European Respiratory Society Guidelines for the Diagnosis of Asthma in Adults

Recommendation

• The TF recommends performing spirometry to detect airway obstruction as part of the diagnostic work-up of adults aged 18 years with suspected asthma (strong recommendation for the test, low quality of evidence)

Remarks

• An FEV₁/FVC <LLN or <75%, higher than the commonly utilised 70% threshold, should be considered supportive of an asthma diagnosis and should prompt further testing (see Algorithm)

• A normal spirometry does not exclude asthma

Background

Spirometry is a non-invasive physiological test, performed since the 19th Century, that measures the volume and flow of air during inhalation and exhalation. A standardised procedure for performing spirometry has been published by the ERS and the ATS [8]. The ratio of the forced expiratory volume in the first second to the forced vital capacity (FEV₁/FVC) is an index reflecting airway obstruction. The TF assessed the FEV₁/FVC ratio to determine whether it could help in the diagnosis of asthma.

Review of the evidence

Our literature search identified 11 potentially relevant studies of which four were suitable to be included (supplementary tables 3a and b) [18–21], all performed in secondary care that assessed the accuracy of the FEV₁/FVC ratio to predict the probability of asthma ascertained by either BdR of 12% and 200mL or 15% reversibility, methacholine BHR (PC20-M <8–16 mg·mL⁻¹), or 20% PEF variability over a two-week period (supplementary table 4).

In their cross-sectional study, Hunter et al., recruited 89 patients (baseline FEV₁>65% of predicted) from primary care with a prior label of asthma, but 20 patients were found to have an alternative explanation for their asthma [19]. Of those diagnosed as asthma (n=69), 46% were receiving concomitant inhaled corticosteroid (ICS) while undergoing diagnostic testing. Asthma was diagnosed based on symptoms combined with at least one of the following: BdR of 15% after 200 µg salbutamol, PC20-M <8 mg·mL⁻¹, or PEF variability of 20% over a 15-day period. A predetermined cut-off of the FEV₁/FVC ratio at 77% based on the 95% LLN found in healthy subjects, yielded a sensitivity and specificity of 61% and 60%, respectively [19]. Stanbrook et al., retrospectively analysed lung function tests of 500 patients referred to secondary care and found a FEV₁/FVC cut-off value of <90% predicted had 53% sensitivity and 27% specificity to identify a positive methacholine test (PC20-M <8 mg·mL⁻¹) [21].

Two retrospective studies conducted in secondary care investigated the best threshold by constructing ROC curves. In 270 patients, where half of the patients were treated with ICS, Bougard et al., found an AUC of 0.62 and a FEV₁/FVC cut-off value at 77% in the training cohort and an AUC of 0.68 with a FEV₁/FVC cut-off value of 79% in the validation cohort [18]. Nekoee et al., recruited steroid-naïve patients (n=702) with symptoms suggestive of asthma, including 19% of current smokers and displaying an average baseline FEV₁ of 95% predicted [20], and found sensitivity ranged from 0.51 to 0.69 with specificity ranging from 0.28 to 0.76 (GRADE table 6, EtD supplementary table 3b)

Justification of the recommendation

Physiological airflow obstruction and fluctuation of airway caliber, that is usually reversible are recognised as hallmarks of asthma. Though the quality of evidence was low, the TF recommends spirometry as the first test to be conducted in the diagnostic work-up. Over-diagnosis, which occurs in approximately 30% of patients with asthma diagnosed in primary care, occurs in part because spirometry in not performed and has a substantial risk of harm due to inappropriate treatment side-effects, costs, and lack of proper diagnosis [4]. Therefore, a strong recommendation can be made despite low quality of evidence. Spirometry is readily available both in primary and secondary care, even though it might not be used sufficiently in primary care. Our research found the FEV₁/FVC cut-off providing the best combination of sensitivity and specificity is close to 75%, a threshold well above the 70% threshold generally recognised as a marker of airway obstruction. However, sensitivity at a cut-off of 75% is close to 50% and much too low to rule out asthma. Likewise, at this cut-off, specificity remains below 80% making spirometry alone insufficient to rule in asthma with confidence.

Patient perspective

Spirometry is non-invasive and generally well-accepted by the patient. The reproducibility of the measure, however, depends on the skill of the operator and the participation of the patient. Indeed, the role of the operator is crucial in putting patients at ease and guiding them through each step [22], where patients value their role: “a sympathetic, helpful and considerate nurse can do wonders during this test”. Patients are also interested in knowing about their breathing performance and individual test results, and how they relate to averages for their age, height and weight.

Key unanswered questions

We know that FEV₁/FVC ratio declines with age so fixing a threshold is inappropriate to apply across a population with varying ages [23]. We did not find any study that expressed the FEV₁/FVC ratio as <LLN and calculated its prediction value. There is an urgent need for prospective studies in both primary and secondary care that would combine specific symptoms with spirometry indices expressed as LLN to make a diagnosis of asthma.

Recommendation

• The TF suggests not recording PEF variability as the primary test to make a diagnosis of asthma diagnosis (conditional recommendation against the test, low quality of evidence)

Remarks

• PEF may be considered if no other lung function test is available including spirometry at rest and bronchial challenge testing

• PEF should be monitored over a two--week period and a variation of >20% considered as supportive of asthma diagnosis

• PEF variability <20% does not rule out asthma

• PEF may be especially useful to support a diagnosis of occupational asthma

Background

Peak expiratory flow (PEF) measurement over a few weeks has been advocated as a test to diagnose asthma for several decades as the tool to evidence airway caliber fluctuation associated with poor asthma control [24].

Review of the evidence

Our literature search identified 15 potentially relevant studies of which six studies (one retrospective, five prospective) met the inclusion criteria (supplementary tables 5a and b) [19, 25–29]. Five studies (three in primary care, two in specialist care referred from primary care) addressed symptomatic patients without any prior investigations or diagnosis, and one study included patients diagnosed with asthma in primary care but referred to secondary care (supplementary table 4).

All the studies assessed the diagnostic performance of pre-specified thresholds of PEF variability with thresholds most often set at 15% or 20% over a two-week period. The way to calculate the PEF variability has a great impact on diagnostic performance with the greatest sensitivity when variability is the difference between the greatest and the lowest value divided by the lowest [26]. Overall, PEF variability provided a highly variable sensitivity ranging from 5% until 93% while the specificity was ranging from 75% to 100% (GRADE table 7, EtD supplementary table 5b). The lower the variability required to define asthma, the greater the sensitivity.

Justification of the recommendation

Results from studies on PEF variability demonstrate a highly variable sensitivity, with lower sensitivities in studies where the prevalence of asthma was low. The most common method used to calculate PEF variability is the average daily amplitude percentage mean with a cut-off of 20%, however, alternatives such as the % amplitude highest PEF may just be as accurate and not require calculating the daily mean PEF [26, 30]. Completion of accurate peak flow diaries was poor, with results as low as 50% in one study [26], challenging the reliability, accuracy and feasibility of home PEF recordings. In addition, reliability of PEF measurement may be even lower in real life than in a research setting. A very recent study has shown that measurement over 5 days compared to 14 days improved diary completion rate from 15% to 94% with no loss of accuracy [30]. In the absence of spirometry defined obstruction and significant BdR, PEF can be monitored over a two-week period particularly if access to bronchial challenge is limited. In the context of a patient with symptoms suggestive of asthma, a positive PEF variability of >20%, that is reliably performed, has a high positive predictive value. Lowering the cut-off at 15% to 10% would increase the sensitivity at the expense of specificity. Thus, PEF monitoring may be of higher value to diagnose asthma in patients with highly variable day-to-day symptoms, where variable airflow obstruction might be easily detected, or in patients with suspected occupational asthma. However, we caution that lack of PEF variability does not rule out asthma and further objective testing should always be performed. Spontaneous and ICS induced FEV₁ variability over time could also have been considered. However, we decided not to conduct a separate PICO due to the limitation of the ERS framework to eight PICO questions, and the low number of longitudinal studies that have evaluated FEV₁ variability over time. Having said that we mention a recent study looking at between visit FEV₁ variability, that provided similar results to PEF, with a poor sensitivity but a high specificity in the order to diagnose asthma [31].

Patient perspective

PEF variability testing has advantages of being cheap and easy to perform even in low-resource settings. Although no undesirable effects of PEF testing were documented, the TF recognises that for some patients performing home PEF twice daily for at least two weeks may become unrewarding and time-consuming, reinforcing the need for proper education and training. Patients may prefer undertaking a one-stop BdR undertaken in 15 min, which if positive would potentially prevent delay in diagnosis and potential treatment. Hence, if available, the TF advises BdR testing, particularly in primary care above PEF testing.

Key unanswered questions

PEF variability between 15% to 20% clearly lacks sensitivity to diagnose asthma compared to bronchial challenge and we advocate prospective studies to establish the threshold of variability that best correlates to a positive bronchial challenge test.

Recommendation

• In patients suspected of asthma, in whom the diagnosis is not established based on the initial spirometry combined with bronchodilator reversibility testing, the TF suggests measuring the fraction of exhaled nitric oxide (FeNO) as part of the diagnostic work-up of adults aged >18 years with suspected asthma (conditional recommendation for the intervention, moderate quality of evidence)

Remarks

• A cut-off value of 40 ppb offers the best compromise between sensitivity and specificity while a cut-off of 50 ppb has a high specificity >90% and is supportive of a diagnosis of asthma

• A FeNO value <40 ppb does not rule out asthma and similarly high FeNO levels themselves do not define asthma

• FeNO values are markedly reduced by smoking, impaired airway calibre, treatment with ICS or anti-IL4/IL13-receptor alpha antibody

Background

Nitric oxide is a gas measurable in exhaled air by chemoluminescence or an electrochemical method, where the measurement has been standardised and endorsed by the ERS/ATS [32]. The fraction of exhaled nitric oxide (FeNO) measures allergic airway inflammation mediated through allergen-driven IL-4 and IL-13 effects on airway epithelial cells and is associated with the extent of airway eosinophilic inflammation [33]. FeNO is dependent on height, gender, atopy and smoking status and airway caliber [34]. FeNO is raised in patients with asthma compared to healthy subjects, and in asthma patients with allergic rhinitis compared to those without rhinitis. FeNO is exquisitely sensitive to ICS, with a sharp decrease in levels a few days after starting treatment [35]. Certain biological treatments, which can be given for other than severe asthma, eg. nasal polyposis, also reduce FeNO [36].

Review of the evidence

Our literature search identified 31 potentially relevant studies of which 21 studies met the inclusion criteria (supplementary tables 6a and b) [9, 20, 37–56]. We exclusively selected studies that measured FeNO at an expiratory flow of 50 mL·sec⁻¹ (supplementary table 7), thus excluding two studies where FeNO was measured at a higher flow [57, 58]. Optimal FeNO cut-off values for a diagnosis of asthma in adults ranged from 15 ppb to 64 ppb, with sensitivity values ranging from 29% to 79% and specificity values ranging from 55% to 95%. The high variability observed across the studies reflected differences in patient inclusion criteria in demographics such as smoking and atopy status, or concurrent ICS treatment during assessment.

Katsoulis et al., found a FeNO cut-off of 32 ppb for the whole population of patients with symptoms suggestive of asthma (n=112), but a low cut-off of 11 ppb when selecting actively smoking asthma patients [46]. Nekoee et al., (n=720) found a FeNO cut-off value of 36 ppb yielded a sensitivity of 30% and a specificity of 85% [20]. The TF derived the sensitivity and specificity for fixed FeNO cut-offs where it was provided by the study authors. A lower cut-off of 25 ppb provided sensitivity and specificity of 0.53 (95% CI: 0.33 to 0.72) and 0.72 (95% CI: 0.61 to 0.81) respectively (GRADE table 8a), where a higher 50 ppb cut-off value ranged from 0.19 to 0.56 and 0.77 to 0.95, respectively (GRADE table 8c). A cut-off of 40 ppb yielded a sensitivity of 0.61 (95% CI: 0.37–0.81) and a specificity of 0.82 (95% CI:0.75–0.87) (GRADE table 8b).

Justification of the recommendation

Measuring FeNO is a point-of-care method that may be particularly useful in both primary and secondary care [59], although it is not yet considered for reimbursement in most of European countries. A cut-off value above 40–50 ppb yields a high specificity (between 0.75 to 0.95), to rule in a diagnosis of asthma with confidence. However, the poor sensitivity (between 0.19 to 0.81) does not allow asthma to be ruled out, for values below 40 ppb. Although the TF recommends using FeNO to help in the diagnosis of asthma, we make it clear that high FeNO levels do not define asthma. High FeNO levels may be observed in patients with eosinophilic chronic bronchitis, allergic rhinitis or eczema who may deny any asthma symptoms and do not show bronchial hyperresponsiveness [3]. Additional factors such as training, cost of device and sensors, and local reimbursement policies may limit use in primary care.

Patient perspective

FeNO is a non-invasive, quick and relatively cheap measurement well accepted by the patient. It is worth noting that some patients are unable to adequately control their expiratory flow to provide a value. Given the strong influence of ICS on FeNO level it is better to measure it when patients have not taken this medication, whenever possible. The cost of paying for FeNO by patients in settings where reimbursement is not available may limit use.

Key unanswered questions

Given the many factors influencing FeNO values, prospective studies are needed defining the best cut-off in different categories of patients taking into account smoking and atopic status.

Recommendation

• The TF suggests not measuring blood eosinophil count to make a diagnosis of asthma (conditional recommendation against the test, low quality of evidence)

Remarks

• Blood eosinophil count does not define asthma but rather contributes to phenotyping

Background

Eosinophilic inflammation is a feature often found, but not specific of asthma, irrespective of the status of atopy [60], that may contribute to asthma exacerbation [61]. Although analysis of the airway compartment by sputum or bronchoalveolar lavage is preferred, measuring the systemic component of eosinophilic inflammation by blood sampling may be a practical alternative. We investigated whether measuring blood eosinophil count (BEC) may help in the diagnosis of asthma.

Review of the evidence

Our search identified 24 potentially relevant studies of which five studies (four prospective, one retrospective) were suitable for analysis (one in primary care, four in specialist care) (supplementary tables 8a, b and 9). Hunter et al., assessed the value of a BEC cut-off of 6.3%, taken as the upper limit of the normal range [19]. Popovic et al., investigated 195 patients with symptoms of dyspnea where asthma was diagnosed in 141 subjects based on a symptom questionnaire and significant BdR (no threshold was provided) and assessed the value of eosinophilia without providing any cut-off [62]. In a prospective observational study, Yurdakul et al., included 123 participants, where 60 had asthma, 40 pseudo-asthma and 23 were healthy. Asthma was diagnosed based on reported symptoms associated with either BdR of 15%, PC20-M <8 mg·mL⁻¹, or PEF diurnal variation of at least 20%. Nearly half (48%) of patients with asthma were receiving ICS before testing. No cut-off for BEC was provided [63]. Two studies constructed ROC curves to determine the performance of BEC and the best BEC cut-offs. Tilemann et al., prospectively investigated 210 patients recruited in primary care with symptoms suggestive of asthma, where 5% were receiving ICS treatment. Asthma was confirmed in patients with BdR of 12% and 200mL improvement, or PC20-M <16 mg·mL⁻¹. The AUC-ROC (95% confidence intervals (CI)) for BEC was 0.60 (0.52–0.68) with an optimal cut-off of 4.1% in the Tilemann's study [54], and 0.58 (0.54–0.62) with a cut-off of 4.4% in the Nekoee's study [20]. Overall, sensitivity ranged between 0.15 and 0.59 while specificity was between 0.39 and 1 (GRADE table 9, supplementary EtD table 8b). A 95% specificity was obtained for a BEC cut-off of 5.9% in Nekoee's study [20].

Justification of the recommendation

BEC lacks sensitivity to diagnose asthma, with sensitivities ranging between 21% to 59% in the reported studies. A BEC does not provide immediate results at the time of the consultation in order to directly help the clinician, although as blood leukocyte differential is a test frequently performed for several indications in routine practice, it may be that a previous test is available at the time of the consultation. BEC cut-offs above 4% and 6% have a specificity greater than 80% and 95% respectively and may help the clinician to be confident in their diagnosis in patients with suggestive symptoms.

Patient perspective

Performing a blood leukocyte differential is relatively cheap, minimally invasive, although some patients may be anxious of venipuncture.

Recommendation

• The TF suggests not measuring total serum IgE to make to make a diagnosis of asthma (conditional recommendation against the test, low quality of evidence)

Remarks

• Total serum IgE does not define asthma but rather contributes to phenotyping

Background

Immunoglobulin (Ig)-E is a key component in mediating type-1 hyper-sensitivity reaction resulting in degranulation of mast cells and basophils, which can lead to symptoms of asthma [64]. There are non-IgE mediated events that can also trigger symptoms. IgE mediated mechanisms can also occur in non-atopic patients [65, 66], where elevated levels of total serum IgE have been reported [67]. We investigated whether assessing total serum IgE could help in the diagnosis of asthma.

Review of the evidence

Our search identified 26 potentially relevant studies of which four studies were considered suitable for analysis (supplementary tables 10a and b), which have been previously described above (supplementary table 8) [20, 54, 62, 63]. Popovic and Yurdakul assessed the value of a predetermined (but not provided) cut-off while Tilemann and Nekoee constructed ROC curves [63]. The AUC-ROC (95% CI) was 0.58 (0.50–0.66) with a cut-off of 90 Ku·L⁻¹ in Tilemann's study [54], and 0.57 (0.53–0.61) with a cut-off value of 132 KU·L⁻¹ in Nekoee's study [20]. Overall, sensitivity ranged between 0.33 and 0.51 and specificity between 0.72 and 0.85 (GRADE table 10, supplementary EtD table 10b). Using a cut-off of 584 Ku·L⁻¹, 95% specificity was obtained [20].

Justification of the recommendation

Total serum IgE should not be used for the diagnosis of asthma because of consistently poor sensitivities across the studies, reaching at best 51%. This is in line with the existence of a significant proportion of non IgE-mediated asthma, also called “intrinsic” asthma. Measuring total serum IgE does not provide immediate results at the time of the consultation. If specificity is better than sensitivity it remains limited at the cut-offs provided by the ROC curves, ranging from 39% to 85%. The value of measuring IgE may vary according to the population of patients investigated, the seasonal manifestations of the symptoms, the coexistence of allergic rhinitis and is likely to be more valid in young patients as IgE levels decline with age [68–70].

Patient perspective

Measuring total IgE is relatively cheap and minimally invasive, although some patients may be anxious of venipuncture. Patients are often keen to know their possible allergies and, although skin tests are the gold standard to define allergic status, measuring total and specific serum IgE may certainly represent a useful approach to assess allergy in primary care.

Recommendation

• The TF suggests not combining FeNO, blood eosinophils and serum IgE to make a diagnosis of asthma (conditional recommendation against the combination of tests, moderate quality of evidence)

Background

Total serum IgE, BEC and FeNO represent facets of the T2 asthma phenotype, although the molecular mechanisms behind these biochemical and cellular variables may be different and eosinophils and IgE dissociated [71, 72]. We therefore investigated whether the combination of these variables could improve their diagnostic value.

Review of the evidence

Our search identified 10 potentially relevant studies of which only one study was suitable to be included (supplementary tables 11a and b). Combination of the three tests provided an AUC-ROC of 0.6 (95 CI:0.56–0.64) while the AUC for individual tests were 0.58 (0.54–0.62), 0.57 (0.53–0.61) and 0.58 (0.54–0.62) for FeNO, IgE and BEC respectively [20]. Overall, sensitivity of the combination was 0.46 (95% CI: 0.37 to 0.52) while specificity was 0 .74 (95% CI: 0.64 to 0.69) (GRADE table 11, supplementary EtD table 11b)

Justification of the recommendation

Although a large study, the only study that met the criteria was a single-centre secondary care assessment. Combining blood eosinophils, total serum IgE and FeNO does not seem to improve diagnostic accuracy as compared to performing one single test. Further studies are needed, particularly those in primary care.

Recommendation

• The TF suggests bronchial challenge testing should be performed in secondary care to confirm a diagnosis of asthma in adults when the diagnosis was not previously established in primary care (conditional recommendation for the test, low quality of evidence)

Remarks

• A provocative concentration of methacholine (PC20-M) or histamine (PC20-H) <8 mg·mL⁻¹ in steroid-naïve patients and <16 mg·mL⁻¹ in patient receiving regular inhaled corticosteroids supports a diagnosis of asthma

• Indirect challenges such as mannitol or exercise may be considered in patients who remain negative with direct constricting agents

Background

Bronchial challenges demonstrate bronchial hyperresponsiveness (BHR), one of the key pathophysiological feature of asthma, and are divided into direct and indirect challenges on the basis of the mechanism leading to airway constriction [73–75]. Challenges with methacholine or histamine are considered direct tests as these mediators bind directly to airway smooth muscle leading to constriction. Exercise or mannitol challenge are considered as indirect airway challenges as they involve local release of constricting mediators such as cysteinyl-leukotrienes in the vicinity of smooth muscle. Indirect challenges are better correlated with the extent of airway inflammation than direct challenges [74, 76]. We investigated whether bronchial challenge could identify patients with asthma diagnosed by BdR and we compared the performance of both tests to confirm a diagnosis of asthma.

Review of the evidence

Our search identified 18 potentially relevant studies of which six studies were suitable for inclusion (supplementary tables 12a and b) (five prospective cross-sectional [19, 26, 29, 63, 77], one retrospective) [78]. Two studies assessed the value of bronchial challenge to identify patients diagnosed as being asthmatic based on both suggestive symptoms and positive BdR test (supplementary table 13). Porpodis et al., prospectively investigated 88 steroid-naive subjects where 67 patients were diagnosed as asthma based on suggestive symptoms and BdR of 12% and 200-mL FEV₁ improvement [77]. Louis et al., assessed 194 steroid-naive patients retrospectively with symptoms suggestive of asthma and baseline FEV₁ >70% predicted, and found 39 patients with a BdR of 12% and 200-mL FEV₁ improvement [78]. Other studies have compared the performance of BdR versus bronchial challenge in patients with symptoms suggestive of asthma (supplementary table 12). Overall, sensitivity ranged between 0.63 and 0.97 while specificity ranged between 0.12 and 1 (GRADE table 12, supplementary EtD table 12b).

Justification of the recommendation

In making a conditional recommendation the TF balanced the desirable effects of making a diagnosis, against any undesirable effects, risks to patients and the resources required to implement and make bronchial challenge testing a feasible test. Although methacholine, histamine and mannitol are very safe, these tests require additional equipment, reagents, time in the laboratory, air source, and trained staff, with access to resuscitation facilities and medical personnel in rare cases of severe bronchoconstriction. This will undoubtedly increase the costs in comparison to BdR testing. Mannitol challenge appeared slightly more specific than methacholine challenge, albeit one study.

Patient perspective

Patients may feel uncomfortable during bronchial challenge testing as using histamine may cause unpleasant facial flushing and headache, and mannitol can induce cough. In addition, prior to bronchial challenge tests, patients on inhaled and oral treatment including anti-histamines (for histamine challenge) will need to be withdrawn in order to reduce the risk of a false negative test. However, some patients, particular those who may been previously diagnosed as moderate or severe asthma, may find treatment withdrawal difficult or unacceptable. Therefore, the TF recommends careful discussion with patients about medication withdrawal for purpose testing.

Key unanswered questions

Several types of bronchial challenge have been validated to confirm the diagnosis of asthma when reversibility of airway obstruction cannot be demonstrated. Whether prognosis, natural evolution and response whilst on treatment are similar irrespective of the method that has been used to make the diagnosis is largely unknown. Prospective trials are needed to answer this important clinical question.

Recommendation

• The TF suggests not measuring sGaw and RV/TLC by whole body plethysmography to make to make a diagnosis of asthma (conditional recommendation against the tests, low quality of evidence)

Remarks

• sGaw does not perform better than FEV₁/FVC ratio to predict positive methacholine challenge in patients with normal baseline FEV₁

• RV/TLC >130% predicted has a high specificity (>90%) but poor sensitivity (25%) to predict a positive methacholine challenge in patient with normal FEV₁/FVC

Background

Temporal fluctuation in airway caliber is linked to variation in airways resistance. Specific airway conductance (sGaw) is a sensitive index to measure airway resistance related to lung volume and does not require the patient to perform a forced effort-dependent maneuver. Topalovic et al., observed 21% of asthma patients may display abnormally low specific airway conductance (<0.63 1/KPas.sec) despite FEV₁/FVC >LLN [79]. Emphasis has been placed on the role of distal airway narrowing and gas trapping in asthma that can be measured by the ratio RV/TLC [80, 81]. We undertook to investigate whether sGaw, a sensitive marker of airway obstruction, and the ratio of residual volume/total lung capacity (RV/TLC), an index of lung hyperinflation measured by whole body plethysmography, could help in the diagnosis of asthma when baseline spirometry appears to be normal.

Review of evidence

Our literature search identified 11 potentially relevant studies of which only two were suitable for inclusion (supplementary table 14a). Both were retrospective and performed in secondary care, where only one undertook a direct comparison between FEV₁/FVC, sGaw and RV/TLC (supplementary table 4) [18, 21]. Stanbrook et al., analysed the lung function results of 500 patients with asthma, chronic obstructive pulmonary disease (COPD), bronchitis and bronchiectasis, where 169 patients had no baseline airway obstruction, defined by FEV₁/FVC >90% of predicted [21]. The authors investigated the relationship between gas trapping, measured by the change in functional residual capacity (ΔFRC)_{body plethysmography}-(minus)-FRC_helium) and RV/TLC with a positive PC20-M <8 mg·mL⁻¹. No details were provided however on the symptom status of the patients, so it is difficult to ascertain if all patients with a positive PC20-M were actually patients with asthma. The authors investigated the diagnostic performance of predetermined values of ΔFRC_{body plethysmography}-(minus)-FRC_helium and RV/TLC. Bougard et al., assessed the lung function indices of sGaw and RV/TLC to predict a positive bronchial methacholine challenge (PC20-M <16 mg·mL⁻¹) by constructing ROC curves in 270 patients referred to a secondary care asthma clinic. All patients had whole body plethysmography prior to their visit at the asthma clinic for the methacholine challenge and were divided into a training cohort (n=129, baseline FEV₁ 95% predicted) and a validation cohort (n=141, baseline FEV₁ 91% predicted), indicating no substantial lung function impairment [18]. Among all plethysmography indices measured, RV/TLC provided the best AUC-ROC in both training and validation cohorts with values reaching 0.74 and 0.75, respectively while AUC-ROC reached 0.69 and 0.62 for sGaw in the training and the validation cohorts respectively. A model combining RV/TLC and FeNO provided an AUC that rose up to 0.79. Overall, sensitivity for sGaw ranged from 0.50 to 0.51 and specificity from 0.71 to 0.74 (GRADE table 13, supplementary EtD table 14b). Sensitivity for RV/TLC ranged from 0.28 to 0.71 while specificity was ranged from 0.68 to 0.86 (GRADE table 14, supplementary EtD table 14b). In patients with RV/TLC >135% predicted and an FEV₁/FVC >90%, provided 95% specificity in Stanbrook's study [21].

Justification of the recommendation

The current evidence with RV/TLC is too limited to recommend using it to ascertain a diagnosis of asthma. The two studies suggest a high RV/TLC might be a useful physiological index to consider asthma diagnosis. Whole body plethysmography can provide sophisticated lung function measurements including the early physiological sign of hyperdistention as a consequence of small airway obstruction, not revealed by spirometry. Where RV/TLC may hold some promise, measuring sGaw does not bring additional value to the measurement FEV₁/FVC ratio by spirometry. Whole body plethysmography, however, requires technical expertise from laboratory personnel and the cost and relatively limited access even in specialist secondary care may preclude use of this test on a large scale.

Key unanswered questions

Prospective studies are needed to further assess the value of RV/TLC, potentially combined with FeNO in patients with normal baseline spirometric indices.

How to investigate patients already receiving regular maintenance medication to make an asthma diagnosis?

In patients receiving ICS maintenance therapy as monotherapy or in combination with LABA, the demonstration of variable airway obstruction may be challenging. Where the influence of LABA disappears in a few days, long-term ICS use may reduce airway responsiveness and normalise airway calibre for longer [82, 83]. For patients established on maintenance therapy, GINA recommends making the diagnosis by the classic criteria of reversibility testing or bronchial challenge testing, being less stringent for the latter and accepting a PC20 <16 mg·mL⁻¹ as valid diagnostic criterion. In patients with a negative BdR, (FEV₁ does not improve by 12% and 200 mL) and a negative methacholine challenge (PC20-M <16 mg·mL⁻¹), ICS maintenance treatment is gradually tapered, and if symptoms do not worsen nor a significant decline in spirometry or PEF monitoring occurs, a bronchial challenge test can be repeated [3, 82].

Objective testing of airflow variability and airway hyper-responsiveness over 12 months is important to address seasonal and occupational asthma or intermittent increases in airway hyper-responsiveness from infections, and asthma is usually excluded if these are normal [84]. Patients should be encouraged to present to the physician if they experience any worsening of respiratory symptoms during this period, and alternative diagnoses should of course be considered and investigated.

How may comorbidities obscure the diagnosis of asthma?

Asthma frequently coexists with co-morbidities that not only affect the control and management of asthma [85], but need to be considered during the diagnostic phase. Some comorbidities can be supportive in diagnosing asthma. The presence of atopy and atopic conditions such as allergic rhinitis or atopic dermatitis increase the probability of the diagnosis of allergic asthma when patients present with respiratory symptoms [86]. The presence of atopy is not specific for asthma [87], nor does its absence rule out asthma, since atopy is not present in all asthma phenotypes. It should be noted that the relevance of allergen exposure in relation to symptoms requires a positive test (skin prick test or serum specific IgE) confirmed by a corresponding history.

Chronic rhinosinusitis and nasal polyposis are more often associated with the late-onset eosinophilic asthma subtype, characterised by onset of disease in adulthood, absence of atopy, airway obstruction without a smoking history and eosinophilic inflammation [88, 89]. In this respect, the presence of chronic rhinosinusitis or nasal polyposis in patients with respiratory symptoms usually alerts physicians to consider the diagnosis of asthma, with the late-onset phenotype.

COPD is the other most common chronic obstructive airway disease. The diagnosis of asthma and COPD may not be mutually exclusive given that many patients with asthma smoke [90] or are exposed to noxious gases and it is common to observe irreversible airway obstruction in moderate to severe asthmatics [91]. Gastro-oesophageal reflux disease (GERD) can cause laryngeal or pharyngeal irritation, chest tightness, and dry cough, symptoms that can easily be misinterpreted as asthma [3], and are often more problematic at night. The diagnosis of GERD may be considered, particularly in patients presenting with non-productive cough as their main symptom, and current consensus suggests an empirical treatment of anti-reflux medication may be used where there is objective evidence of reflux or a history suggestive of reflux symptoms [44].

A particular challenge is the diagnosis of asthma in people with obesity. Obesity itself can cause shortness of breath, wheezing due to breathing at lower volume and reduced exercise tolerance, and may be accompanied by GERD or obstructive sleep apnoea, which in turn can cause asthma-like symptoms. People with obesity are shown to be at risk of both over- and under-diagnosis of asthma [92], and need an objective diagnosis of asthma to prevent unwanted over- or under-treatment.

Inducible laryngeal obstruction (ILO), hyperventilation and dysfunctional breathing all may cause asthma-like symptoms and lead to an incorrect asthma diagnosis. Patients with inducible laryngeal obstruction have an inappropriate, transient, reversible narrowing of the larynx in response to diverse triggers [93], that may result in inspiratory breathing difficulties, sometimes with coarse to high-pitched inspiratory breath sounds, and repetitive attacks of acute dyspnea (mimicking exacerbations of asthma). Dysfunctional breathing is characterised by irregular breathing patterns and patients with this condition often present with dyspnea or "air hunger", together with non-respiratory symptoms such as dizziness and palpitations [94]. Valid, accessible and quantifiable tests for diagnosing dysfunction breathing is missing, making it difficult to distinguish from asthma, although continuous laryngoscopy during exercise (CLE) is considered a reliable test to diagnose or rule out exercise-induced laryngeal obstruction [95]. In these patients, symptoms do not improve on asthma medicines and it is preferable to consider alternative options, such as breathing exercises, speech therapy, biofeedback strategies or psychological support.

Does lung imaging help in the work up of asthma diagnosis?

Beyond the physiological abnormalities defining asthma, additional investigations may be worthwhile to demonstrate co-morbidities that may be contributing to the symptom burden of the patient. High-resolution computed tomogram (HRCT) of the lungs provides a diagnosis of additional conditions in 40% of cases in patients with severe asthma, including bronchiectasis, emphysema and lung nodules [96]. HRCT can identify classical radio-pathological patterns of airway wall thickening, airway distensibility, bronchiectasis, lung distension and air trapping, where most of these changes can overlap with each other and present in varying proportions. The radiological presence of emphysema (or “pseudo-emphysema”) increases the complexity of differentiating asthma from COPD, and air trapping can be challenging to discriminate from emphysema. Assessing HRCT lung changes before and after treatment (bronchodilation, anti-inflammatory treatment) or airway challenge (bronchoconstriction) are potentially insightful [97–100]. However, it appears that as an increasing number of radiological features are incidentally detected (e.g. interstitial lung abnormalities), which may make the diagnosis of asthma a challenge. Beyond an alternative diagnosis, additional studies are needed to assess whether HRCT is able to identify particular phenotypes and predict treatment response [98, 99]. and potentially whether radiological features can predict future risk of disease exacerbation and lung function decline. Noteworthy, sinus CT can not only identify asthma-related comorbidities such as nasal polyposis, but also has the potential to support phenotypic characterisation.

Do we need to phenotype airway and systemic inflammation in the patient with asthma?

Asthma is a heterogeneous disease that encompasses different clinical phenotypes and endotypes that share excessive airflow fluctuation [101, 102]. In particular, there is now clear evidence of differing patterns of airways inflammation in people with asthma. Although not applicable in primary care setting the development of the technique of induced sputum has been pivotal to airway inflammatory phenotyping in asthma [103–105]. When available in secondary care, induced sputum may complement the diagnostic work-up in severe patients [3]. Some authors have advocated to classify the patients based on the granulocytic airway content [106–108]. In large cohorts of patients across the whole severity spectrum pauci-granunocytic and eosinophilic asthma were found to be the two most frequently encountered phenotypes where the proportion of eosinophilic asthma increases with disease severity [106, 107, 109]. In contrast, paucigranulocytic asthma is the most prevalent inflammatory phenotype in mild asthma [78, 106, 110], even if sputum analysis suggests that paucigranulocytic asthma are actually low-grade eosinophilic airway inflammation [111]. Although sputum eosinophils were shown to provide acceptable accuracy to diagnose asthma [19], the main interest of identifying airway cell content is that it may provide valuable information regarding several clinical asthma outcomes beyond the diagnosis [112]. Sputum eosinophilia predicts a good response to ICS or to a course of OCS [103]. The persistently mixed granulocytic profile is associated with lung function decline and relative resistance to ICS in contrast to the pure highly variable eosinophilic pattern, which shows propensity to exacerbation but generally a good response to corticoids preventing decline in lung function [113]. Biomarkers such as blood eosinophils and FeNO have shown consistent relationship with sputum eosinophil counts and were found to be good predictors of the response to ICS in steroid-naïve patients [51, 114–116], making them suitable tools to phenotype asthma in primary care setting. We currently lack of user-friendly biomarkers to identify neutrophilic asthma, a phenotype found to be associated with signs of innate immunity activation [117, 118], often induced by dysbiosis [119, 120] and resistant to ICS [121]. Analysis of VOCs has recently shown some promise in this respect [122].

Categorisation of asthma according to the inflammatory profile has proved to be invaluable in the appropriate targeting of expensive biological treatments in difficult asthma, where use of T2 biomarkers differentiates those likely to respond from those unlikely to benefit [123]. Furthermore, the growing recognition of the need for personalised [124], precision medicine, based on categorisation and appropriate response to the variety of drivers of disease at an individual level, has led to the proposal for a “treatable traits” strategy in airways disease [125]. There is preliminary evidence that this is a successful strategy in hospital-based care [126], with calls from the ERS for more research into wider clinical implementation of this approach [127].

What are the patient perspectives of asthma diagnosis in adults?

A review of published and grey literature explored patient experiences of adult asthma diagnosis. Details of the search strategy available in the supplement.

Patients are often uncertain about starting treatment without first having a definitive diagnosis [6]. In the absence of a diagnosis, some patients may want to trial treatment to check if they experience any benefit (table 5). Patients describe the surprise of being diagnosed later in life as an adult. They often considered asthma to be a childhood illness, and thought it was possible to “grow out of” asthma. Patients express frustration at not knowing why they develop asthma at this point in life (table 5).

Patients describe the psycho-social impact of diagnosis where for some, getting a diagnosis can be positive, finally pinpointing the underlying cause of their poor health and providing tools to manage it. Depression, feeling scared and having anxiety about how asthma will affect other aspects of their life are common. Patients have complex emotions about how their condition impacts their loved ones, and how their relationships have changed as a result. Overall, patients describe coming to terms with the diagnosis, accepting it as something they have to live with long term, recognising that asthma can be life-threating, and their role in self-management. Professionals have an important role in supporting their patients with the psycho-social impact (table 5). If a diagnostic test is done in hospital, results need to be communicated to the family doctor and ideally followed up in community care [128].

Patients would benefit from further research on the actual diagnostic pathways of asthma patients. Professionals have an important role in improving the patient experience of diagnostic testing and supporting individuals to manage the wider impact of diagnosis. The diagnostic process can be long and confusing for adult patients who would benefit from clear patient-centred information which takes into account variation in access to diagnostic testing across Europe.