Systematic literature review of the clinical, humanistic, and economic burden associated with asthma uncontrolled by GINA Steps 4 or 5 treatment

Abstract

Objective: This study sought to characterize the epidemiologic, clinical, humanistic, and economic burden of patients with asthma uncontrolled by GINA Steps 4 or 5 treatment (severe, uncontrolled asthma [SUA]).

Methods: A systematic literature review adhering to PRISMA guidelines was performed. Relevant publications were searched for in MEDLINE and EMBASE from January 2004 to September 2016 and in a conference proceedings database from January 2012 to October 2016. Studies were screened using the Population, Intervention, Comparator, Outcomes, Study Design, and Time (PICOS-T) framework. Studies of SUA with observational (prospective and retrospective), randomized, or nonrandomized study designs; adult patient populations; sample sizes ≥20 patients; epidemiologic or clinical outcomes, patient-reported outcomes (PROs), or economic outcomes were included. For our analysis, SUA was defined as inadequate control of asthma, despite the use of medium- to high-dosage inhaled corticosteroids and at least one additional treatment.

Results: A total of 195 articles reporting unique study populations were included. Prevalence of SUA was as great as 87.4% for patients with severe asthma, although values varied depending on the criteria used to define asthma control. Compared with patients with severe asthma who were controlled, patients with SUA experienced more symptoms, night-time awakenings, rescue medication use, and worse PROs. SUA-associated costs were 3-times greater than costs for patients with severe, controlled disease.

Conclusion: Despite the availability of approved asthma treatments, this literature analysis confirms that SUA poses a substantial epidemiologic, clinical, humanistic, and economic burden. Published data are limited for certain aspects of SUA, highlighting a need for further research.

Keywords:

• Severe
• uncontrolled asthma
• economic burden
• epidemiology
• clinical burden
• humanistic

Introduction

Asthma is a chronic disorder of the airways characterized by inflammation associated with bronchial hyperreactivity and spasm, resulting in airflow obstruction. Features of asthma include wheezing, chest tightness, breathlessness, and coughing. More than 300 million individuals worldwide are affected by asthma, and this number is projected to exceed 400 million by 2025¹. Asthma occurs in all countries, irrespective of socioeconomic level, but intercountry prevalence varies. In the US, the prevalence of asthma is 7.6% and 8.4% for adults and children, respectively²; in the European Union, the prevalence is 8.2% and 9.4%, respectively^3,4.

Asthma presents with varying degrees of severity, from mild and moderate to severe disease. Severe asthma can be debilitating and even life-threatening. Previously based on symptoms, the definition of asthma severity now has evolved to one that is focused on the intensity of treatment required to achieve good symptom control^1,5. According to international guidelines such as the Global Initiative for Asthma (GINA), asthma is severe if, despite the elimination of modifiable factors (e.g. poor inhaler technique/adherence, persistent environmental exposures), it requires high-dosage inhaled corticosteroids (ICS) plus a second controller with or without oral corticosteroids (OCS; GINA Steps 4 or 5) to prevent it from becoming uncontrolled, or remains uncontrolled despite this treatment⁵. Patients are assessed as having uncontrolled disease if they have at least three of the following: daytime symptoms (more than twice a week), night-time symptoms, activity limitation, or the need for rescue medication more than twice a week⁵. Patients with asthma uncontrolled by GINA Steps 4 or 5 treatment (subsequently described as severe, uncontrolled asthma [SUA]) respond poorly to standard therapies and often rely on regular intake of OCS for symptom control⁵. SUA accounts for a disproportionate use of asthma-related healthcare resource utilization, including inpatient care, and remains a significant unmet need in respiratory medicine^6,7.

We conducted a systematic literature review (SLR) to further understand the burden of SUA and to identify unmet needs associated with this disease. The following aspects of SUA were evaluated: epidemiology, clinical burden (including clinical outcomes of pharmacologic treatments, consequences of long-term/frequent OCS use), humanistic burden (patient-reported outcomes [PROs]), and economic burden (including direct and indirect costs). Data for the overall SUA population were obtained from the literature and are summarized in this review.

Methods

We performed an SLR adhering to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines⁸. MEDLINE, via PubMed, and EMBASE were searched for relevant studies published from January 1, 2004 (to capture publications following approval of the first biologic therapy for severe, uncontrolled asthma) to September 12, 2016. The search algorithm included keywords for SUA paired with outcome terms of interest (Supplementary Tables 1–4). We manually checked the bibliographies of relevant SLRs and meta-analyses published since 2010 contained within both databases to confirm comprehensive retrieval of literature of interest.

In addition to MEDLINE and EMBASE, scientific conference proceedings published from January 2012 through October 2016 were reviewed to identify potential studies of interest using the MedMeme Online Portal for Clinical Intelligence database. Supplementary Table 5 shows the terms used for this search.

Study selection

Inclusion criteria

Studies of asthma uncontrolled by GINA Steps 4 or 5 treatment with the following characteristics were included: observational (prospective and retrospective), randomized, or non-randomized study designs; adult patient populations; sample sizes ≥20 patients; and epidemiologic, clinical, patient-reported, and/or economic outcomes. For our review summary, uncontrolled asthma was defined as asthma that was inadequately controlled despite the use of medium- to high-dosage ICS and at least one additional treatment, such as a long-acting β₂-agonist (LABA), as specified in the 2016 GINA guidelines (referred to as severe, uncontrolled asthma [SUA])⁵. Criteria for defining asthma control varied between studies and included exacerbation/hospitalization history or validated symptom questionnaires. To gain insights from a global perspective, we included studies published in English, Chinese, Dutch, French, German, Italian, Japanese, Portuguese, Russian, and Spanish languages.

Exclusion criteria

We excluded studies with only pediatric or adolescent patients (as disease characteristics differ from those of adults) and studies of seasonal asthma, genetics of asthma, and non-pharmacologic treatments for asthma. Animal studies, studies using in vitro models, and data from published letters, commentaries, editorials, and narrative reviews were excluded.

Review process

Following the literature search, a single reviewer manually screened all titles and abstracts identified from MEDLINE, EMBASE, and the scientific congress proceedings databases against the inclusion and exclusion criteria using the PICOS-T (Population, Intervention, Comparator, Outcomes, Study Design, and Time) framework⁹. The full-text articles of accepted studies passing abstract screening were retrieved, and an investigator screened them using the same inclusion and exclusion criteria that had been applied at the abstract level. All excluded studies were confirmed by a second, senior investigator. Any discrepancies between the two investigators were resolved by a third investigator. Full-text articles published in non-English languages were translated into English and reviewed following the above-described process to determine their eligibility for inclusion in the review.

Data abstraction

We performed data abstraction of the included studies using an evidence table template. For each study, data were abstracted into a data input grid by two investigators independently trained in the critical assessment of evidence. A senior investigator resolved any discrepancies.

Quality assessment of included studies

We evaluated all selected publications to determine the quality of the studies according to criteria modified from the University of Oxford’s Centre for Evidence-Based Medicine (CEBM)¹⁰. CEBM ratings were only given to studies published in the peer-reviewed literature (i.e. MEDLINE and EMBASE).

Results

Study selection

The MEDLINE (via PubMed) and EMBASE searches for literature were performed on September 12, 2016, and yielded 4,986 publications from MEDLINE and 6,067 from EMBASE, with overlap between the two databases. After removing duplicate publications, we identified 8,187 unique articles and screened them at the abstract level, selecting 1,718 for further review of the full text. Of these, 169 primary studies were accepted for inclusion in the SLR.

A review of the conference proceedings identified 59 additional meeting abstracts, resulting in 228 included studies. Thirty-three of these reported data on the same study population and were designated “related publications” to avoid counting patients multiple times in the review’s results. Many of the related publications were either sub-group analyses or conference abstracts that were also available as full publications. When accounting for the presence of related publications, 195 articles on unique study populations were included for summary in the SLR. Figure 1 presents a summary of study attrition at each level of the SLR.

Figure 1. Study selection.

* Conference proceedings identified from the MedMeme Online Portal for Clinical Intelligence Database.

Epidemiology of asthma uncontrolled by GINA Steps 4 or 5 treatment

Our SLR identified 83 primary studies with epidemiology data relevant to SUA; all were observational studies. Three studies were international (involving multiple countries from different continents)^11–13, and the remaining studies were from Asia^14–19, Australia^20,21, Germany^22,23, France^24–27, Italy^28–35, Spain^36–40, the UK^6,41–48, Europe (other)^49–68, the Middle East^69,70, South America^71–75, and the US^76–90. Table 1 shows a summary of the epidemiology of SUA, and Supplementary Table 6 provides details of the study designs and main results. Of the identified studies, only one study reported on severe asthma-related mortality; however, this study reported risk factors for mortality in severe asthma patients rather than the prevalence of mortality from severe disease⁹¹. All other epidemiologic studies reported on the prevalence of SUA, most often in the asthma patient population. These studies reported that a small but clinically important segment of patients with severe asthma often has uncontrolled symptoms, despite available maintenance therapies (high-dosage ICS/LABA).

Table 1. Summary of key studies reporting the prevalence of asthma uncontrolled by GINA Steps 4 or 5 treatment (severe, uncontrolled asthma [SUA]).

Country	Number of studies (number of patients)	How asthma control was defined	SUA in asthma population	SUA in uncontrolled asthma	SUA in severe asthma	Prevalence in the general population
International (France, Germany, Italy, Spain, UK, US)^12	1 (n = 9,734)	ACT < 20	10.0%	NR	79.0%	NR
US^Citation76,Citation78,Citation79,Citation81,Citation83,Citation85,Citation89	7 (n = 320–12.6 million)	Self-assessment of control in the previous 4 weeks; EPR-3 definition: the presence of either risk or impairment; ACT <20 at baseline and 6 months; ≥2 exacerbations^a; ACT < 20 at baseline and every 6 months to 24 months; ≥2 exacerbations^a and high-dosage therapy; ≥3 exacerbations^a; NAEPP guidelines	1.06% (EPR-3)– 49.2% (ACT)	29.9 (EPR-3)–88.0% (≥ 3 exacerbations)	19.9 (≥2 exacerbations)–49.2% (ACT)	0.03 (≥2 exacerbations)
UK (Scotland)^45	1 (n = 267)	BTS guideline	16.9%	NR	NR	NR
France^Citation24–26	2 (n = 491–15,534)	ACT score (<15 very poorly controlled; 15–20 poorly controlled); ANAES (French adaptation of the GINA guidelines); GINA 2006 classification; patient judgment	3.2% (ACT)–18.4% (GINA)	51.2% (GINA)	26.6 (patient judgment)–87.4% (ANAES)	NR
Germany^23	1 (n = 150)	GINA 2009 classification	30.0%	66.2%	38.1%	NR
Italy^28	1 (n = 2,835)	ACT ≤ 15	NR	34.6%	14.2%	NR
Spain^Citation36,Citation37	2 (n = 1,224–36,649)	ACQ >1.5; clinician judgment; GINA guidelines	1.8% (GINA)–3.9% (clinical judgment)	NR	55.7% (ACQ)	NR
Denmark^176	1 (n = 61,583)	High SABA use (≥600 doses of SABA) and/or at least 1 exacerbation in the 1-year period of observation	3.0%	11.7%	36.4%	NR
The Netherlands^53	1 (n = 32,163)	Hospitalization for asthma or use of more than one short (<30 days) course of oral corticosteroids in 2004	0.7%–0.8%	NR	17%	NR
Brazil^Citation71,Citation72,Citation75	3 (n = 275–183 million)	Patient self-assessment; ACQ ≥ 1.5; GINA 2006 classification	38.9% (GINA)	69.9% (GINA)	8.0 (self-assessment)–74.3% (GINA)
South Korea^17	1 (n = 2,298)	ATS criteria for refractory asthma	4.3%	NR	NR	NR
Israel^70	1 (n = 123)	GINA 2006 classification	NR	NR	43.9%	NR

Abbreviations. ACT, Asthma Control Test; ACQ, Asthma Control Questionnaire; ANAES, L’Agence Nationale d’Accreditation et d’Evaluation en Sante; ATS, American Thoracic Society; BTS, British Thoracic Society; EPR-3, The Expert Panel Report 3; GINA, Global Initiative for Asthma; NAEPP, National Asthma Education and Prevention Program; NR, not reported; SABA, short-acting β₂-agonist.

^a Exacerbations per year.

Prevalence of severe, uncontrolled asthma

The prevalence of SUA within the general population was reported as 0.03% in a study from the US in which SUA was defined by ≥2 exacerbations per year⁹². The prevalence of SUA as a percentage of all asthma varies among studies and countries, largely because of variation between clinical and epidemiologic definitions of SUA. In identified studies, criteria used to define asthma severity and control included GINA⁴ or British Thoracic Society⁹³ guidelines, Asthma Control Test (ACT)⁹⁴ score cut-off of 16–19 (not well controlled) or <15 (very poorly controlled), Asthma Control Questionnaire (ACQ)⁹⁵ score >1.5 (uncontrolled), clinician judgment, and patient self-assessment. The percentage of the overall asthma patient population with SUA reported in these identified publications ranged from 0.7%⁵³ in the Netherlands, where SUA was defined by hospitalization/short course of OCS, to 49.2% in the US, where SUA was defined by ACT <20 at baseline and 6 months⁷⁹. Studies using GINA guidelines criteria reported an SUA prevalence of 1.8³⁷ to 38.9%⁷² within the asthma patient population. Among eight large population studies (sample size ranging from 4,755–36,649 patients), the percentage of all patients with asthma with SUA ranged from 3.2% in France²⁶ to 10% in the US TENOR study population⁸¹ for studies with the same definition of disease (treatment with GINA Steps 4 or 5 [or equivalent]; uncontrolled status assessed using a standard questionnaire).

Studies reporting the percentages of patients with severe asthma and uncontrolled disease used varying criteria to determine the degree of control. These criteria included exacerbation history, GINA guideline definition, and use of validated asthma control questionnaires. For patients with severe asthma, up to 87.4% were reported to have SUA (French study, asthma control defined by adapted GINA guidelines)²⁴. As in the overall asthma population, the prevalence of SUA in patients with severe asthma varied greatly depending on the criteria used. For example, in a Brazilian study, the prevalence of SUA increased from 8% when using self-assessment to 38.8% when control was assessed using the ACQ-6⁷¹.

Patient characteristics and risk factors for severe, uncontrolled asthma

The SLR identified only one study that assessed mortality associated with SUA. It was a nested case-control study of patients in Brazil with severe asthma⁹¹. Multivariate analysis identified male sex, uncontrolled asthma in the past year, number of asthma exacerbations in the past year, and prebronchodilator forced expiratory volume in 1 second (FEV₁) < 60% as significant and independently associated risk factors for mortality for the population with severe asthma. This study suggested that the risk of respiratory-related mortality was approximately 8-times greater for patients with SUA than patients with controlled disease⁹¹.

Approximately 50% of patients who had SUA despite the use of high-dosage ICS had elevated blood eosinophil counts (≥300 cells/µL)⁹². For patients with SUA, greater eosinophil counts were significantly associated with persistent airway obstruction, absence of atopy, late-onset asthma, and age younger than 65 years⁴². Patients’ body mass index had an inverse relationship with both sputum and blood eosinophil counts^32,35,65.

Clinical burden of asthma uncontrolled by GINA Steps 4 or 5 treatment

Our SLR identified 113 publications that included data related to the clinical burden of SUA. Twenty-three of these studies, reported in 21 publications, were international studies^13,96–115; others reported data from Asia^{14–16,19,116–124}, Australia^21,125, Germany^22,126–129, France^{50,54,113,130–133}, Italy^{29–31,34,134,135}, Spain^{37,39,40,136–138}, the UK^{6,41–48,139–146}, Europe (other)^{56,62,64,67,68,122,131,147–155}, Mexico^156,157, South America¹⁵⁸, the Middle East^159–164, Africa¹⁶⁵, the US^{77–79,88,92,115,166–171}, Canada¹⁷², and the US and Canada combined^173,174. These studies evaluated asthma exacerbation rates and the effect of active treatment on asthma exacerbation rates, lung function, and asthma medication use. Supplementary Table 6 provides details of the study designs and main results.

Our review found that, as expected, exacerbations were common for people with SUA, and patients with severe asthma had more than twice as many asthma-related hospitalizations as those with non-severe disease⁹². However, estimates of annual asthma exacerbation rates varied depending on inclusion and exclusion criteria and the definition of asthma exacerbation used by the reporting studies. Consistent with current treatment guidelines^5,175, most of the studies defined exacerbation by the need for corticosteroid use and/or emergency department (ED) visit or hospital admission. The annualized exacerbation rate from observational studies of patients with SUA ranged from 0.48 in Japan (n = 54)¹⁴ to 6.96 in Ireland (n = 30)⁶², and the differences in these reported ranges reflect significant variations in study designs and inclusion criteria. The exacerbation rate identified for patients from Ireland was reported before the availability of omalizumab as a treatment option. Omalizumab was also not available for a portion of the time period covered in the report from Japan¹⁴. The rate of severe exacerbations was more consistent across studies, with an annualized rate of 2.27 in Greece (n = 60)⁶⁴ to 3.60 in Spain (n = 263)^39,40. Exacerbation rates were provided as a 12-month prevalence percentage in two studies and ranged from 19.6% for Denmark¹⁷⁶ (n = 61,583) to 31.2% for Taiwan (n = 282)¹¹⁶. No non-treatment-associated observational studies were identified that reported on the lung function status of patients with SUA.

Impact of biologic therapies on clinical burden

Omalizumab treatment was approved for the treatment of patients with severe asthma for a substantial period covered by this review. We identified 58 publications of reports of omalizumab from clinical trials and observational settings^{15,19,21,22,27,29–31,34,39,40,48,56,62,64,67,68,79,98,99,104,117,122,124–131,134,136,138,139,146,149,152,153,156–158,160–164,167–169,174,177–182}. Of these publications, most trials and all observational studies showed a reduction in exacerbations with treatment^{29,30,39,48,98,127,128,177,178,183}. Two German studies reported the greatest and the smallest change in annual exacerbation rates after omalizumab treatment; the greatest change was a 93% reduction¹²⁷, and the smallest was a 22% reduction (although this analysis only included patients 50 years or older)²². However, trials of omalizumab showed inconsistent results for improvement in lung function^{56,64,96,100,131,152,153,156,157}. Omalizumab treatment was associated with a decrease in the use and dosage of other asthma medications, including a reduction in OCS use^{48,116,124,129,136,138,139,146,177} and ICS use^{62,64,136,168}. Omalizumab therapy also reduced rescue medication use in several trials^15,40,62,136. Although overall the extensive studies reporting on the treatment of severe asthma with omalizumab show this drug to be effective, not all patients with SUA are suitable for treatment with omalizumab, and not all those identified as candidates respond to treatment¹⁷⁵.

The impact of active treatment on the risk of exacerbations with other biologic treatments varied. Reductions in annual exacerbations were observed in randomized controlled trials (RCTs) of benralizumab (reduction vs placebo: 29–70%^97,102,114), mepolizumab (reduction vs placebo: 47–52%¹¹¹, 39–52%¹¹⁰ [depending on dosage and administration route]), dupilumab (reduction vs placebo: 33–87%^13,171 [depending on treatment schedule]), and reslizumab (reduction vs placebo: 46%¹¹³). In contrast, there were no significant differences between the effects of placebo and active treatment exacerbations for trials of golimumab¹¹², etanercept¹⁸⁴, KB003¹⁰⁸, or tralokinumab¹⁰⁰.

Improvements in lung function were observed only with certain active biologic treatments, and not consistently in all studies of each treatment. Improvements in lung function compared with placebo, although not statistically significant in all cases, were reported in RCTs for dupilumab^13,171, benralizumab^97,102,114, and reslizumab¹¹³. Golimumab¹¹² and etanercept¹⁸⁴ were not significantly more effective than placebo at improving lung function outcomes in any trial reported. Tralokinumab showed inconsistent results, with significant improvements in some, but not all, lung function measures associated with active treatment¹⁰⁰.

The impact of treatment on medication use (asthma and rescue medication) varied for patients with SUA. As discussed previously, omalizumab treatment was associated with a decrease in the use and dosage of other asthma medications, including rescue medication. Dupilumab¹⁷¹ treatment also reduced rescue medication use, while golimumab¹¹² treatment, compared with placebo, did not. During the review period covered by this SLR, the impact of newer biologic treatments on OCS use was not reported.

Impact of non-biologic therapies on clinical burden

Identified studies reported no significant differences between the effects of placebo and active treatment on exacerbations in trials of masitinib¹⁸⁵, AZD5069¹⁰⁹, or mometasone furoate¹⁴⁴. Improvements in lung function compared with placebo, although not statistically significant in all cases, were reported in RCTs of prednisolone¹⁵¹. In observational studies, lung function improved significantly for patients receiving mycophenolate mofetil¹⁴⁴, tiotropium with salmeterol/fluticasone¹⁵⁴, and pranlukast¹¹⁹. Azithromycin¹⁴⁷, mometasone furoate¹⁵⁴, and sodium cromoglicate¹²¹ were not significantly more effective than placebo at improving lung function outcomes in any trial reported. Ciclesonide demonstrated inconsistent results, with significant improvements in some lung function measures associated with treatment⁹⁶.

Humanistic burden of asthma uncontrolled by GINA Steps 4 or 5 treatment

We identified 103 primary studies that included data related to PROs for patients with SUA. Twenty-three international studies in 19 publications were identified^{11,13,97,98,100–102,104–114,132}. Additional studies were from Asia^{15,16,18,19,117,119–124}, Australia^21,125, Germany^{126–129,186}, France^179,181,187, Italy^34,183, Spain^39,40, the UK^{41–43,45,46,48,139–142,144–146,177,182,184,188,189}, Europe (other)^{50,51,55–57,59–61,64,69,147–150,152,153,161–163,184,190}, the Middle East¹⁶⁷, South America^73,158, the US^{77,81,87,115,147,167,191}, Canada^173,174, and Africa¹⁶⁵. Validated questionnaires used in these studies to measure PROs included the Asthma Quality of Life Questionnaire (AQLQ)¹⁹², ACT⁹⁴, and ACQ, and the minimally clinically important differences were 0.5, 3, and 0.5 points, respectively⁹⁵. Supplementary Table 6 provides details of study designs and the main results.

Observational studies reported that, compared with patients with controlled disease, patients with SUA experienced more symptoms, night-time awakenings, and rescue medication use, as well as worse health-related quality-of-life (HRQoL)^37,132. In the identified observational publications, HRQoL was most often assessed by AQLQ score, but some studies also used the European Quality of Life-Five Dimensions Questionnaire and the St. George’s Respiratory Questionnaire. Compared with patients with moderate-to-severe, controlled disease, patients with SUA scored an average of 1.4 points greater (worse control) on the ACQ³⁷, 7 points greater (worse control) on the ACT³⁴, and 1.3 points lower (worse HRQoL) on the AQLQ³⁷.

In RCTs of treatments for patients with SUA, patient questionnaires were frequently used to assess treatment-associated improvements in PROs. RCTs have reported statistically significant treatment improvements in HRQoL for patients with immunoglobulin E (IgE)-mediated allergic SUA, as assessed by the AQLQ for patients treated with mometasone furoate (non-biologic) or omalizumab (biologic)^{15,104,117,145,150,174}. No significant effects on HRQoL were reported in the identified publications for patients treated with the biologic therapies etanercept¹⁸⁴ and golimumab¹¹², or the non-biologic AZD5069¹⁰⁹. Mixed results were seen for trials of patients treated with the biologic therapies reslizumab¹¹³, benralizumab^97,102, tralokinumab¹⁰⁰, dupilumab¹³, and mepolizumab^{11,111,145,150}, and the non-biologic tiotropium^106,150.

In the publications we identified, prospective cohort studies reported significant improvements in disease control associated with omalizumab biologic treatment using the ACQ, and several RCTs across geographic areas reported significant treatment effects on PROs for omalizumab^{15,21,48,98,124,126,138,181}. RCTs evaluating other biologic treatments, including benralizumab^97,102 and dupilumab^13,171, and the non-biologic masitinib¹⁸⁵, also found significant effects of treatment on symptom control, as assessed by ACQ score. However, RCTs evaluating the biologics reslizumab^113,173, mepolizumab^110,111, and golimumab¹¹² and the non-biologics AZD5069¹⁰⁹ and azithromycin¹⁴⁷ found no significant treatment effects on ACQ score. Trials evaluating tiotropium (nonbiologic)^105,106 and tralokinumab (biologic)¹⁰⁰ did not report consistent results. RCTs and several cohort studies reported an association between omalizumab therapy and beneficial changes on the ACT scores of patients with SUA (including for those with IgE-mediated allergic SUA)^{23,36,39,64,71,145,147,173,178,179,193}.

Overall, the results of these studies highlight that there is a considerable, worldwide humanistic burden associated with SUA. In many studies, there was evidence that this burden remained after treatment. Although treatment was associated with some improvements in PROs, these improvements were not consistently statistically significant compared with placebo for all outcomes measured and across studies.

Economic burden of asthma uncontrolled by GINA Steps 4 or 5 treatment

We identified 65 primary studies that included data related to the economic burden of SUA. Nine international studies were identified^{97,98,102,104,105,111,112,114,194}, as well as studies from Africa¹⁶⁵, Asia^{19,116,122,124,195}, Australia¹²⁵, Canada¹⁹⁶, France^27,178,179, Germany^22,127,128, Italy^{30–32,34,197}, Spain^37,39,40,138, the UK^{6,41,42,44,46,48,80,139–142,145,146,184,189,198,199}, Europe (other)^{49,51–53,55,62,63,68,131,147–149,152,200,201}, Mexico^156,157, the Middle East^159,164, South America^71,73,202, and the US^{71,73,76,77,84,85,87,88,92,166–169,171,202–204}. Supplementary Table 6 provides details of study designs and the main results.

Across studies, this review identified that a substantial economic burden was associated with SUA, which was reduced but not eliminated by treatments approved for SUA. This burden was measured most often by healthcare resource utilization, such as hospitalizations, ED visits, and intensive care unit (ICU) admissions, as well as unscheduled and scheduled physician appointments (typically asthma exacerbation/symptom worsening related). Patients with SUA had an average of 3–3.5 office visits, 0.6–2.0 ED visits, and 0.5–0.6 hospitalizations annually^37,179. More than 80% of studies reported that ≥17% of patients with SUA had been hospitalized in the year prior to study entry.

A Canadian study (1996–2000) reported that patients with SUA accounted for 60% of total asthma-related healthcare costs, estimated at $25 million per year (inflation-adjusted 2006 Canadian Dollars) in British Columbia, Canada¹⁹⁶. A European study performed in The Netherlands reported that total direct medical costs per patient year for all patients with SUA were 3-times as high as for patients with severe asthma, but whose disease was controlled⁵³. Overall, for a representative population of patients with asthma from The Netherlands (n = 32,163), medication and hospitalization costs for all patients with severe asthma (n = 1,518) totaled €1.6 million in 2004. Above standard maintenance costs, therapy drug costs totaled roughly €700 per patient per year for patients with SUA with hospitalization and €300 per patient per year for patients with SUA with multiple courses of OCS. Costs for patients with SUA requiring hospitalization (including admission- and drug-associated costs) were more than €10,000 per patient per year⁵³. Patients with SUA received frequent or long-term OCS, with an average of 43% greater associated direct healthcare treatment costs than patients not receiving maintenance OCS⁶. Other studies non-specific to SUA have associated OCS treatment with short- and long-term detrimental adverse effects, including diabetes, osteoporosis, high blood pressure, skeletal complications, ocular complications, and Cushingoid features^205,206.

Identified studies reported that SUA severely impaired quality-of-life and was associated with substantially greater healthcare resource use⁹². A US study evaluating SUA-associated healthcare costs reported that a sub-set of patients with severe eosinophilic asthma (defined as ≥1 laboratory result with a blood eosinophil count ≥300 cells/µL) incurred greater costs than patients with severe asthma without an elevated blood eosinophil count. Persistent asthma was defined as asthma that was inadequately controlled, and severe, persistent asthma was defined as asthma in patients who had evidence of ≥2 exacerbations during the study period (Figure 2)⁹².

Figure 2. Mean all-cause and asthma-related costs for patients with persistent severe, uncontrolled; and severe, eosinophilic asthma⁹⁰.

SUA was also associated with greater indirect costs. In RCTs of omalizumab from France and the UK, patients with SUA had mean baseline annual days of absenteeism of 14.6 and 49.4 days, respectively^48,179. A small (n = 92) study from Korea found that, for patients with severe asthma, the indirect costs were 3.5-times greater for those with uncontrolled vs well-controlled disease ($4,909.50 vs $1,217.80 per year)¹⁹⁵. In a cross-sectional study from the European Network for Understanding Mechanisms of Severe Asthma, which included patients from nine European countries, patients with SUA reported that work affected their breathing and that they often had to change jobs due to their asthma. In addition, fewer patients with SUA were employed at the time of the study than patients who had mild disease (odds ratio = 0.39)⁵⁵.

Discussion

This SLR provides a broad overview of the burden associated with SUA (defined as asthma uncontrolled by GINA Steps 4 or 5 treatment), including epidemiologic, clinical, humanistic, and economic outcomes. In the identified literature, the prevalence of SUA varied widely based on the definition of asthma control used and the study population. Historical differences in the criteria used to define SUA, patient and physician perceptions of control, and the application of the GINA guidelines may all have contributed to the variation in reported estimates. Bearing in mind these limitations, from the available literature identified, the reported prevalence of SUA ranged from 0.7%⁵³ to 49.2%⁷⁹ within the asthma patient population⁸¹, and was up to 87.4% within the severe asthma patient population²⁴. The prevalence of SUA within the asthma patient population tended to be smaller for larger population studies (>2,000 patients) and greater for studies that defined SUA according to GINA guidelines. Clinician and patient self-assessments tended to over-estimate the degree of patients’ asthma control.

Our review identified a large body of evidence confirming that patients with SUA have a significant burden of disease, characterized by poorly controlled symptoms and frequent exacerbations, reduced HRQoL, and substantial healthcare-associated costs. This indicates that SUA is associated with a significant impediment to patients’ everyday lives. Compared with patients with moderate-to-severe disease, those with SUA have more persistent symptoms, more frequent exacerbations, and a greater number and rate of comorbidities. Identified studies also reported that SUA is associated with a reduced HRQoL, which was most frequently assessed using the AQLQ, a patient questionnaire designed to evaluate the physical, emotional, social, and occupational burden of patients with asthma. Overall, the studies we identified indicate that the symptom burden associated with SUA is considerable, globally consistent, and persists even after treatment. These results are consistent with the results of another review that demonstrated that many patients treated according to the GINA guidelines continue to have symptoms that adversely affect their HRQoL²⁰⁷.

We identified 40 studies evaluating the efficacy of omalizumab for the treatment of SUA. Omalizumab, the first targeted biological therapy for the treatment of SUA, was first approved in 2003. The extent of the literature evaluating its use is not unexpected. The identified studies, which were published from 2004–2016, included observational studies and RCTs. Omalizumab was associated with a reduction in exacerbations in all observational studies and most RCTs, but results for lung function were inconsistent between studies (both observational and RCTs). Overall, these results provide evidence that SUA continues to contribute a significant clinical and economic burden, despite the introduction of omalizumab. Reports in the literature of more recently approved targeted biological therapies, such as mepolizumab, reslizumab, and benralizumab, are mostly restricted to RCTs. As these therapies become more widely used over time, it will be interesting to see if and how they affect the clinical, humanistic, economic burden associated with SUA at the population level.

The identified studies reporting on the economic burden of SUA most often assessed healthcare resource utilization by analyzing the frequency of patient hospitalizations, ED visits, ICU admissions, and physician appointments. These studies reported a considerable economic burden associated with SUA compared with controlled asthma. This is in agreement with a recent report from the UK, which stated that mean per-patient healthcare resource and associated costs were 4-times greater for patients with eosinophilic SUA than the general asthma patient population²⁰⁸. SUA has also been associated with a greater risk of exacerbations, hospitalizations, deaths, and OCS dependency²⁰⁹. Improved asthma control would decrease the need for hospitalizations and lower the overall costs associated with SUA.

Our review identified several data gaps in the available literature relating to SUA. In particular, data were lacking for the mortality associated with SUA and the SUA-related indirect costs caused by patient loss of productivity, work absenteeism, and caregiver burden.

Limitations

As is the case for all SLRs, this review has certain limitations. For example, it was subject to selection bias, although this was minimized by using a stringent case definition of SUA (i.e. inadequate control of asthma despite the use of medium- to high-dosage ICS and at least one additional treatment). However, during the timeframe of studies included in this review, the generally accepted definition of SUA changed from symptom-based to treatment-based. Although we made every attempt to standardize the definition of SUA, it is possible that unidentified misclassifications were present in our results. For the assessment of PROs, variation existed between studies on how clearly outcomes were described and whether actual values, relative change from baseline, or differences between treatment and placebo were reported. Furthermore, many reports from RCTs provided limited details on the methods of administering the instruments and how estimates of results were derived. There was also inconsistent reporting of how patient discontinuations and missing data were handled. This could significantly impact the reported results, because, for some studies, dropout rates were significant and may have been conceivably greater for placebo study arms.

Conclusions

This SLR identified a substantial epidemiologic burden, as well as a clinical, humanistic, and economic burden, associated with SUA, despite the availability and use of current treatments. Although this review is subject to limitations, most notably the lack of a consistent definition of severe, uncontrolled asthma, it highlights the considerable unmet need worldwide for patients with SUA. This is important, because many of the 250,000 annual asthma-related deaths worldwide can be attributed to avoidable factors (e.g. lack of referral for specialist care)^210–212. Thus, clinicians should understand the heterogeneous nature of asthma and improve treatment and HRQoL for patients with SUA.

Transparency

Declaration of funding

This analysis was funded by AstraZeneca.

Declaration of financial/other relationships

Stephanie Chen, Sarowar Golam, and Xiao Xu are employees of AstraZeneca. Julie Myers, Chris Bly, and Harry Smolen are employees of Medical Decision Modeling, Inc. Peer reviewers on this manuscript have received an honorarium from CMRO for their review work, but have no other relevant financial relationships to disclose.

Acknowledgments

Editorial support was provided by Debra Scates, PhD, and Lourdes Yun, MD, MSPH, of JK Associates Inc., a member of the Fishawack Group of Companies, and Michael A. Nissen, ELS, of AstraZeneca. This support was funded by AstraZeneca.