⁵³ between ICS -treated subjects and controls (Table 25).

II. Growth

Four head-to-head RCTs comparing fluticasone to beclomethasone, ³¹, fluticasone to budesonide,^{44, 249}, or ciclesonide to budesonide⁶² assessed differences in growth. A fair 1-year multinational head-to-head trial determined differences in growth velocity comparing a medium dose of fluticasone (400 mcg/day) to a medium dose of beclomethasone (400 mcg/day) in 343 pre-pubertal children with asthma.³¹ ITT analysis revealed that adjusted mean growth velocity was significantly greater in fluticasone than in beclomethasone-treated patients (+0.70 cm/year; 95% CI: 0.13 to 1.26; P < 0.02). Another fair RCT compared growth velocity in 60 children treated with either a low dose of fluticasone (200 mcg/day) or a low dose of budesonide (400 mcg/day) over one year.²⁴⁹ Fluticasone-treated children had less reduction in growth velocity than the budesonide-treated group (height standard deviation score: 0.03 compared with 0.23; P < 0.05); the authors did not provide absolute numbers in centimeters of differences in growth. The third RCT compared differences in growth velocity in 333 children treated with a medium dose of fluticasone (400 mcg/day) or a medium dose of budesonide (800 mcg/day) over 20 weeks.⁴⁴ Linear growth velocity was greater for fluticasone-treated children compared to those treated with budesonide (adjusted mean increase in height: 2.51 cm compared with 1.89; difference 6.2 mm (95% CI: 2.9–9.6, P = 0.0003). The forth RCT compared growth in 621 children (age 6–11) treated with either a low dose of ciclesonide (160 mcg/day) or a low dose of budesonide (400 mgc/day) over 12 weeks. Ciclesonide-treated subjects had a greater mean body height increase (1.18cm vs. 0.70cm, P = 0.0025).

Four additional studies provide general evidence of growth retardation for ICSs (Table 26). These included two meta-analyses^{246, 247} and three RCTs. ^{124, 254–256} A good quality meta -analysis assessed differences in short-term growth velocity in 273 children with mild to moderate asthma treated with either beclomethasone (mean 400 mcg/day) or placebo for 7 to 12 months.²⁴⁶ The meta-analysis reported a statistically significant decrease in linear growth velocity of children treated with beclomethasone (−1.54 cm per year; 95% CI: −1.15, −1.94) compared to the placebo group. Another good-quality meta-analysis assessed short-term growth velocity in 855 children treated with beclomethasone or fluticasone compared to placebo. Growth velocity was statistically significantly reduced in those treated with beclomethasone (1.51 cm/year; 95% CI: 1.15, 1.87; four studies) and in those treated with fluticasone (0.43cm/year; 95% CI: 0.1, 0.85; 1 study) compared to placebo.²⁴⁷

Table 26

Summary of studies on growth retardation.

The best longer-term evidence of linear growth delay comes from the Childhood Asthma Management Program (CAMP) study, a good quality RCT with median follow-up of 4.3 years that randomized 1,041 asthmatic children to budesonide, nedocromil, or placebo.^{255, 256} The mean increase in height was significantly less in budesonide-treated patients than in placebo-treated patients (−1.1 cm; 22.7 cm compared with 23.8 cm; P = 0.005). This analysis was performed on an intent-to-treat basis, providing a more conservative result than an “as treated” analysis. The differences in growth occurred, however, primarily during the first year of treatment. After two years of treatment growth velocity was approximately the same between groups.

Another placebo controlled trial assessing growth velocity under low-dose fluticasone treatment (100 mcg/day; 200 mcg/d) did not find any significant differences in linear growth compared to placebo after one year of treatment.^{254, 270} One additional fair quality RCT (N = 360) compared linear growth rates in prepubertal children treated with montelukast, beclomethasone, or placebo over 56 weeks and found that the mean growth rate of subjects treated with beclomethasone was 0.78 cm less than that of subjects treated with placebo and 0.81 cm less than that of subjects treated with montelukast (P < 0.001 for both).¹²⁴

III. Acute adrenal crisis

The use of ICSs includes the risk of altered hypothalamic-pituitary axis (HPA axis) functioning and the rare possibility of resultant adrenal suppression. We did not find any studies meeting our inclusion/exclusion criteria reporting on the comparative frequency of clinical adrenal insufficiency in patients treated with ICSs. However, multiple studies report on adrenal suppression during ICS therapy using urinary or serum cortisol levels and results of stimulation tests as intermediate outcomes. It is unclear to what extent results from sensitive studies of HPA axis suppression can be extrapolated to assess differences in risks for clinically significant adrenal suppression.

Various case reports indicate that acute adrenal crisis is an extremely rare but potentially fatal adverse event of ICS treatment.^271–273 However, in most cases dosing was likely outside approved labeling. These case reports did not meet eligibility criteria for this report.

IV. Cataracts

Systemic corticosteroid-induced cataracts typically are located on the posterior side of the lens and are referred to as posterior subcapsular cataracts (PSC); we reviewed studies that compared the risk of PSC in ICS-treated populations to non-ICS-treated populations (Table 27).

Table 27

Summary of studies on posterior subcapsular cataracts.

No study compared the risk of developing PSC between one ICS and another. One head-to-head RCT evaluated the effect of ciclesonide and beclomethasone on eye lens opacity.²⁵⁷ One placebo-controlled trial^{255, 256} and five observational studies^263–267 evaluated the risk of developing cataracts between ICS- and non-ICS-treated patients. One RCT^{255, 256} and one observational study²⁶³ compared budesonide to placebo; the other studies all compared nonspecific ICS use to no ICS use. Two studies were conducted in pediatric populations,^{255, 256, 263} one in a mixed population of children and adults,²⁶⁶ and four evaluated adult populations (≥40 years).^{257, 264, 265, 267}

The single head-to-head RCT²⁵⁷ evaluating eye lens opacity found ciclesonide to be non-inferior to beclomethasone (both delivered at high doses). Both treatments were found to have minimal impact on lenticular opacities development and/or progression. Both trials conducted in children reported no significant differences in the development of PSC between budesonide-treated patients and placebo or matched controls.^{255, 256, 263} One of these was the CAMP study, a good quality RCT with median follow-up of 4.3 years that allocated 1,041 asthmatic children to budesonide, nedocromil, or placebo.^{255, 256} The single study that included a mixed population of adults and children reported no increase in the risk of developing cataracts between ICS-treated patients and controls in persons younger than 40 years; a dose-, duration-, and age-related increase in risk was observed for persons older than 40 years of age.²⁶⁶

Consistent evidence from two case-control studies^{265, 267} and one cross -sectional study²⁶⁴ conducted in adult populations reported an increased risk of cataracts for ICS-treated patients compared to controls. Both case-control studies found the risk of cataracts increased at higher ICS doses and longer duration of treatment; one study reported a higher relative risk for ICS doses greater than 1,600 mcg/day²⁶⁷ and one study reported a higher relative risk for budesonide or beclomethasone doses greater than 1,000 mcg/day.²⁶⁵

Most studies did not control for or did not report previous exposure to systemic corticosteroids, a known cause of cataracts. Only one observational study controlled for previous exposure to systemic corticosteroids; controlling for systemic corticosteroid use and other potential confounders had little effect on the magnitude of the associations in this study.²⁶⁴

V. Ocular hypertension and open-angle glaucoma

No study compared one ICS to another for the risk of ocular hypertension or open-angle glaucoma. One fair-rated case-control study of 48,118 Canadians age 66 years and older²⁶⁵ and one cross-sectional population-based study of 3,654 Australians 49 to 97 years of age²⁶⁸ compared the risk of increased intraocular pressure or open-angle glaucoma between ICS- and non-ICS-treated patients. The populations in these studies were not limited to asthmatics. Both studies reported a dose-related increase in the risk of open-angle glaucoma for ICS-treated patients compared to patients that had not used an ICS. In one study this relationship was observed only among current users of high doses of ICSs prescribed regularly for three or more months (OR 1.44; 95% C.I. 1.01 to 2.06).²⁶⁵ The other study found an association between ever using ICSs and findings of elevated intraocular pressure or glaucoma only in subjects with a glaucoma family history (OR 2.8; 95% CI: 1.2 to 6.8).²⁶⁸ Both studies adjusted for age, sex, oral steroid use, history of diabetes, and history of hypertension (Table 28).

Table 28

Summary of studies on ocular hypertension or open-angle glaucoma.

Systematic review

We found 2 systematic reviews of good quality that compared the fixed-dose combination of an ICS plus a LABA with another ICS/LABA combination for controller therapy.^{94, 281} One review included only randomized, controlled, parallel-design trials and required that only single inhaler devices were used to administer study drugs;⁹⁴ the other allowed administration by either single or multiple inhalers. Studies lasting fewer than 12 weeks or administering “adjustable maintenance dosing” or “single inhaler therapy” rather than fixed doses were excluded from both reviews.

One review has been described in detail in Key Question 1 (section IE) ⁹⁴. The other included eight studies, seven of which compared BUD/FM with FP/SM. The eighth compared FP/SM with beclomethasone/FM, a comparison not relevant to this section of the report. Among the seven relevant studies in the 2010 review,²⁸¹ four were also included in the earlier review and in the RCT section of this report.^{95, 97, 98}. An additional trial is also included in our RCT section but not the earlier review due to its delivery of study medications via separate inhalers¹⁰¹, and results of one unpublished trial and one trial we deemed poor quality¹⁰² were included in the earlier review but not in our report. Results from a second unpublished trial were not reported in either the earlier review, nor are they reported in our RCT section.

Doses of BUD and FM in the included trials ranged from 400–800 (320–640 ex-mouthpiece) mcg/day and 12–24 (9–18 ex -mouthpiece) mcg/day, respectively. All of the published studies administered 500mcg and 100mcg of FP and SM per day; the two unpublished studies administered 12mcg of FM daily and either 200 or 500mcg of FP daily. Included studies ranged from 12 weeks to 30 weeks and took place in the United States and Europe. The total number of participants in the seven relevant trials was 5,935. All included studies enrolled adolescents and adults (no studies in children were identified), and neither restricted asthma severity or current treatment. All included studies were funded by pharmaceutical manufacturers.

Randomized controlled trials

The studies that examined the efficacy of one fixed-dose combination treatment relative to another (described in Key Question 1) also reported tolerability and adverse events. All trials included adolescents and adults; Study duration ranged from 12 weeks to seven months. Methods of adverse events assessment differed greatly. Few studies used objective scales such as the adverse reaction terminology from the World Health Organization (WHO). Most studies combined patient-reported adverse events with a regular clinical examination by an investigator. Often it was hard to determine if assessment methods were unbiased and adequate; many trials reported only those adverse events considered to be related to treatment. Rarely were adverse events prespecified and defined.

1. Budesonide/formoterol for maintenance and relief (BUD/FM MART) compared with Inhaled corticosteroid/Long-Acting Beta Agonist (ICS/LABA) for maintenance and Short-Acting Beta-Agonist (SABA) for relief

The results of the four RCTs contributing five comparisons (one study compared BUD/FM MART with BUD/FM for maintenance and SABA for relief and with FP/SM for maintenance and SABA for relief) are described below under the appropriate drug comparisons. Overall, no studies reported statistical significance of differences between treatments. However, the reported frequencies of adverse events suggest either no difference or a trend toward favoring BUD/FM MART.

Most of the trials reported a numerical trend favoring BUD/FM MART when considering withdrawals due to adverse events. The few trials reporting occurrences of specific adverse events found no difference between treatments.

2. Budesonide/formoterol for maintenance and relief (BUD/FM MART) compared with budesonide/formoterol (BUD/FM) for maintenance and Short-Acting Beta-Agonist (SABA) for relief

Neither trial comparing BUD/FM MART to BUD/FM for maintenance with a SABA for relief^{98, 100, 103, 105} found a difference in adverse events between treatments. The percentage of patients experiencing at least one serious adverse event ranged from 3% to 7% among adults. A subset analysis of the pediatric population of a larger study¹⁰³ found a trend favoring BUD/FM MART (2% of patients had a serious adverse event compared with 14%).

Rate of withdrawal due to adverse events was numerically higher in the BUD/FM+SABA arms of both trials. The magnitude differed between them, possibly due to inconsistency in the definition of an event. In one trial, 1.0% of patients in the BUD/FM MART arm and 1.2% in the BUD/FM+SABA arm withdrew due to adverse events.⁹⁸ In the other, 2.0% (BUD/FM MART) and 4.4% (BUD/FM+SABA) of patients withdrew due to adverse events.

Specific adverse events were reported in only one of the two trials.^{103, 105} The most frequently reported events (those occurring in at least 5% of patients) were respiratory infection, pharyngitis, rhinitis, bronchitis, sinusitis and headache. There were no major qualitative differences between treatments for occurrence of those events, nor were there major qualitative differences in reports of tremor, palpitation, tachycardia, candidiasis or dysphonia, reports of which were rare. In the subset of children within that trial, there was a trend favoring BUD/FM MART for occurrences of serious adverse events, fractures, and pneumonia.

3. Budesonide/formoterol for maintenance and relief (BUD/FM MART) compared with fluticasone/salmeterol (FP/SM) for maintenance and Short-Acting Beta-Agonist (SABA) for relief

Three trials compared BUD/FM MART to FP/SM for maintenance with a SABA for relief.^{98, 100, 104, 106} The percentage of patients experiencing at least one serious adverse event ranged from 3% to 8.2% among adults and adolescents. None of the three included children.

Rate of withdrawal due to adverse events was numerically higher in the FP/SM+SABA arms of two of the three trials.^{104, 106} One percent and 1.2% of participants receiving BUD/FM for maintenance and relief withdrew due to adverse events, compared with 1.7% and 2.0% of patients receiving FP/SM+SABA. One trial¹⁰⁴ reported withdrawals due to “class effect,” a composite measure that included dysphonia, oral candidiasis, oral fungal infection, tremor, tachycardia, palpitations and headache. Fewer patients in the BUD/FM for maintenance and relief arm withdrew due to class effects compared with those receiving FP/SM+SABA, although the rate was <1% in each. In one trial,¹⁰⁶ 27 (2.5%) and 28 (2.6%) patients in the BUD/FM MART and FP/SM+SABA arms, respectively, discontinued the study drug but remained in the trial.

In the third trial, the difference in withdrawals due to adverse event was 0.1% in favor of FP/SM+SABA. Deaths were reported in all three trials, though occurrence was rare. A total of 2 patients treated with BUD/FM MART and three patients receiving FP/SM+SABA treatment died during the trials. In the BUD/FM arms, one death was from severe typhoid fever and the other was due to respiratory failure. One of the patients receiving FP/SM died from cardiac failure; causes of the other two deaths were not specified.

ICS+LTRA compared with same dose ICS

For ICS plus LTRA compared with the same dose of ICS, the systematic review reported no significant differences in overall adverse events (2 trials, RR 1.01, 95% CI: 0.88 to 1.15), specific adverse events (including elevated liver enzymes, headache, and nausea), or withdrawals due to adverse effects (3 trials, RR 0.63, 95% CI: 0.29 to 1.37) among trials using licensed doses of LTRAs (Evidence Tables A).

One fair 16 week trial²³⁰ (N = 639) reported similar rates of overall adverse events (41% compared with 44%; P = NR) and withdrawals due to adverse events (2% compared with 3%; P = NR) in those treated with BUD and those treated with BUD+ML.

ICS+LTRA compared with increased ICS

For ICS plus LTRA compared with increased doses of ICS, the systematic review reported no significant differences in overall adverse events (2 trials, RR 0.95, 95% CI: 0.84 to 1.06), risk of elevated liver enzymes (2 trials, RR 0.8 95% CI: 0.34 to 1.92), headache (2 trials, RR 1.07, 95% CI: 0.76 to 1.52), nausea (2 trials, RR 0.63 95% CI: 0.25 to 1.60), or withdrawals due to adverse events (2 trials, RR 1.14, 95% CI: 0.55 to 2.37) among trials using licensed doses of LTRAs. The trials that used two to four-fold higher than licensed doses of LTRA had a five-fold increased risk of liver enzyme elevation (3 trials, RR 4.97 95% CI: 1.45 to 17).

One fair 16 week trial^{228, 229}(N = 889) reported similar rates of overall adverse events (37.1% compared with 41.3%; P = NR) between groups, but found a slightly increased rate of respiratory infections (11.6% compared with 16.6%; P < 0.05) in those treated with BUD compared to those treated with BUD+ML.