Health economics of asthma: assessing the value of asthma interventions

Authors


Jonathan D. Campbell
Pharmaceutical Outcomes Research and Policy Program
University of Washington
1959 NE Pacific Avenue
H375Q
Box 357630
Seattle
WA 98195
USA

Abstract

The aim of this systematic review was to summarize and assess the quality of asthma intervention health economic studies from 2002 to 2007, compare the study findings with clinical management guidelines, and suggest avenues for future improvement of asthma health economic studies. Forty of the 177 studies met our inclusion criteria. We assessed the quality of studies using The Quality of Health Economic Studies validated instrument (total score range: 0–100). Six studies (15%) had quality category 2, 26 studies (65%) achieved quality category 3, and the remaining eight (20%) studies were scored as the highest quality level, category 4. Overall, the findings from this review are in line with the Global Initiative for Asthma clinical guidelines. Many asthma health economic studies lacked appropriate long term time horizons to match the chronic nature of the disease and suffered from using effectiveness measures that did not capture all disease related risks and benefits. We recommend that new asthma simulation models: be flexible to allow for long term time horizons, focus on using levels of asthma control in their structure, and estimate both long term asthma specific outcomes like well-controlled time as well as generic outcomes such as quality adjusted survival.

Health economics is the study of how scarce healthcare resources are used and should be used. Because resources are scarce, healthcare decision makers must decide, given the available evidence, how to deploy resources in a manner that optimizes value for money. The use of health economic tools aid decision makers in their deployment strategies by: clearly identifying the relevant alternatives, transparently evaluating the perspectives and inputs of the strategies, and modeling uncertainty and what if scenarios (1). Health economics is important because of the desire to make the decision process and its outcomes, more rational and transparent. This desire is revealed through the emergence of government and regional health bodies such as the National Institute for Health & Clinical Excellence (NICE), Scottish Medicines Commission (SMC), Medical Products Agency, Canadian Agency for Drugs and Technologies in Health (CADTH), and Pharmaceutical Benefits Advisory Committee (PBAC). These agencies use health economics to help inform the technology coverage decisions of England and Wales, Scotland, Sweden, Canada, and Australia, respectively.

Previous reviews have summarized the literature on the economic evaluations of asthma interventions (2–8). Sullivan and Weiss observed that the literature was evolving, but did not meet the standards of the time (4). They commented that amongst the studies reviewed, time horizons were not long enough and relevant comparators were not always used. They also argued that there was a lack of standardized outcomes in asthma economic evaluations. Sculpher and Price in their review that extended to 2002 remarked on many of the same issues and made suggestions for overcoming the common weaknesses observed in the literature (3). They advocated for longer term pragmatic trials, explicit statement of the perspective, and the use of cost-consequence models. They noted that only two of the 33 studies reviewed at that time reported utilities and suggested the increased use of generic measures alongside disease specific measures. Lee and Weiss added in their one year update that pharmaceutical agents are the primary component of direct medical expenditures and stressed the importance of evaluating the costs and outcomes of pharmaceuticals and their alternatives (2). Feenstra et al. compared the cost-effectiveness literature for children with asthma to four different asthma guideline bodies (5). They concluded that although cost-effectiveness was not explicitly included in the guidance, cost-effectiveness evidence was consistent with current treatment advice. Akazawa and Stempel (6), Shih et al. (7), and Shepherd et al. (8) reviewed the cost-effectiveness evidence of single inhaler combination therapy for persistent asthma.

We acknowledge the publication of previous asthma burden reviews. These reviews did not focus on the incremental benefit of asthma interventions, but rather focused on geographic specific burden (9, 10), pediatric burden (11–14) or work-related burden (15).

The aim of this systematic review was to summarize and assess the quality of asthma intervention health economic studies from 2002 to 2007, compare the study findings with clinical management guidelines, and suggest avenues for future improvement of asthma health economic studies. No outside funding was provided for the conduct of this study.

Review methodology

Health economic analyses

Health economic analyses can take many forms, but generally consider two main characteristics: assessing the value of the viable treatment alternatives, and comparing the costs and consequences of the alternatives. A cost-effectiveness analysis (CEA) is a formal method for comparing the benefits and costs of a medical intervention to its next best alternative in order to determine whether it is of sufficient value to adopt or reimburse (16). The main output from a cost-effectiveness study is the incremental cost-effectiveness ratio (ICER) which compares two alternative interventions’ average costs (C1 and C2) and effects (E1 and E2) in the form of the following ratio:

image

Cost-consequence analysis is an alternative to cost-effectiveness whereby the analyst presents the disaggregated resource uses, unit costs, and consequences for all of the alternatives. Cost-consequence analysis leaves the weighting of the costs and consequences up to the given decision-maker. Armed with the results of an incremental economic analysis such as a CEA or a cost-consequence analysis, decision makers are better equipped to address an intervention’s value.

Literature search

We performed a systematic search in July of 2008 using multiple databases including MEDLINE (PubMed) and the National Health Services Economic Evaluation Database querying for the MeSH terms: ‘asthma’ and ‘cost–benefit analysis.’ The MeSH term ‘cost–benefit analysis’ includes the following nested entry terms: cost–benefit, cost-effectiveness, and benefits and costs. No other MeSH term is comprehensive enough to require elements of both costs and health outcomes. We limited the search to articles with more than 50 subjects that were published in English from 2002 to 2007. We focused on primary data sources that describe economic evaluations of asthma interventions and their alternatives. Studies that considered only costs with no evidence of equivalence between comparators were excluded from our main findings.

Quality assessment of health economic analyses

We quantitatively evaluated the quality of the asthma health economic studies in a systematic and transparent way by using the 16-item Quality of Health Economic Studies (QHES) validated instrument (17). Two health economists scored each study that met our entry criteria using the QHES. We report the average of the two scores. The total score on the QHES ranges from 0 to 100, with 100 being a perfect quality score. Four quality levels have been established: quality category 1 (total score: 0–25), quality category 2 (total score: 25.1–50), quality category 3 (total score: 50.1–75), and quality category 4 (total score 75.1–100). Below, we have highlighted four main topics to focus the discussion surrounding levels of quality of asthma health economic studies.

Choice of economic analysis (structure, objective, perspective, and time horizon) The consensus in economic evaluations is to include both estimates on costs and consequences of alternatives in order to help inform policy decision making. Cost-effectiveness, cost–benefit, cost-consequence, and cost-utility analyses all include costs and consequences of at least two alternatives. The outcome in the case of asthma cost-effectiveness analyses is generally disease specific such as symptom free days (SFD), exacerbations averted et cetera. The preferred outcome in cost-utility analyses is quality adjusted life years (QALYs). This outcome estimates the alternative specific survival and assigns utility weights ranging from zero (death) to one (perfect health) for each life year. A cost–benefit analysis converts all consequences of the alternatives into a monetary metric by way of assuming a specific willingness-to-pay for a unit of the consequence(s). A net benefit calculation can determine the best policy decision with respect to the evaluated alternatives given the assumed willingness-to-pay. Out of the different approaches, a cost-minimization analysis is the most restrictive because it assumes equality for all of the possible consequences of each of the alternatives compared. The equality assumption in any cost-minimization analysis ultimately biases the results because no two alternatives are exactly alike in terms of each measured and unmeasured consequence. Ideally cost-minimization analyses would show that the benefits of the comparators are the same. However, data that do not show a difference in benefit may only be due to a lack of statistical power.

The QHES instrument measures whether the economic model choice (including structure), the study objective, the perspective, and the time horizon of the analysis were stated/displayed and justified (QHES items 1, 2, 8, 12, 13, and 14). The transparency of the economic analysis is paramount to enable an unbiased peer review process. Justification of the model framework and its assumptions along with the validation of the analysis are key factors for the usability of the model.

Asthma model inputs: costs and effectiveness measures Items (3, 7, 9, 10, and 11) from the QHES instrument assess the transparency and justification of model inputs. The assessment of the type of data abstraction methods and the evidence that underlies the model helps users replicate and validate the analysis. For costs, high quality analyses will describe and justify both the resource utilization and unit costs. Asthma models differ as far as how they estimate health outcomes.

The ideal outcome measure for any CEA captures all of the risks and benefits for each of the comparators. Past asthma specific outcome measures used for cost-effectiveness analyses include SFD, exacerbations averted, improvements in lung function, or asthma quality of life instruments (3, 4). While these outcomes independently capture some of the important aspects of asthma, they do not capture adverse events and other health changes that may occur due to treatment. Symptom free days weights each of the symptoms evenly; a patient who has an exacerbation is counted equally to a patient who has shortness of breath. Another shortcoming of the use of SFD is the lack of description of the symptoms included. Without this list, it is unclear whether two studies reporting cost per SFD are comparable. A standardized questionnaire, such as the Asthma Quality of Life Questionnaire (AQLQ) (18) may provide valid and reliable results across studies, but a one unit change in the AQLQ remains difficult to interpret in terms of one’s willingness-to-pay.

Model uncertainty The uncertainty in economic analyses should be presented in order for decision makers to evaluate the inputs that have a particularly strong influence on the output (e.g. sensitivity analysis) and the need for further research (QHES item 5). Using the decision analytic approach (19, 20), the first step is to decide what alternative to support given the current information in the form of a cost per unit outcome and a willingness-to-pay for the unit outcome. Based on the goal of maximizing the health benefit for a given budget, the mean net benefit answers the question of what alternative to support without regard to the net benefit’s uncertainty. The second step in the decision making approach examines the net benefit’s uncertainty and expected value of information methods may be used in this step to answer questions about the need for further research (21).

Stated conclusions and disclosures The main result of many economic analyses is the incremental cost effectiveness ratio (ICER). This ratio describes the incremental cost of an intervention for each additional unit of benefit. A negative ratio means that one of the alternatives is more effective and less costly than the other. This alternative is considered to dominate the other. A positive ratio means that one alternative is more effective and more costly or less effective and less costly. An intervention is considered cost-effective if it dominates its comparator or in the case of a more effective more costly intervention the ICER falls below the decision maker’s willingness-to-pay threshold.

The justification of the stated conclusions is evaluated with QHES item 15. Disclosing the source of funding for the study can highlight possible biases in the analysis (QHES item 16).

Results

Literature search results

Of the 177 total studies that met our initial search criteria, we included 40 in our review. The most common reasons for excluding studies that met our initial search were: not including both costs and outcomes in their analysis, and not using primary data (i.e. reviews, methods papers, letters, and editorials were excluded).

Results by intervention class

Tables 1–4 contain detailed information on each of the economic evaluations organized by their intervention of interest: policy interventions (Table 1), monotherapy controller medications (Table 2), combination controller medications (Table 3), and add-on medications (Table 4). Of the 40 papers that we reviewed, 14 evaluated a policy intervention (22–35) and the other 26 were pharmaceutical comparisons (36–61). Of the 14 policy evaluations six considered the economic effects of implementing an education program (22, 24, 25, 28, 30, 32). Other studies evaluated management programs (23, 26, 29, 33, 35), guidelines (27), and surveillance (31, 34). All 26 of the pharmaceutical comparisons were of controller medications. Seven studies compared combination vs combination (37, 43, 48, 50, 52, 54, 59). Seven additional studies compared combination vs monotherapy (36, 40, 44, 49, 53, 55, 61). Eight studies compared monotherapy vs monotherapy (38, 41, 42, 45, 46, 51, 56, 58). Four studies compared add-on omalizumab vs usual care (39, 47, 57, 60). There was one study over the time period that considered the costs of reliever therapies, but since no health outcome was reported this study was excluded (62).

Table 1.   Economic evaluations of policy interventions 2002–2007
Author (pub. year)Model type perspectiveComparatorsOutcomesResultsStated conclusionQHES scoreSponsor
  1. AQLQ, Asthma Quality of Life Questionnaire; BFo, budesonide–formoterol combination; Bu, Budesonide; CBA, cost–benefit analysis; CEA, cost-effectiveness analysis; CMA, cost-minimization analysis; CUA, cost-utility analysis; CFC-BDP, chlorofluorocarbon-beclomethasone; ER, emergency room; Fo, formoterol; FP, fluticasone propionate; FPS, fluticasone propionate salmeterol combination; HFA-BDP, hydrofluoroalkane 134a-beclomethasone dipropionate; ICER, incremental cost-effectiveness ratio; ICS, inhaled corticosteroids; IEP, internet-based education program; LM, leukotriene modifier; M, montelukast; MCO, managed care organization; PACI, planned asthma care intervention; PLE, peer leader-based physician behavior change intervention; QALY, quality adjusted life years; QHES, The Quality of Health Economic Studies Instrument; S, salmeterol; Sb, salbutamol; SFD, symptom free day; SPMP, standardized patient management program.

Schermer et al. (26)
(2002)
The Netherlands
Trial based CUA
payer
1. Self-management
2. Usual care
1. QALYSelf-management has 0.015 higher QALYs and 13 Euro lower total costSelf-management is the dominant alternative72Netherlands organization for scientific research
Sullivan et al. (29)
(2002)
USA
Trial based CEA
payer
1. Education program
2. Usual care
1. SFD$9.20/SFDEducation program is cost-effective overall and cost-saving in severe population81.5National Institute of Allergy and Infectious Disease
de Asis et al. (22)
(2004)
USA
Model based CEA
payer
1. Education plus self-management
2. Symptom based self-management
3. No management plan
1. ER visits averted
2. Hospitalizations prevented
$60.57, $31.46/ER averted 1 vs 3, 1 vs 2
$300, $311/hospitalization prevented 1 vs 3,1 vs 2
Interventions are cost-effective55Not reported
Kamps et al. (23)
(2004)
The Netherlands
Trial-based
CMA
unspecified
1. Nurse led care
2. Pediatrician led care
1. Cost onlyNurse led management is 4.1% less expensiveNurse led management is the dominant strategy55.5GSK
Kattan et al. (24)
(2005)
USA
Trial-based CEA
(payer)
1.Environmental intervention
2. Usual care
1. SFD$27.57/SFDIntervention was cost-effective60.5National Institute of Health
Sullivan et al. (28)
(2005)
USA
Trial-based CEA
(societal)
1. PLE
2. PACI
3. Usual care
1. Symptom free days$18/SFD for 1 vs 3
$68/SFD for 2 vs 3
Improvements were realized but at an increased cost72.5Agency for Healthcare Research and Quality
Tschopp et al. (30)
(2005)
Switzerland
Pre–post CEA
(unspecified perspective)
1. Education program1. AQLQ
2. Severity
Cost–benefit ratio = 1.96Education program dominates45Lancardis Foundation
Wild et al. (31)
(2005)
USA
Model based CUA
(employer and societal)
1. Surveillance
2. Usual care
1. QALY
2. SFD
$24 000/QALY
$13.33/SFD
Surveillance is good value for money64National Institute of Health
Runge et al. (25)
(2006)
Germany
Trial-based
CBA
(societal)
1. SPMP
2. SPMP and IEP
3. Usual care
1. Morbidity CostsIntervention costs exceeded reduction in morbidity costs except in moderate/severe subgroupIEP plus SPMP in moderate/severe subgroup54GSK
Simonella et al. (27)
(2006)
Australia
Unspecified model based CEA (government)1. Evidence based guidelines 2. Usual care1. Years lived with disability (YLD) averted$A 3064/YLD avertedOptimal treatment based on guidelines is cost-effective41National Health and Medical Research Council of Australia
Steuten et al. (33)
(2007)
The Netherlands
Model based CUA (societal)1.Management program
2. Usual care
1. QALYThe management program has 0.7 higher QALYs and is Euro 329 less expensiveThe management program dominates usual care89.5Dutch Council for Scientific Research
Karnick et al. (32)
(2007)
USA
Pre–post CEA
(medicaid)
1. Individualized asthma education
2. Reinforced asthma education
3. Reinforced asthma education plus case management
1. Hospitalization
2. ER visits
The programs saved from $4021 to $4503Each of the three interventions were cost-saving53.5Michael Reese Health Trust and the Crown Foundation
Polisena et al. (35)
(2007)
Canada
Retrospective questionnaire CMA
(societal)
1. Self-management program
2. Usual care
1. Nights with symptoms
2. Asthma attacks
Similar health outcomes and $860 per patient per year more expensive The self-management program is not cost-effective66Allergen NCE Inc.
Pinnock et al. (34)
(2007)
UK
Trial-based CEA
(practice)
1. Telephone reviews
2. Face-to-face reviews
3. Usual care
1. Mini-AQLQ
2. ACQ
3. mPEI
4.ABPm
5. Proportion reviewed
No significant difference in mini-AQLQ and ACQ. Telephone reviews had highest mPEI and ABPm and was less expensive than face-to-face.Telephone reviews are cost-effective method for enhancing access49.5Scientific Foundation Board of the Royal College of General Practitioners
Table 2.   Economic evaluations of monotherapy 2002–2007
Author (pub. year)Model type perspectiveComparatorsOutcomesResultsStated conclusionQHES scoreSponsor
  1. For abbreviations see Table 1.

Everden et al. (41)
(2002)
UK
Trial based CEA
payer
1. Eformoterol
2. Salmeterol
1. SFDEformoterol dominates SEformoterol is recommended over S59AstraZeneca
Price et al. (51)
(2002)
UK
Trial based CEA
payer
1. HFA-BDP
2. CFC-BDP
1. SFD
2. AQLQ
HFA-BDP is £5 less expensive and had a higher percentage of SFD (42.4%vs 20%)HFA-BDP is the dominant alternative693M Health Care
Malone et al. (46)
(2003)
USA
Trial based CEA
payer
1. HFA-BDP
2. CFC-BDP
1. SFDHFA-BDP has 7.8 more SFDs and is $309 less expensiveHFA-BDP is the dominant alternative693M Pharmaceuticals
Sullivan et al. (56)
(2003)
USA
Trial based CEA
societal and payer
1. Budesonide
2. Usual care
1. SFD$11.30/SFD
$3.70/SFD societal perspective
Bu appears to be cost-effective79AstraZeneca
Buxton et al. (38)
(2004)
Multinational
Trial based CEA
societal and payer
1. Budesonide
2. Usual care
1. SFDPayer perspective cost-saving in Australia. Other countries ranged from $2.4 to 11.3/SFD. Societal perspective was cost-saving in Canada and SwedenMay be cost-effective depending on WTP75AstraZeneca
Lindgren et al. (45)
(2005)
Sweden and Spain
Trial-based CEA
(societal)
1. Formoterol
2. Salbutamol
1. Exacerbations
2. Symptom free days
Range of ICERS where Formoterol was more effective but was either cost neutral or costs slightly moreFormoterol is favored53.5AstraZeneca
Heaton et al. (42)
(2006)
USA
Retrospective claims database
CBA (medicaid)
1. LM
2. non-LM
1. ER visits
2. Hospitalizations
3. Steroid bursts
LM use was associated with higher costs and no additional benefitLM use was dominated by non-LM use49No reported outside funding
Stempel et al. (58) (2007)
USA
Retrospective claims database CEA (payer)1. FP
2. Montelukast
1. Treatment failure
2. Hospitalization
Treatment failure OR = 0.39
Mean cost $472 more for montelukast
FP dominates montelukast in those 3–17 years old.45.5GSK
Table 3.   Economic evaluations of combination therapy 2002–2007
Author (pub. year)Model type
perspective
ComparatorsOutcomesResultsStated conclusionQHES
score
Sponsor
  1. For abbreviations see Table 1.

Price et al. (53)
(2002)
UK
Model based CEA
payer
1. FPS
2. FP
1. Successfully controlled week
2. QALY
£20.83/successfully controlled weekFPS is cost-effective77.5GSK
Sheth et al. (55) (2002)
USA
Trial based CEA
payer
1. FPS
2. Montelukast
1. SFD
2. >12% improvement in FEV1
$1.69/SFDFPS is cost-effective69GSK
Rosenhall et al. (54)
(2003)
Sweden
Trial-based
CMA
societal
1. BFo
2. Bu and Fo (not in combination)
1. Cost onlyBFo has SEK 1884 lower costsThe single inhaler dominates51AstraZeneca
Jonsson et al. (44)
(2004)
Sweden
Trial based CEA
societal
1. BFo
2. Budesonide
1. SFDSEK 21/SFDBFo improved effectiveness at modest additional cost62.5AstraZeneca
O’Connor et al. (48)
(2004)
Trial based CEA
payer
1. FPS
2. FP and M
1. >12% improvement in FEV1
2. SFD
Higher proportion on FPS achieved >12% improvement in FEV1 and similar SFD with $1 lower daily cost per patientFPS is cost-effective and possibly cost-saving65.5GSK
O’Connor et al. (49)
(2004)
Trial based CEA
payer
1. FPS
2. FP
3. Montelukast
1. >12% improvement in FEV1
2. Control
3. Exacerbations
FPS has greater increase in FEV and fewer exacerbations it also had lower costs per exacerbationFPS is cost-effective42GSK
Price et al. (52)
(2004)
UK
Trial-based
CMA
(healthcare system)
1. BFo (adjustable dose)
2. BFo (fixed dose)
1. AQLQNo difference in outcome.
£65.70/patient year cost savings (1 vs 2)
Adjustable maintenance dose dominates65AstraZeneca
Borker et al. (36)
(2005)
USA
Trial-based CEA
(payer)
1. FPS
2. Montelukast
1. >12% improvement in FEV1
2. Symptom free days
$1.79/day (Outcome 1)
$2.87/day
(Outcome 2)
FPS is cost-effective61.5GSK
Bruggenjugen et al. (37)
(2005)
Germany
Trial-based
CMA
(healthcare system)
1. BFo (adjustable dose)
2. BFo (fixed dose)
1. AQLQNo difference in outcome.
Seventy-one Euros cost savings over 12 weeks (1 vs 2)
Adjustable maintenance dose dominates52.5AstraZeneca
Pieters et al. (50)
(2005)
The Netherlands
Trial-based CEA
(healthcare system)
1. FPS
2. FP and M
1. Successfully treated weeks (improvement in morning PEF >5% over baseline)Twenty-five percentage more successfully treated weeks.
Sixteen percentage lower costs
FPS dominates65GSK
Briggs et al. (61) (2006)
UK
Trial-based
CUA
(payer)
1. FPS
2. FP
1. QALY£7600–13 700/QALY dependent on IC use FPS is cost-effective78.5GSK
Ericsson et al. (40)
(2006)
Germany
The Netherlands
Trial based CEA (societal)1. BFo
2. FP
1. Episode free days
2–7. Six other clinical outcomes
BFo was cost neutral and more effectiveBFo dominates55.5AstraZeneca
Johansson et al. (43)
(2006)
Italy, Germany, France, UK
Trial based CEA (European societal)1. BFo
2. FPS and Sb
1. Exacerbations0.24 vs 0.31 exacerbations/patient year. Cost neutral or cost savings depending on perspectiveBFo may be cost-effective62AstraZeneca
Miller et al. (59)
(2007)
Trial based CEA (societal)1. BFo
2. FPS and Sb
1. Time to first severe exacerbationBFo dominates FPS and SbBFo dominates61.5Unknown
Table 4.   Economic evaluations of add-on therapy 2002–2007
Author (pub. Year)Model type perspectiveComparatorsOutcomesResultsStated conclusionQHES scoreSponsor
  1. For abbreviations see Table 1.

Oba et al. (47)
(2004)
USA
Trial based CEA
payer
1. Omalizumab plus standard care
2. Usual care
1. Successfully controlled day
2. AQLQ
3. Severe exacerbations
$523/additional successfully controlled dayOmalizumab is not cost-effective59Unspecified
Dewilde et al. (39)
(2006)
Sweden
Model based CUA
(societal)
1. Omalizumab
2. Usual care
1. QALYEuro 56 091/QALYAttractive ICER86Novartis
Brown et al. (57)
(2007)
Canada
Model based
CUA (payer)
1. Omalizumab
2. Usual care
1. QALYEuro 31 209/QALYOmalizumab is cost-effective76Novartis
Wu et al. (60)
(2007)
USA
Model based
CUA (societal)
1. Omalizumab plus ICS and quick reliever
2. ICS and quick reliever
1. QALY$821 000/QALYOmalizumab is not cost-effective for most patients with severe asthma75.5National Heart, Lung, and Blood Institute

The most commonly tested interventions were fluticasone propionate (FP) and budesonide (Bu) both as monotherapies and in combination with long-acting beta agonists. Of the reviewed analyses, FP was the drug of interest (three times), comparator (two times) or both (five times). Studies that included FP as both the drug of interest and the comparator tested different combinations such as combination FP and salmeterol vs FP alone. Budesonide was the drug of interest (five times), comparator (zero times) and both (four times). Those studies that included Bu as both the drug of interest and the comparator included two analyses that tested Bu–formoterol fixed dose combination vs adjusted dose.

Results by geographic region

Out of the 40 studies, 21 of them assumed a European perspective and/or included information from European data sources. European perspectives included populations from: UK, Germany, Sweden, Netherlands, Italy, France, Spain, and Switzerland. Outside of Europe, analyses were mostly performed from perspectives in the USA. Other country specific analyses were done for Canada, Australia, and China.

Quality assessment results

The average QHES total score for each health economic study is presented within Tables 1–4. Six studies (15%) had quality category 2, 26 studies (65%) achieved quality category 3, and the remaining eight (20%) studies were scored as the highest quality category 4.

Choice of economic model (objective, perspective, structure, and time horizon) Our search yielded 27 cost-effectiveness analyses, four cost-minimization analyses, two cost–benefit analyses, and seven cost-utility analyses. The study objective was presented in a clear, specific and measurable manner in 38 out of the 40 studies (QHES item 1). The rationale for the stated perspective was only observed in seven studies (QHES item 2). The choice of economic model, including limitations and assumptions, was stated and justified in 18 studies (QHES item 13). Only three out of 40 studies assumed a time horizon that allowed for all relevant and important outcomes (QHES item 8).

Asthma model inputs: costs and effectiveness measures The methodology for data abstraction was stated in 39 of the 40 studies (QHES item 7). Costs and health outcomes were clearly described and appropriate in 35 studies and nine studies, respectively (QHES items 9 and 10, respectively). Fourteen of the studies used SFD as the primary benefit measure. Six studies used exacerbations as their primary benefit measure. Four studies used the percentage of patients achieving >12% improvement in forced expiratory volume in 1 s (FEV1). The AQLQ was used five times. Eight studies reported QALYs, while seven used QALYs as their primary benefit outcome and were therefore considered cost-utility analyses.

Model uncertainty Nineteen of the 40 economic studies included a one-way or multi-way sensitivity analysis. Uncertainty around the outputs was reported in 20 of the 40 studies. Twenty-four studies displayed elements of uncertainty in their analyses per item 5 of the QHES. The need for further research informed from value of information methods was not reported in any of the studies.

Stated conclusions and disclosures Thirty-five of the reviewed studies reported that the intervention of interest was cost-effective whereas 36 studies justified their conclusions (QHES item 15). Only five studies did not come to this conclusion. Of these, one study of leukotriene modifiers concluded that they were dominated by non-LM use (42). Two studies concluded that the additional cost of omalizumab was not worth the observed benefits for most patients with severe asthma (47, 60). In an analysis of two asthma care strategies compared to usual care Sullivan et al. (28) reported ICERs without making an assumption of the willingness-to-pay. Finally, an asthma management intervention was found to have higher costs, but comparable benefits compared to standard-of-care and therefore, were not deemed cost-effective (35).

A statement disclosing the source of funding for the study was reported in 33 of the 40 studies (QHES item 16).

Discussion

Clinical guidelines compared to review findings

We now compare our findings with the recommendations of the Global Initiative for Asthma (GINA). Global Initiative for Asthma guidelines put forth the following six goals for the successful management of asthma irrespective of one’s age: (1) achieve and maintain control of symptoms, (2) maintain normal activity levels, including exercise, (3) maintain pulmonary function as close to normal as possible, (4) prevent asthma exacerbations, (5) avoid adverse effects from asthma medications, and (6) prevent asthma mortality (63). Global Initiative for Asthma’s management approach for all patients is based on a patient’s level of control and age. For all patients irrespective of their level of control and age, GINA recommends asthma education and environmental control. This broad recommendation is in line with the policy intervention cost-effectiveness findings. Namely, most well designed education based interventions are considered good value for money by their authors, but it is often difficult to put into practice such policy interventions. Polisena et al. found that their studied asthma action plan was associated with greater costs compared to a control group without demonstrating superior outcomes (35). Polisena et al. called for further research to understand the most cost-effective asthma education method for children given the participation of parents as proxy reporters and educators. Further, research should focus on the applicability of such policy interventions in order to bridge the gap between policy intervention research and asthma management in practice.

With regard to pharmaceutical treatment, GINA recommends stepping up treatment for those that are not controlled, but once controlled GINA recommends maintaining the lowest step of treatment. Per the GINA guidelines, if patients are not controlled on inhaled short-acting beta-2 agonists (SABA), then inhaled corticosteroids (ICS) should be considered. For patients who are not controlled on ICS, the addition of a long-acting beta-2 agonist (LABA) is recommended. After an unsuccessful trial of ICS and LABA, options include increasing the dose of ICS, or adding a leukotriene antagonist, or theophylline tablet. For those that are not controlled on the highest dose of ICS and LABA combination, a low dose of oral glucocorticosteroids or anti-IgE treatment (omalizumab) is recommended as the final step.

Overall, the findings from this systematic review of economic evidence are in line with the pharmaceutical management recommendations of GINA. The cost-effectiveness of ICS therapy is widely accepted for those whom treatment with an ICS is appropriate. Further, the published literature indicates that ICS plus LABA combinations have favorable value for money (6–8). Leukotriene antagonists have more questionable value for money and this is consistent with their lower emphasis in GINA guidelines. National Institute for Health & Clinical Excellence recently published a health technology assessment of omalizumab (64). This assessment recommends that omalizumab only be used for those with severe unstable allergic asthma and be terminated after 16 weeks of use if the patient has not responded to therapy. The evidence in the literature on the cost-effectiveness of omalizumab is inconclusive, but may suggest that omalizumab is good value for money in responders with severe but uncontrolled allergic disease.

Health economic recommendations (past and present)

Clinical treatment guidelines such as GINA and recent trial based cost-effectiveness models like the Briggs et al. CEA of the Gaining Optimal Asthma Control (GOAL) study (61) have shifted away from an emphasis on lung function or SFD and moved towards a focus on achieving or maintaining control. These reasons along with the lack of quality for primary outcome measures used in the reviewed studies (22.5% of studies had appropriate outcome measures per QHES item 10), supports the need to develop new asthma simulation models that include comprehensive measures like asthma control in the structure (e.g. levels of asthma control as health states of a Markov model) and as an output of cost-effectiveness models. As part of his dissertation research, Campbell and his doctoral committee developed and validated a contemporary asthma policy model that simulates a cohort’s costs and benefits based on transitions through levels of asthma control (65). Although this research focused on an adult population using The Epidemiology and Natural History of Asthma: Outcomes and Treatment Regimens (TENOR) cohort (66) to estimate transition patterns of control, similar analyses could be performed in the less severe or pediatric populations given available data sources in order to simulate the long term costs and benefits of asthma interventions.

Past reviewers of asthma economic evaluations advocated for longer term pragmatic trials to help inform cost-effectiveness analyses (3, 4). We observed that six out of 40 studies included information from trials with over one year of follow-up (24, 26, 28, 29, 38, 56). Only two of these six studies evaluated a pharmaceutical intervention. Therefore, the past reviewers advice remains germane to the present economic evaluations in particular with an emphasis on advocating longer term pragmatic trials of pharmaceuticals. The best available evidence need not come from only randomized controlled trials. Treatment selection and other potential biases notwithstanding, epidemiological and registry databases provide a rich source for economic evaluations due to their long term follow-up and more naturalistic setting compared to controlled trials.

Sullivan and Weiss stressed the importance of using the next best comparator while performing an economic evaluation (4). The clinical guidelines from GINA indicate that most of the reviewed pharmaceutical interventions assumed relevant comparators. However, only one cost-effectiveness analyses was published on the incremental benefit of combination products compared to both of the controller components not taken in a combined formulation. Future study is needed to reduce the uncertainty surrounding the real world incremental cost and benefit estimates of combination products compared to their collective components. The next best comparator for the policy interventions was mostly assumed to be usual care or care prior to the administration of the policy intervention. Temporal trends can bias a pre–post analysis. We advocate for policy interventions to include a control group that is consistent with the next best comparator.

In many situations the outcomes used in cost-effectiveness analyses are chosen based on the availability of data derived in large part from clinical trials. However, clinicians concerned with outcomes for a clinical trial consider different issues compared to an economist designing the CEA. An outcome chosen for use in a CEA must be clinically meaningful and resonate well with the patient population. The health outcome must as completely as possible incorporate all of the relevant aspects of the disease being studied, but not be subjected to double-counting (i.e. capturing the monetary value of that same outcome in the estimate of costs) (67). Since the goal of the CEA is to provide value information, a CEA outcome must be meaningful to decision makers. This means decision makers must assume a willingness-to-pay for the outcome if it is to be used for adoption or reimbursement decisions.

To determine whether an incremental cost of $3/additional patient-day with a >12% improvement in FEV1 is good value for money, the decision maker must have a willingness-to-pay for an additional patient-day with a >12% improvement in FEV1. To determine this willingness-to-pay, often comparisons are made between other treatments measured on the same outcome. However, to make results comparable is not a sufficient justification for the use of an outcome. Most importantly is the ability of an outcome to capture all of the meaningful differences of a treatment. This suggests the need to use generic measures as has been previously recommended (3). Generic measures are also comparable across disease areas allowing decision makers to make higher level comparisons. For the purposes of aiding decisions on adoption or reimbursement, we recommend the use of a generic measure sensitive to treatment differences in asthma patients and standardized across disease areas. However, due to the lack of methods for measuring QALYs in young children and the noted other shortcomings of QALYs (68), we recommend the reporting of generic as well as asthma specific metrics, ideally with a composite measure like asthma control.

Consistent with the findings of Sculpher and Price (3), we found that many of the studies were not explicit about the perspective of their analysis. However, most of the studies listed the resources included in evaluations. With the transparent reporting of resources/costs, it is generally possible to determine the perspective of the analysis. The inclusion of only direct medical related costs suggests a payer perspective. For the most part, the difference between societal costs and payer costs was the inclusion of productivity costs. Although it is helpful for readers to have a clearly stated perspective it is more important to transparently present costs that can help inform or justify the perspective.

We were interested in determining whether Sculpher and Price’s recommendation of using a cost-consequence model was followed. While 17 studies included more than one outcome, only one satisfies the recommendation. Ericsson et al. reported a table of six clinical variables as well as episode free days and a list of direct and indirect costs and therefore meets the requirements of a cost-consequence analysis, but was primarily considered to be a CEA because they also reported ICERs (40). O’Connor et al. reported a table of 10 efficacy variables, however the cost data was insufficient to be classified as a cost-consequence analysis (49). The costs reported by O’Connor et al. were the cost of an exacerbation and the mean cost per exacerbation; total direct medical costs were not included. We recommend taking a cost-consequence approach as well as reporting ICERs in order for the decision makers to determine which outcomes and costs are important from their perspective.

When interpreting the results of a CEA many authors proclaim a treatment to be cost-effective. In order to determine if an intervention is cost-effective a willingness-to-pay must be assumed. However, few studies justified the assumed willingness-to-pay threshold. Good examples of such justification can be found in Oba et al. (47) and Sullivan et al. (56). In the absence of a well justified willingness-to-pay threshold, a neutral reporting of the ICER is preferred.

The rapid emergence of health technology assessment bodies that require evidence of clinical and economic benefit of new technologies for coverage and reimbursement decisions has expanded the market for economic studies of medical interventions. From this review, it is clear that this has impacted the availability of economic studies for asthma and asthma treatments. Current health economic studies in asthma lack appropriate long term time horizons to match the chronic nature of the disease and suffer from using effectiveness measures that do not capture all disease related risks and benefits. We recommend that new asthma simulation models focus on using levels of asthma control in their structure and estimate both long term asthma specific outcomes like well-controlled time and generic outcomes such as QALYs.

Acknowledgments

Two health economists, J. Babigumira and D.E. Spackman, used The Quality of Health Economic Studies instrument to score the asthma health economic publications.

Ancillary