4 Outcomes

The diagnostics advisory committee (section 9) considered evidence from a number of sources (section 10). Full details of all the evidence are in the committee papers.

How outcomes were assessed

4.1 The assessment consisted of a systematic review of the evidence on test performance and clinical‑effectiveness data for ImmunoCAP ISAC, Microtest and standard clinical assessment (comparator).

4.2 Different versions of the ImmunoCAP ISAC tests are named according to the number of allergen components tested, indicated by the number at the end of the name. For example, ImmunoCAP ISAC 112, the most recent version, tests for 112 allergen components. All 15 of the studies included in the review evaluated versions of ImmunoCAP ISAC:

  • ImmunoCAP ISAC 112: 1 study

  • ImmunoCAP ISAC 103: 5 studies

  • ImmunoCAP ISAC 96: 1 study

  • ImmunoCAP ISAC 89: 1 study

  • ImmunoCAP ISAC 51: 1 study

  • ImmunoCAP ISAC 50: 1 study

  • ImmunoCAP ISAC version unspecified: 5 studies.

4.3 In the included studies, 'standard diagnostic work‑up' includes a combination of an allergy‑focused clinical history, and single specific‑IgE testing, skin prick testing or oral‑food‑challenge testing. The combination used in each study varies by study setting and design.

4.4 Of the 15 included studies:

  • 8 studies compared the diagnostic accuracy of ImmunoCAP ISAC with other testing options (single specific‑IgE testing or skin prick testing) to predict clinical reactivity as defined by clinical history, and skin prick test alone or with oral‑food‑challenge testing.

  • 1 study assessed the effects on clinical diagnosis of adding ImmunoCAP ISAC 103 to the standard diagnostic work‑up.

  • 1 study assessed the effects on clinical diagnosis, specific immunotherapy prescription and the value of the additional information gained by adding ImmunoCAP ISAC 103 to the standard diagnostic work‑up.

  • 4 studies assessed the effects on managing the patient's condition by adding ImmunoCAP ISAC to the standard diagnostic work‑up.

  • 1 study looked at the levels of IgE using ImmunoCAP ISAC before and after specific immunotherapy.

4.5 In addition, 2 studies that used ImmunoCAP ISAC to determine sensitisation rates to various allergens were identified. These studies did not meet the inclusion criteria for the systematic review but are described as examples of studies in which positive results do not always link with clinical reactivity.

4.6 Two of the included studies were done in the UK, 12 were done in other European countries and 1 study did not report location. Of the 15 included studies, 4 were funded by or received reagents and consumables or testing services from the company. Five studies were publicly funded, and 6 did not report funding sources.

4.7 The external assessment group did not identify any studies that reported clinical outcomes (that is, allergy symptoms, incidence of acute exacerbations, mortality, adverse events of testing and treatment, healthcare presentations or admissions, health‑related quality of life, patient anxiety, or patient preferences).

4.8 Studies were generally of unclear quality because of limitations in reporting, and 6 studies were reported as conference abstracts only. All studies in the review are considered to be at 'high' or 'unclear' risk of bias. The main areas of bias were participant selection (inappropriate exclusions) and application of testing procedures (variation in testing procedures between study participants and within‑study optimisation of the diagnostic threshold).

4.9 It was not possible to conduct a meta‑analysis of the studies because of the heterogeneity of the included studies and lack of reported data.

Evidence on diagnostic accuracy of ImmunoCAP ISAC

4.10 Of the 8 studies identified, 6 compared the accuracy of ImmunoCAP ISAC with commonly used diagnostic tests (skin prick tests or single specific‑IgE tests) in people with food allergies and 2 studies in people with allergies to aeroallergens. None of the studies used ISAC 112; 2 used ISAC 103; 1 used ISAC 89; 2 used ISAC 50/51; and 3 used unknown ISAC versions.

Diagnosis of food allergy

4.11 De Swert et al. (2012) investigated soy flour allergy. The diagnostic accuracy of an unknown ImmunoCAP ISAC version to measure the soy flour component rGly m 4 was compared with the single specific‑IgE test for the same component and with a skin prick test for soy flour. Cut‑off values were reported separately for each test and oral‑food‑challenge testing was used as the reference standard. ImmunoCAP ISAC had the highest sensitivity (86%; 95% confidence interval [CI] 42% to 100%) but the lowest specificity (80%; 95% CI 28% to 100%). The single specific‑IgE test and skin prick test had similar sensitivity (75%) and specificity (100%).

4.12 Alessandri et al. (2011) investigated allergy to boiled or raw egg. The diagnostic accuracy of ISAC 103, when used to measure 3 individual egg components (Gal d 1, Gal d 2 or Gal d 3), was compared with the accuracy of single specific‑IgE tests (egg yolk or egg white) and compared with the accuracy of skin prick tests (egg white extract, raw egg white, boiled egg white, egg yolk extract, raw egg yolk and boiled egg yolk). Cut‑off values were reported separately for each test and oral‑food‑challenge testing was used as the reference standard. The skin prick test had the highest sensitivity for prediction of allergic response to raw egg white (88%; 95% CI 71.8% to 96.6%), whereas Gal d 3 measured using ImmunoCAP ISAC 103 had the highest specificity (100%; 95% CI 90% to 100%). Results for raw egg were similar to those for boiled egg. In general, single specific‑IgE tests worked the same as a skin prick test, (both measured whole extracts), whereas ImmunoCAP ISAC 103 gave much more variable results for the 3 different components measured. No measure of the overall diagnostic performance of ImmunoCAP ISAC 103 (all components combined) was reported.

4.13 D'Urbano et al. (2010) compared the accuracy of ImmunoCAP ISAC 89, used to measure 2 individual components (Gal d 1 or Bos d 8), with the accuracy of single specific‑IgE tests (egg white or cow's milk). Cut‑off values were reported separately for each test and oral‑food‑challenge testing was used as the reference standard. Specificity was consistent (96%) for both ImmunoCAP ISAC 89 components and for cow's milk and egg white single specific‑IgE tests. Sensitivity values were higher for ISAC 89 components (78% for Bos d 8 and 73% for Gal d 1) than for the corresponding whole allergen single specific‑IgE tests (41% for cow's milk and 27% for egg white). When whole allergen single specific‑IgE tests and ImmunoCAP ISAC 89 were used in series (that is, ImmunoCAP ISAC 89 results were only considered in people with single specific‑IgE negative results), the combined sensitivity was greater than that for single specific‑IgE tests alone (84% compared with 41% for cow's milk allergy and 73% compared with 27% for hen's egg allergy); specificity was 92% in both cases. Ott et al. (2008) compared the accuracy of ImmunoCAP ISAC 51, used to measure 8 individual components (alpha casein, beta casein, kappa casein, Bos d 4, Bos d 5, Gal d 1, Gal d 2, Gal d 4) with the accuracy of single specific‑IgE tests (hen's egg or cow's milk extract) and with the accuracy of skin prick tests (native hen's egg or native cow's milk). Cut‑off values were reported separately for each test and oral‑food‑challenge testing was used as the reference standard. The results were highly variable between tests. The skin prick test had the highest sensitivity for cow's milk allergy (93.6%; 95% CI 78.5% to 99%). The ImmunoCAP ISAC 51 components all had low sensitivity for cow's milk allergy (ranging from 23.9% to 50% for the 5 components assessed). Conversely, all 5 ImmunoCAP ISAC 51 components had high specificity for cow's milk allergy (ranging from 88.4% to 97.7%), whereas the skin prick test had low specificity (48.2%; 95% CI 28.7% to 68%). Single specific‑IgE testing had the highest sensitivity for hen's egg allergy (71.1%; 95% CI 55.7% to 83.6%). All 3 ImmunoCAP ISAC 51 components had low sensitivity (ranging from 17.8% to 57.8%) and high specificity for hen's egg allergy (the individual specificities of the ImmunoCAP ISAC 51 components were 100% for Gal d 4, 86.7% for Gal d 1 and 80% for Gal d 2). Single specific‑IgE testing and skin prick testing had comparable specificity (86.7% and 100% respectively). No measure of the overall diagnostic performance of ImmunoCAP ISAC 51 (all relevant components combined) was reported for either cow's milk or hen's egg allergy.

4.14 Sokolova et al. (2009) investigated milk allergy. The diagnostic accuracy of an unknown ISAC version, used to measure 9 individual components (Bos d 4, Bos d 6, Bos d 7, Bos d 8, casein alpha‑S1, casein beta and casein kappa, Bos d lactoferrin, Bos d 5.0101), was compared with the accuracy of single specific‑IgE tests for 4 allergens (whole milk, alpha‑lactalbumin, beta‑lactoglobulin and casein). For both methods, a positive result was defined as positive for at least 1 component or whole allergen; the cut‑off values used to define positivity for individual components and allergens were not reported. Oral‑food‑challenge testing was used as the reference standard. Both combined ImmunoCAP ISAC testing and combined single specific‑IgE testing had 100% sensitivity, but ImmunoCAP ISAC testing had much higher specificity (91.7%; 95% CI 73% to 99%) than the single specific‑IgE testing (37.5%; 95% CI 18.8% to 59.4%).

4.15 Albarini et al. 2013 investigated hazelnut allergy. The diagnostic accuracy of an unknown ImmunoCAP ISAC version, used to measure 4 individual components (Cor a 1 1010, Cor a 1 0401, Cor a 8, Cor a 9), was compared with the accuracy of single specific‑IgE tests (hazelnut) and with skin prick testing. Cut‑off values were not reported for the ImmunoCAP ISAC test. Oral‑food‑challenge testing was used as the reference standard. Both the skin prick test and the single specific‑IgE test had 100% sensitivity, whereas the ImmunoCAP ISAC components generally had low sensitivity (ranging from 6.3% to 56.3%). However, the ImmunoCAP ISAC components had higher specificity (ranging from 73.7% to 100%) than either single specific‑IgE (21.1%) or skin prick testing (52.6%).

Diagnosis of aeroallergy

4.16 Wohrl et al. (2006) investigated 5 different aeroallergens (house dust mite, cat dander, birch pollen, grass pollen and mugwort pollen). The diagnostic accuracy of ImmunoCAP ISAC 50, used to measure the presence of 1 or more aeroallergens (up to 5), was compared with the accuracy of single specific‑IgE tests of whole allergens. When multiple ImmunoCAP ISAC components were assessed, a positive result was defined as positive for at least 1 component. The cut‑offs for each test were not reported. Skin prick testing was used as the reference standard. The specificity of ImmunoCAP ISAC 50 was high for all aeroallergens investigated, regardless of whether a single component or multiple components were assessed (range 89.9% to 98.1%), and, except for mugwort pollen, was comparable with the specificity estimate for the corresponding whole allergen single specific‑IgE test for all aeroallergens investigated. The sensitivity of ImmunoCAP ISAC 50 was lower than that of single specific‑IgE tests for house dust mite, cat, and mugwort pollen. The sensitivities and specificities of the individual components ImmunoCAP ISAC 50 components were not reported.

4.17 Cabrera‑Freitag et al. (2011) investigated 2 different pollens (grass pollen or Ppratense and cypress pollen or Carizonica). Two cut‑off points (recommended by the companies and ROC optimised) were reported for each test; skin prick test was used as the reference standard. The diagnostic accuracy of ImmunoCAP ISAC 103, when used to measure the 8 components for grass pollen (rPhl p 1, rPhl p 2, nPhl p 4, rPhl p 5, rPhl p 6, rPhl p 7, rPhl p 11, rPhl p 12) was compared with the accuracy of a single specific‑IgE test to measure Ppratense; a positive result was defined as positive for at least 1 component. The sensitivity and specificity for ImmunoCAP ISAC 103 and the single specific‑IgE test were similar, whatever the cut‑off point used. Sensitivity and specificity estimates for individual grass pollen ImmunoCAP ISAC 103 components were not reported. The accuracy of ImmunoCAP ISAC 103 was also used to measure the presence of a single component for cypress pollen (nCup a 1) compared with the accuracy of single specific‑IgE tests to measure Carizonica. The sensitivity estimates for the 2 tests were equal at both cut‑offs (91.7%), but specificity was higher for ImmunoCAP ISAC 103 at both cut‑offs (91.3% and 95.6%) than for the single specific‑IgE test (80.4% to 89.1%).

Evidence on clinical diagnosis using ImmunoCAP ISAC

4.18 Heaps et al. (2014) investigated 110 people with a diagnosis of idiopathic anaphylaxis (based on clinical assessment, skin prick test, single specific‑IgE testing and mast cell tryptase), from 5 UK specialist allergy centres. Study participants were re‑assessed using ImmunoCAP ISAC 103 and clinicians were asked to score the additional information provided. Information from ImmunoCAP ISAC 103 was given the highest score (new heat and digestion stable sensitisations found, which were thought to have a strong association with anaphylaxis) for 22 (20%) participants, but in these 22 people, 168 sensitisations that were not thought to be associated with anaphylaxis were also identified. Also, for a further 35 (32%) participants, the information from ImmunoCAP ISAC was thought to have only identified additional sensitisations that were not thought to be associated with anaphylaxis (322 in total).

Evidence on clinical diagnosis and management using ImmunoCAP ISAC

4.19 Passalacqua et al. (2013) investigated 318 consecutive people with polysensitised (at least 2 positive skin prick tests) respiratory allergy in 6 allergy units in Italy. Participants first gave a clinical history, and had skin prick testing and single specific‑IgE testing (including mites, grass, olive, parietaria, birch, cypress, ragweed, mugwort, cat and dog dander, alternaria and aspergillus), and were then assessed using ImmunoCAP ISAC 103 (no details reported of components assessed or interpretation, but cross‑immunoreactive allergens were considered). Clinicians were asked to review their diagnosis or treatment based on the ImmunoCAP ISAC 103 results and to judge the value of any additional information gained from the test. New information was classified as 'remarkable' if it could not be obtained using standard diagnostic work‑up and could affect the accuracy of diagnosis or the specific immunotherapy prescription. The authors reported that new information related to managing the patient's condition was classified as 'remarkable' for 299 (95%) participants and 'to some extent' (not defined) in 232 (73%) patients. Details of the new information were not reported.

4.20 Passalacqua et al. (2013) also reported detailed information on changes to diagnostic category using 5 classifications when ImmunoCAP ISAC 103 testing was used. The number of people classified as:

  • polysensitised with only 1 clinically relevant sensitisation decreased from 56 to 33

  • true polysensitised with greater than 1 clinically relevant sensitisation decreased from 176 to 117

  • polysensitised with suspected cross‑reactivity increased from 44 to 99

  • sensitised to inhalants and foods increased from 34 to 69

  • non‑classifiable decreased from 8 to 0.

4.21 Passalacqua et al. (2013) also reported changes in specific immunotherapy prescriptions. Eighty‑five people with respiratory allergy, who would not have had specific immunotherapy based on a standard diagnostic work‑up (skin prick test or single specific‑IgE test), were given a new prescription for specific immunotherapy after testing with ImmunoCAP ISAC 103. The existing specific immunotherapy prescription was also changed in a further 3 people with respiratory allergy, after ImmunoCAP ISAC 103 testing. No details of the specific immunotherapy prescriptions or any subsequent clinical outcomes were reported.

Evidence on management using ImmunoCAP ISAC

Discontinuation of restrictive diets

4.22 Two studies investigated using ImmunoCAP ISAC to guide discontinuation of restrictive diets in children with food allergies (Hermansson et al. 2014; Noimark et al. 2014). Both studies were only reported as conference abstracts and so gave only limited study details and results.

4.23 Hermansson et al. (2014) used a database to identify 199 schoolchildren in Harkatie, Finland, having special diets in school catering. The details of this study are confidential at the time of writing this document. No information on clinical outcomes after changes to dietary management was reported.

4.24 Noimark et al. (2012) investigated 12 children selected from people attending an east London allergy clinic (no details of the selection criteria were reported). Participants were investigated using skin prick testing with or without single specific‑IgE testing, and an unspecified version of ImmunoCAP ISAC. The authors reported that ImmunoCAP ISAC testing helped with potential food reintroductions (peanut n=4; soy n=2; wheat n=4), in addition to those indicated by single specific‑IgE testing alone; the numbers of potential reintroductions based on standard diagnostic work‑up (skin prick testing with or without single specific‑IgE testing) were not reported. No details were given on which single specific‑IgE tests or skin prick tests were done or which ISAC components were assessed. The number of food reintroductions that happened after testing, or clinical outcomes after any changes to dietary management were not reported.

Value of additional information

4.25 Luengo et al. (2010) did ImmunoCAP ISAC 103 testing in 55 people who had well‑characterised, poly‑sensitisation (as assessed by skin prick test and single specific‑IgE tests) with various allergies; no details were given on which ImmunoCAP ISAC components were assessed or how these were interpreted. Participating clinicians judged that ImmunoCAP ISAC 103 gave useful new information for managing the condition in 50 (91%) participants. The added value was the ability of ImmunoCAP ISAC to differentiate between protein homologues and so help in differentiating allergens that were cross‑immunoreactive from those responsible for sensitisation. The clinicians considered that it would have been useful to do ImmunoCAP ISAC 103 testing before skin prick testing in 34 (62%) patients, because several protein homologues can be investigated at once using ImmunoCAP ISAC.

Changes in specific immunotherapy prescriptions

4.26 Sastre et al. (2012) investigated 141 people with respiratory allergy (with or without associated food allergy) in 1 allergy outpatient clinic in Spain. Clinicians first assessed the indications for giving patients specific immunotherapy (Olea e, Platanus a, Cupressus a, grass mix, Cynodon d, Phragmites c, Artemisia v, Salsola k and Plantago I) based on clinical history and skin prick test, blind to the results of ImmunoCAP 96 testing (Ole e 1, Cup s 1, Cry j 1, Pla a 1, Pla a 2, Phl p 1, Phl p 5, Phl p 4, Phl p 6, rPhl p 11, Phl p 12, Cyn d 1, Sal k 1, Aln g 1, Bet v 1, Cor a 1.0101, Amb a 1, Art v1, Art v 3 and Par j 2). Clinicians then re‑assessed specific immunotherapy indications based on standard diagnostic work‑up and ImmunoCAP ISAC 96 results. There were disagreements between the prescription based on standard diagnostic work‑up and that based on all information, including ImmunoCAP ISAC, for 79 (54%) of study participants. No details were reported on which specific immunotherapy prescriptions were actually used, or any subsequent clinical outcomes.

Evidence on assessment of IgE levels before and after specific immunotherapy

4.27 Gay‑Crosier et al. (2010) assessed the relationship between change in IgE levels, measured by single specific‑IgE testing, and change in IgE levels, measured by an unspecified version of ImmunoCAP ISAC, before and after a 3‑year course of specific immunotherapy, and the clinicians' evaluation of the benefit of specific immunotherapy. This study included only 9 participants who had a total of 31 courses of specific immunotherapy (no details of diagnosis were reported). The location of this study is not reported. The median specific‑IgE levels, measured by an unknown ImmunoCAP ISAC version, decreased from 5.6 ISU/ml at the beginning of specific immunotherapy to 0.01 ISU/ml at the end of specific immunotherapy and this change correlated with clinicians' judgements of the clinical benefit of specific immunotherapy (Spearman r=0.46; p=0.02). Conversely, allergen‑specific, single specific‑IgE measurements did not show a decrease from the beginning to the end of specific immunotherapy.

Additional studies reporting sensitisation rates that did not meet inclusion criteria

4.28 Two studies, done in Spain, which did not meet the original inclusion criteria for the systematic review, looked at sensitisation rates to various plant food allergens in people with and without previous allergy symptoms. They have been included because they give examples of studies with positive results that do not always link with clinical reactivity

4.29 Pedrosa et al. (2012) assessed 123 children with food allergy, of whom 55 were classified as having peanut allergy and 68 as tolerating peanuts (skin prick test and single specific‑IgE test), and used ImmunoCAP ISAC 103 to assess sensitisation to a range of allergenic components. There were no significant differences between children with peanut allergy and those who could tolerate peanuts in the rates of sensitisation to pathogenesis‑related protein family PR‑10 allergens (Ara h 8, Act d 8, Cor a 1, Gly m 4, Mal d 1, Pru p 1), profilins (Bet v 2, Ole e 2, Hev b 8, Mer a 1, Phl p 12), some lipid‑transfer proteins (Par j 2, Pru p 3), cross‑reactive carbohydrate determinate Ana c 2, or pollens (Ole e 1, Phl p 1).

4.30 The study by Pascal et al. (2015), reported in a pre‑publication manuscript at the time of guidance development, included 130 children with plant‑food allergy and lipid transfer protein sensitisation. They found that sensitisation to a particular plant‑food lipid‑transfer protein, identified with ImmunoCAP ISAC 112, was not always associated with clinical symptoms of allergy to that plant food: 69% (40/58) and 63% (17/27) of children who could tolerate peach and walnut were sensitised to Pru p 3 and Jug r 3 respectively; 60% (21/35) of children without seed or nut allergy were sensitised to storage proteins.

Costs and cost effectiveness

4.31 The external assessment group searched for existing studies on the cost effectiveness of ImmunoCAP ISAC and Microtest, in combination with standard clinical assessment, to help diagnose allergy and predict the grade of allergic reaction. A de novo economic model could not be developed because of the lack of clinical‑effectiveness data.

Systematic review of cost‑effectiveness evidence

4.32 Nine publications from 4 studies were considered eligible for inclusion in the systematic review. All included studies were only reported as conference abstracts, so the methods and assumptions used were largely unclear. Essential inputs to the models in the studies were based on expert opinion or inaccessible references, or no references were reported.

4.33 Hermansson et al. (2014; 2 publications) considered the cost effectiveness of using ImmunoCAP ISAC with standard diagnostic work‑up compared with standard diagnostic work‑up alone, for Finnish school children on a restricted diet because of suspected food allergy (community setting). Data from 24 children drawn from a larger database (including a total of 2,317 school children) were analysed. The results showed an unnecessary restricted diet for 63% of the children, resulting in a cost per avoided unnecessary diet of €480 for ImmunoCAP ISAC compared with standard diagnostic work‑up alone.

4.34 Another study by Hermansson and colleagues (Hermansson et al. 2012; Hermansson et al. 2013) examined the cost effectiveness of an unknown ImmunoCAP ISAC version compared with double‑blind placebo‑controlled food challenge (DBPCFC) and skin prick testing for children with suspected peanut allergy. For this purpose, a Markov model was constructed with a 5‑year time horizon. Health states included non‑allergic and allergic, and mild and severe allergic reactions were modelled as events. The costs were considered for Sweden, the USA and China. The results showed that ImmunoCAP ISAC testing was least expensive whereas skin prick testing was most expensive for all 3 countries. ImmunoCAP ISAC was also most effective, leading to 3.97 quality‑adjusted life years (QALYs) gained, whereas the DBPCFC strategy was least effective (2.54 QALYs). So, ImmunoCAP ISAC dominated both the skin prick test and DBPCFC strategies.

4.35 Glaumann et al. (2013) examined the cost effectiveness of ImmunoCAP ISAC compared with DBPCFC, open (non‑blinded) oral food challenge and skin prick test for children with suspected peanut allergy in Sweden. A Markov model with a 5‑year time horizon was developed, and included non‑allergic and allergic health states, and mild and severe allergic reactions were modelled as events. The results showed that ImmunoCAP ISAC is least expensive whereas skin prick test is most expensive. ImmunoCAP ISAC was also found to be most effective (4.34 QALYs) whereas the oral‑food‑challenge strategy was considered least effective (2.23 QALYs). So ImmunoCAP ISAC dominated all 3 alternative strategies.

4.36 Mascialino et al. (2013; 2 publications) and Hermansson et al. (2012) examined the cost effectiveness of an unknown ImmunoCAP ISAC version with skin prick test compared with a skin prick test alone for Spanish people sensitised to pollen in a complex pollen area. The analysis was based on a Markov model with a 9‑year time horizon, and an assumption that people on specific immunotherapy continue the treatment for 3 years and stay healthy for the following 6 years, or stop specific immunotherapy and move to symptom‑management treatment until year 9. A dataset from 141 people, with allergic rhinoconjunctivitis with or without asthma, sensitised to pollen was analysed. The results showed that adding ImmunoCAP ISAC to a skin prick test reduces specific immunotherapy prescriptions and so results in cost savings compared with skin prick test only (€2,538 compared with €2,608). ImmunoCAP ISAC with a skin prick test was also more effective (7.03 QALYs) compared with skin prick test only (6.88 QALYs). So ImmunoCAP ISAC with skin prick test dominated skin prick test only.

4.37 The study by Rogriguez‑Ferran et al. (2011), reported in a conference abstract, was originally excluded from the review because it did not include effectiveness outcomes, but a description is included for completeness. The study considered the costs of skin prick testing, Phadiatop and ImmunoCAP Rapid used for screening for respiratory allergy in children in primary care. Their results showed that skin prick testing is least expensive (€10–15), followed by ImmunoCAP Rapid (€30) and Phadiatop (€36–67). The authors stated that they believe skin prick testing is cost effective.

Economic analysis

4.38 Because of the lack of data on the clinical consequences of adding multiplex allergen testing to current clinical practice, the external assessment group developed a long‑term cost‑effectiveness model, explored current and potential diagnostic pathways, and built a concept model structure instead of a de novo economic model.

Current and potential diagnostic pathways

4.39 Current clinical diagnostic pathways for people referred for specialist allergy investigation in secondary or tertiary care settings may include skin prick testing, single specific‑IgE testing and an oral‑food‑challenge test if needed, combined with clinical history. A skin prick test is often the first investigation in allergy diagnostics. Based on consultations with clinical experts, it was assumed that single specific‑IgE testing would be done if skin prick test results are not consistent with the clinical history of a patient. Inconsistency can happen if the skin prick test for the most likely allergen (based on clinical history) is negative, or if a skin prick test is positive for an allergen that does not seem to explain the symptoms completely. An oral‑food‑challenge test may also be needed to confirm or rule out allergy to a specific food‑related allergen or allergens. If a skin prick test is not acceptable or practical (for example, in children with atopic eczema), single specific‑IgE testing might be the first‑line investigation, using confirmatory oral‑food‑challenge testing or skin prick testing as needed. It might also be possible to do an oral food challenge based on a skin prick test (and patient history) alone. The exact sequence of testing in clinical practice in secondary or tertiary care settings in the UK is unclear.

4.40 When considering people with difficult‑to‑diagnose allergic disease who have been referred for assessment in secondary or tertiary care settings, multiplex allergen testing may be chosen as a further diagnostic test (assuming that all the allergens of interest are included). Its role would be to identify the allergens to which a patient is sensitive. The possible advantage of the multiplex testing is that it can test for homologous and cross‑sensitive proteins at the same time and so can help the clinician to decide which confirmatory tests are needed. For example, if the test is negative for particular proteins this might rule out the need for oral food challenge, but it is also possible that a negative test with an unclear clinical history might result in a decision to also do an oral‑food‑challenge test. In the proposed pathway, it is unclear if single specific‑IgE testing will always be done before multiplex allergen testing (if single specific‑IgE testing is applicable) or if multiplex allergen testing may also be done instead of single specific‑IgE testing. The most important point is that multiplex allergen testing would be likely to reduce the number of single specific‑IgE tests by ruling out particular allergens, so reducing the need for an oral food challenge.

Concept model structure

4.41 This section describes a model structure that could be used to assess the cost effectiveness of multiplex allergen testing compared with current clinical practice for people with difficult to manage allergic disease in secondary or tertiary care settings. Three comparators would be evaluated in the economic model:

  • ImmunoCAP ISAC testing

  • Microtest testing

  • standard clinical assessment.

4.42 The health economic model would possibly consist of a decision tree and a state‑transition (Markov) model. The decision tree could be used to model the short‑term outcomes, based on test results and the accompanying treatment decision. These outcomes consist of 'at risk of allergic reaction (treated)', 'not at risk of allergic reaction (treated)', 'at risk of allergic reaction (untreated)', and 'not at risk of allergic reaction (untreated)'. Potential adverse events of testing can be also considered in the decision tree.

4.43 The long‑term consequences in costs and QALYs could be estimated using a state‑transition cohort model with a lifetime time horizon. The first health state in the state‑transition model would be determined by the short‑term outcome from the decision tree. The following health states were included in the state‑transition model:

  • at risk of allergic reaction

  • not at risk of allergic reaction or remission

  • allergic reaction (experienced during cycle)

  • death.

4.44 Different types and severities of allergic reactions could be included in the model separately. Given the diversity of allergy reactions, which depend on the type of allergy, separate models would ideally be developed for separate populations, for example, those suspected of having clinical reactivity to an inhaled compared with an ingested allergen.

Model inputs

4.45 To inform the decision tree for the diagnostic pathway, data on the following parameters would be needed, but were not available for the assessment:

  • proportion of people who have a particular test (that is, a skin prick test, single specific‑IgE test, multiplex allergen test with or without oral‑food‑challenge test) as well as the number of skin prick tests with or without single specific‑IgE tests per patient

  • accuracy of the diagnostic pathways (that is, proportion of true positives, false positives, false negatives and true negatives as a result of the combined diagnostic performance of skin prick testing, single specific‑IgE testing or multiplex allergen testing)

  • the treatment decision.

4.46 To inform the long‑term state‑transition model, the following parameters would be needed (all conditional on the test result) but were not available for the assessment:

  • probability of allergic reactions (might be multiple allergic reactions and population specific)

  • probability of remission

  • probability of dying.

Health‑state utilities

4.47 The systematic review of health‑state utilities found 14 studies reporting health‑state utilities for allergic conditions. Ten studies, reported in 13 publications, used the EuroQol instrument, and reported either the EQ‑5D utility score or the visual analogue scale (VAS) score. One study reported utilities collected with the HUI Mark III instrument. Three studies used a direct utility elicitation technique. Ten studies reported on 28 populations: 14 with rhinitis, rhinosinusitis, rhinoconjunctivitis or asthma; 11 with eczema; 2 with food allergy; and 1 with mixed allergies except food allergies.

4.48 Six studies describing 10 populations comparing health‑state utility scores for people with and without allergic disease were found. The evidence on utility values for allergic conditions in the UK population was limited, and no utility values for food allergies were found. For seasonal allergic rhinoconjunctivitis, EuroQol VAS scores from Pitt et al. (2004) or EQ‑5D scores from a European study (Poole et al. 2004; Bachert et al. 2007; Currie et al. 2014) could be used. Stephens et al (2004) used standard gamble to get utility values for atopic eczema in UK children. Only Stephens et al. (2004) reported utilities according to the degree of severity of the allergic conditions. Utility values for complications of allergies, such as anaphylactic shock, could not be found in the literature, apart from the assumption by Armstrong et al. (2013) that the impact of anaphylactic shock on quality of life was equal to 0 utility for a maximum duration of 9 days.

Resource use and costs

4.49 To estimate the costs of the individual tests, a detailed cost calculation was done that considered test costs, capital costs (if applicable), service and maintenance costs, and personnel costs for doing and interpreting the tests. For ImmunoCAP ISAC and Microtest testing, minimum and maximum prices were calculated and then averaged. For ImmunoCAP ISAC testing, the main differences between the minimum and maximum prices were due to the difference in time (5–60 minutes) needed to interpret the test results. This was the same for Microtest testing, but the range was smaller (5–10 minutes). It was also assumed that the sample for Microtest testing would be sent to Microtest Dx where the test would be done (most conservative scenario), whereas ImmunoCAP ISAC testing would be done at the service provider's laboratory. So for ImmunoCAP ISAC testing, capital costs were included whereas it was assumed that these costs for Microtest testing would be included in the test costs. Capital costs were annuitised using a cost discount rate of 3.5%.

4.50 Additional costs that would be considered in a long‑term cost‑effectiveness analysis might include the costs of specific immunotherapy, health‑state costs for being at risk of allergic reaction, health‑state costs for having had an allergic reaction, and health‑state costs for adverse events associated with testing. These costs are likely to be very specific for the population to be considered. Different types of specific immunotherapy might also be provided within a specific population. So, the specific type(s) of specific immunotherapy prescribed and the specific immunotherapy duration would be needed to calculate these costs.

Base‑case results

4.51 In the base case, a cost comparison of 3 diagnostic strategies was assessed: ImmunoCAP ISAC compared with Microtest compared with the standard diagnostic pathway without multiplex allergen testing.

4.52 The cost analyses were carried out using 2‑way threshold analyses for single specific‑IgE and oral‑food‑challenge tests in addition to ImmunoCAP ISAC or Microtest because the proportion of people having these tests was unclear. Specifically, in pairwise comparisons of 2 test strategies, the minimal reduction (that is, the threshold) in proportions of single specific‑IgE and oral‑food‑challenge tests that was needed for the most expensive test strategy to become cheaper than the alternative test strategy was identified, assuming that everything else stayed equal. Here, 100% for both tests was defined as all people have 8 single specific‑IgE tests on average and all people having on average 1 oral‑food‑challenge test. So, for example, if it was assumed that using multiplex allergen testing would result in no single specific‑IgE testing then this would imply a 100% reduction in single specific‑IgE testing compared with the standard diagnostic pathway. Given that multiplex allergen testing was more costly than single specific‑IgE testing, threshold analysis could then show what percentage reduction in oral‑food‑challenge tests would be needed to give the multiplex allergen diagnostic pathway the same cost as the standard diagnostic pathway. But if it was assumed that there was no reduction in single specific‑IgE testing by using multiplex allergen testing, this would result in a different threshold for the percentage reduction in oral‑food‑challenge tests needed to give the multiplex allergen pathway the same cost as the standard diagnostic pathway.

4.53 The following assumptions were made in the base case:

  • Number of allergens by skin prick test per patient: 8.

  • Cost of skin prick test per patient: £62.28.

  • Number of allergens by single specific‑IgE testing per patient: 8.

  • Cost of single specific‑IgE test per patient: £136.37.

  • Cost of oral‑food‑challenge test: £570.

  • Minimum cost per ImmunoCAP ISAC 112 test: £154.41.

  • Maximum cost per ImmunoCAP ISAC 112 test: £284.60.

  • Minimum cost per Microtest test: £140.37.

  • Maximum cost per Microtest test: £173.33.

4.54 It was assumed that everything, except the number of single specific‑IgE tests and the number of oral‑food‑challenge tests, stayed equal for all test strategies (including the proportion of people having any skin prick test). Skin prick testing is considered a simple, safe and quick test (giving results within 15–20 minutes) that is often used as the first‑line investigation in allergy diagnostics. ImmunoCAP ISAC is intended to be used with standard clinical assessment.

4.55 The base‑case analysis showed that for ImmunoCAP ISAC and Microtest to be cost saving compared with the standard clinical assessment, the absolute proportion of oral‑food‑challenge tests should be reduced by at least 15% and 4% respectively (for example, from 50% to 35% or from 50% to 46% respectively) if there was a 100% reduction in single specific‑IgE tests (that is, from 100% to 0%). On the other hand, if there was no reduction in the proportion of single specific‑IgE tests (assuming an average of 8 tests per person), the reduction in oral‑food‑challenge tests should be at least 39% and 28% for ImmunoCAP ISAC and Microtest respectively. Also, for ImmunoCAP ISAC compared with Microtest, the proportion of oral food‑challenge‑tests for ImmunoCAP ISAC should be reduced by at least 11% if there was no reduction in the proportion of single specific‑IgE tests. When assuming no reduction in the proportion of oral‑food‑challenge tests, the proportion of people having an average of 8 single specific‑IgE tests for ImmunoCAP ISAC should be reduced by at least 44%.

Analysis of alternative scenarios

Scenario analysis 1: Assumption on the number of days that the LuxScan 10K reader is used

4.56 The LuxScan 10K reader (scanner recommended for measuring the fluorescence of ImmunoCAP ISAC) might be used for other purposes, so scenario analysis 1 explored the impact of using the reader 253 days per year. This reduced the cost of ImmunoCAP ISAC testing to £201.91 per patient tested, a decrease of £18. At this reduced cost, for ImmunoCAP ISAC testing to be cost saving compared with standard clinical assessment or Microtest testing, the proportion of single specific‑IgE and oral‑food‑challenge tests would need to be as follows:

Compared with standard clinical assessment:

  • The proportion of oral‑food‑challenge tests should be reduced by at least 11% (for example from 50% to 39%) if there was a 100% reduction in single specific‑IgE tests.

  • The proportion of oral‑food‑challenge tests should be reduced by at least 35% if there was no reduction in single specific‑IgE tests.

    Compared with Microtest testing:

  • The proportion of oral‑food‑challenge tests should be reduced by at least 8% if there was no reduction in the proportion of single specific‑IgE tests.

  • The proportion of single specific‑IgE tests should be reduced by at least 33% if there was no reduction in the proportion of oral‑food‑challenge tests.

Scenario analysis 2: Assumption that Microtest is run at service provider's laboratory

4.57 Microtest testing might be done at the service provider's laboratory instead of at the Microtest Dx laboratory (as assumed in the base‑case analysis) so scenario analysis 2 explored the impact of this. The cost of Microtest testing reduces by £7 to £149.37 per patient tested in this scenario. At this reduced cost, for Microtest to be cost saving compared with standard clinical assessment or ImmunoCAP ISAC, the proportion of single specific‑IgE tests and oral‑food‑challenge tests would need to be as follows:

Compared with standard clinical assessment:

  • The proportion of oral‑food‑challenge tests should be reduced by at least 2%, if there was a 100% reduction in single specific‑IgE tests.

  • The proportion of oral‑food‑challenge tests should be reduced by at least 26%, if there was a no reduction in single specific‑IgE tests.

    Compared with ImmunoCAP ISAC:

  • The proportion of oral‑food‑challenge tests should be reduced by at least 15% if there was no reduction in the proportion of single specific‑IgE tests.

  • The proportion of single specific‑IgE tests should be reduced by at least 39% if there was no reduction in the proportion of oral‑food‑challenge tests.

Scenario analysis 3: Assumption on the number of allergens tested by single specific‑IgE testing varies

4.58 The number of allergens that may be tested by single specific‑IgE testing is uncertain, so the third scenario analysis explored the impact of varying the number of allergens tested using single specific‑IgE testing (base‑case value: 8 allergens tested per person). For testing with ImmunoCAP ISAC or Microtest to be cost saving compared with standard clinical assessment or each other, the proportion of single specific‑IgE tests and oral‑food‑challenge tests would need to be as follows:

Assuming 1 allergen being tested:

  • For ImmunoCAP ISAC compared with standard clinical assessment, the proportions of oral‑food‑challenge tests should be reduced by at least 35% if there was a 100% reduction in single specific‑IgE tests.

  • For Microtest compared with standard clinical assessment, the proportions of oral‑food‑challenge tests should be reduced by at least 24% if there was a 100% reduction in single specific‑IgE tests.

  • For ImmunoCAP ISAC compared with Microtest, the proportion of oral‑food‑challenge tests for ImmunoCAP ISAC should be reduced by at least 8% if there was a 100% reduction in single specific‑IgE tests.

    Assuming 20 allergens being tested:

  • For ImmunoCAP ISAC compared with standard clinical assessment, the proportion of single specific‑IgE tests should be reduced by at least 64% assuming no reduction in oral‑food‑challenge tests.

  • For Microtest compared with standard clinical assessment, the proportion of single specific‑IgE tests should be reduced by at least 46% assuming no reduction in oral‑food‑challenge tests.

  • For ImmunoCAP ISAC compared with Microtest, the proportion of single specific‑IgE tests for ImmunoCAP ISAC should be reduced by at least 18% if there was no reduction in oral‑food‑challenge tests.

Scenario analysis 4: Assumption on the cost of oral‑food‑challenge tests being £256

4.59 The cost of oral‑food‑challenge in the NHS in England may be lower than that used in the base‑case analysis, so scenario analysis 4 explored the impact of using a lower price of £256 instead of £570. At this reduced cost, for testing with ImmunoCAP ISAC or Microtest to be cost saving compared with standard clinical assessment or each other, the proportion of single specific‑IgE tests and oral‑food‑challenge tests would need to be as follows:

ImmunoCAP ISAC testing compared with standard clinical assessment:

  • The proportion of oral‑food‑challenge tests should be reduced by at least 32% if there was a 100% reduction in single specific‑IgE tests.

  • The proportion of oral‑food‑challenge tests should be reduced by at least 86% if there was no reduction in single specific‑IgE tests.

    Microtest testing compared with standard clinical assessment:

  • the proportion of oral‑food‑challenge tests should be reduced by at least 8% if there was a 100% reduction in single specific‑IgE tests.

  • the proportion of oral‑food‑challenge tests should be reduced by at least 61% if there was no reduction in single specific‑IgE tests.

    ImmunoCAP ISAC testing compared with Microtest testing:

  • the proportion of oral‑food‑challenge tests for ImmunoCAP ISAC should be reduced by at least 24% if there was no reduction in single specific‑IgE tests

  • the proportion of oral‑food‑challenge tests for ImmunoCAP ISAC should be reduced by at least 46% if there was no reduction in oral‑food‑challenge tests.

Threshold analyses

4.60 For the situation in which ImmunoCAP ISAC or Microtest are used as replacements for single specific‑IgE testing (rather than as an add‑on test), a threshold analysis was done to examine the minimum number of allergens tested with single specific‑IgE tests so that single specific‑IgE testing is equally or more expensive than multiplex allergen testing, assuming that everything else stays equal. This analysis was also done for skin prick testing. In these analyses, it was assumed that there was no reduction in oral‑food‑challenge testing with multiplex allergen testing. For standard clinical assessment to be as expensive as the ImmunoCAP ISAC and Microtest pathways, the minimum number of allergens tested using single specific‑IgE tests were 13 and 10 respectively. This means that, if multiplex testing replaced single specific‑IgE testing then it would have to replace at least 13 or 10 tests respectively to be cost saving. For skin prick testing these numbers were 39 and 27 respectively.

  • National Institute for Health and Care Excellence (NICE)