The ARIA classification was clearly a great step forward in 2001, since it acknowledged the impact of a disease that was often qualified as “trivial" and emphasized the need to assess patient needs and treat accordingly. The ARIA "mild" vs. "moderate/severe" classification 13 14 15 has a number of strengths and weaknesses (Table 1). It has the advantage of being very simple to administer, since it is based on "yes"/"no" answers to each of the following statements: "My symptoms disturb my sleep", "My symptoms restrict my daily activities (sports, leisure, etc.)", "My symptoms restrict my participation in school or work" and "My symptoms are troublesome". In ARIA, "mild" AR corresponds to "no" answers to all four questions, whereas the presence of a "yes" for one or more items corresponds to "moderate/severe" AR. The ARIA duration and severity classifications have been implemented in several countries and patient populations. For example, the ADRIAL and PEDRIAL cohort studies of adult and paediatric AR patients in Spain 20 21 found that symptom, Rhinitis Quality of Life Questionnaire (RQLQ) and visual analogue scale (VAS) scores were significantly higher in "moderate/severe" than in "mild" AR. In France, Bousquet et al. studied 3052 patients consulting PCPs and who were classified according to the ARIA classification (mild intermittent: 11%; mild persistent: 8%; moderate/severe intermittent: 35%; moderate/severe persistent: 46%). All patients were scored for the RQLQ, the Jenkins sleep questionnaire and the Allergy-Specific Work Productivity and Activity Impairment questionnaire 22 . Impairments were correlated more strongly with severity than duration. Eighty per cent of the patients with moderate-to-severe AR reported impaired activities, as opposed to only 40% of the patients classified as having mild AR.

Levels of awareness and application of the ARIA severity classification are less than satisfactory. In a study of 943 French PCPs and 277 ear, nose and throat (ENT) specialists, Demoly et al. found that only about 54% of the physicians were aware of the ARIA classification. Over 90% of these actually applied the classification in practice - although there were significant differences between PCPs and ENT specialists 17 . Furthermore, knowledge of the ARIA classification by PCPs did not appear to influence the use of H1-antihistamines (H1As) and/or intranasal corticosteroids (ICSs) as a function of the patient's disease severity 17 . In another cohort study performed in France 23 , researchers found that ARIA severity did not significantly influence medication prescription. Although ARIA suggests that patients with mild and intermittent AR should receive H1As and those with moderate/severe and persistent AR should receive ICSs, Ramirez et al.'s analysis of 3026 patients with intermittent AR and 3507 patients with persistent AR (enrolled by 1346 practitioners) revealed that immediate prescription of H1A+ICS combination therapy was surprisingly frequent 23 .

For AR patients seen by specialists, the "mild" vs. "moderate/severe" distinction has less value since by definition, a patient with "mild" AR is not bothered by his/her symptoms and is unlikely to consult a specialist. In Valovirta et al.'s survey of patients aged 16 and over 9 , 87% of those with persistent conditions and 79% of those with intermittent conditions reported that at least one daily activity was moderately or severely affected “all the time”. Similarly, Van Hoecke et al. observed that 89.3% of patients consulting PCPs in Belgium were classified as "moderate-severe" 24 , as was the case for 92.2% of the patients in the above-mentioned French cohort 17 25 .

However, “moderate-severe” AR is extremely broad (with as few as one or all four ARIA items impaired) and encompasses a highly heterogeneous group of patients. It includes patients with severe chronic upper airway disease (SCUAD) 26 , candidates for allergen specific immunotherapy or nasal obstruction surgery and patients with severe disease in the absence of treatment but who respond well to ICSs or even H1As alone. As such, classification of a patient as suffering from “moderate-severe” AR is of little help in guiding the physician's treatment recommendations. In this respect, it has even been suggested that term "moderate/severe" should be replaced by "severe" alone 22 . Lastly, the ARIA classification does not take account of past and present treatment; this is an important shortcoming, since most AR patients seen for the first time by a physician will have already taken prescription and/or over-the-counter medications for their condition. In summary, consideration of the ARIA classification's particular features has highlighted unmet needs in the routine clinical management of AR.

Since the publication of ARIA's initial work in 2001, several attempts have been made to refine the "mild" vs. "moderate/severe" classification. In 2007, Valero et al. suggested drawing a distinction between 1 to 3 affected ARIA items on one hand and 4 affected items on the other, with the assignment of "moderate" and "severe" disease grades to these respective situations 27 . In a study of over a thousand treatment-naïve patients, classification as "mild", "moderate" or "severe" was correlated with the disease-specific "Cuestionario ESPañol de Calidad de Vida en RINiTis" (ESPRINT-15) quality of life score 28 .

In a study in Belgium, Van Hoecke et al. suggested the use of just two ARIA questions (one question on sleep disturbance and one on impairment in daily life) and the introduction of an additional, "moderate" category 24 . Two "no" replies equated to "mild" AR, one "yes" and one "no" equated to "moderate" AR and two "yes" replies equated to "severe" AR 24 . However, in a cohort of 5140 patients in France classified as having mild (n=357, 7.0%), moderate (n=2498, 48.6%) or severe AR (n=2285, 44.4%) according to Van Hoecke et al.'s suggestion, Demoly et al. found that "no clear clinically relevant trends were observed that could support the need for a distinction between mild, moderate and severe patients" 29 . In response to these proposals to classify AR severity as “mild”, “moderate” and “severe”, the ARIA 2008 authors stated that this "makes it more complex for the practising physician without bringing significant improvement to the patient since this more complex classification does not translate to a difference in therapeutic options" 14 .

Visual analogue scales have been suggested as simple tools for assessing AR severity. Firstly, Bousquet et al. 30 found that a 0 to 10 cm VAS score and the RQLQ score were significantly correlated (rho = 0.46; p <0.0001) and that a patient with a VAS score below 5 cm could be classified as having "mild" AR (negative predictive value: 93.5%), whereas a VAS score over 6 cm was equated with "moderate/severe" AR (positive predictive value: 73.6%). Indeed, the ARIA 2008 guidelines subsequently included a classification in which "mild" AR = 0–3 cm, "moderate" AR = 3.1–7 cm and "severe" AR = 7.1–10 cm. Secondly, the United States' Joint Task Force on Practice Parameters 31 suggested that six VASs (sneezing, runny nose, congestion, itchy nose, postnasal drip and total nasal symptoms) could be used. However, the Task Force's 2003 publication did not present any results on application of these VASs and the validation status of this instrument is uncertain. Thirdly, two of the present authors (PD and IAM) and colleagues compared a VAS severity score with a numerical severity score (both assessed by a physician) in a sample of 36,000 patients with diagnosed, non-complicated, untreated, intermittent AR 32 . Although the two scores were correlated, the absolute values differed and 23.86% of the patients were classified as "severe" according to one scale but not the other.

Disease control is now being considered as an alternative to disease severity in the management of patients with respiratory disease. Indeed, the assessment of disease control in AR was briefly mentioned in the ARIA 2008 update 14 , since "as for asthma, one of the problems to consider is to replace severity by control". However, the ARIA authors noted that, at the time the update was drafted, "sufficient data are not yet available" and that "control questionnaires or methods are still undergoing validation" 14 . Since then, however, a number of AR control questionnaires have been built and validated according to the ideal methodological sequence 33 . These questionnaires are summarized in Table 2.

The Control of Allergic Rhinitis and Asthma Test (CARAT) was initially developed by Nogueira-Silva et al. in Portuguese 34 . After a literature search, Nogueira-Silva et al. included 17 questions in a questionnaire with a Likert scale. The reference period was four weeks – long enough to monitor control over time but short enough to be unaffected by recall bias. A ten-question version of CARAT (CARAT10) was subsequently validated in a cross-sectional study of 193 adults by Fonseca et al. 35 . The range of possible scores for CARAT10 is 0–30, 0 being the complete absence of control. The CARAT10 served as a guide to patient management, since the mean [95% confidence interval] scores associated with the intensification, maintenance and reduction of treatment were 15 [13.6-16.5], 21 [19.4-21.9] and 24 [21.4-26.6], respectively.]. Most recently, Fonseca et al. investigated the CARAT in 62 patients included at 4 outpatient clinics in Portugal 36 . At two visits 4 to 6 weeks apart, a total of 51 patients (aged between 18 and 70) completed the CARAT10, the Asthma Control Questionnaire 37 , three VASs (on airways symptoms, bronchial/pulmonary symptoms and nasal symptoms) and an overall self-assessment of control. Lung function was also tested. The test-retest reliability (intra-class correlation coefficient) was 0.82. A significant change in the CARAT10 score was observed in clinically unstable patients. In terms of the change over time in the CARAT10 score, the correlation ranged from 0.49–0.65 for the ACQ5 and VAS symptom scores and from 0.31 to 0.41 for the physician's assessment of control 36 .

Nathan et al. developed a 26-item Rhinitis Control Assessment Test (RCAT) and refined it to 6 items. After testing the RCAT in 410 AR patients 38 39 , six of the 26 initial items (nasal congestion, sneezing, and watery eyes, sleep interference, activity avoidance and self-assessed control) were most predictive (p <0.001 for all) of the allergist’s overall rating of rhinitis symptom control.

By working with a multidisciplinary group associating allergists, pulmonologists, ENT physicians and methodologists, Demoly et al. developed a five-item, self-assessment Allergic Rhinitis Control Test (ARCT) 40 with similarities to the Asthma Control Test (ACT) 41 . Two ACT questions (on rescue medication and overall assessment of the disease) were incorporated into the ARCT. The questionnaire was validated by testing in 902 patients (selected by 411 PCPs and allergists) before treatment and two weeks after treatment. The score at inclusion correlated significantly (p<0.0001) with the patient's overall clinical status and the impact of AR on social and sporting activities. A significant (p<0.0001) increase in the score was observed after two weeks of treatment (from 14.9±4.0 at inclusion to 21.5±2.9 after treatment). Using a receiver operating characteristic curve, a score of 20 was found to be the optimal cut-off for poor vs. well-controlled rhinitis (sensitivity: 67%; specificity: 82%; negative predictive value: 32%; positive predictive value: 95%).

The "Allergy-Control-SCORE"^TM 42 measures (i) the severity of 10 nasal and non-nasal symptoms on a 4-point scale) and (ii) medication use (out of a catalogue of 745 different medications) and combines the two metrics with equal weighting. In a study of 81 patients with allergic rhinoconjunctivitis and 40 healthy controls, the Allergy-Control-SCORE was significantly correlated with the global assessment of allergy severity, the RQLQ score and the number of medical consultations due to allergy within the previous year. In our opinion, "Allergy-Control-SCORE" is something of a misnomer because this approach is conventionally referred to as a "combined score" (i.e. a combination of a symptom score and a medication score) and has been extensively used in clinical trials for many years 43 .

Lastly, Scadding et al. took a different approach by asking experts to complete an online survey about what they regarded as rhinitis control in relation to the seven questions in Juniper's mini-RQLQ 44 . All the respondents considered that more than “somewhat troubled” was unacceptable. Indeed, most of the experts defined control as being “hardly troubled at all” by each symptom.

Unfortunately, head-to-head comparisons of these various control-based tools have not been reported.

While the ARIA classification of the severity of AR is useful, it is not an optimal guide for making everyday patient management decisions, especially in patients already on therapy. Experience in asthma suggests that there are good reasons to consider measuring control on a routine basis in AR, as a complement to ARIA's severity-based approach. There is a need to compare existing tools and perhaps develop new ones. Regardless of the details of control-based classifications in AR, the key challenge for any instrument will be to achieve high levels of physician awareness, uptake and application - which should ultimately lead to better patient outcomes.

ACS: Allergy-Control-SCORE; AR: Allergic rhinitis; ARC: Allergic Rhinitis Control; ARCT: Allergic Rhinitis Control Test; ARIA: Allergic Rhinitis and its Impact on Asthma; CARAT: Control of Allergic Rhinitis and Asthma Test; ENT: Ear, nose and throat; GINA: Global Initiative for Asthma; H1A: H1-antihistamine; ICS: Intranasal corticosteroid; PCP: Primary care physician; RCAT: Rhinitis Control Assessment Test; RQLQ: Rhinitis Quality of Life Questionnaire; VAS: Visual analogue scale.

Pascal Demoly is a consultant and a speaker for Stallergenes, ALK and Chiesi and was a speaker for Merck, Astra Zeneca, Menarini and GlaxoSmithKline in 2010–2012. Moises Calderon has received consulting fees, honoraria for lectures and/or research funding from ALK-Abello, Merck, Stallergenes and Allergopharma. Glenis Scadding has received consulting fees, honoraria for lectures and/or research funding from ALK-Abello, GSK, Meda, Merck and Stallergenes. Thomas Casale has received consulting fees from Stallergenes and has been an investigator on grants awarded to Creighton University by Stallergenes and Schering-Plough-MSD. This work was supported by an educational grant from Stallergenes.

PD, MC, TC, GS, IAM, JJB, BD, TH, OM, FT and ES all made substantial contributions to the (i) the conception and design of the disease control classification and (ii) the identification and review of relevant literature on ARIA and published proposals for modification. Likewise, PD, MC, TC, GS, IAM, JJB, BD, TH, OM, FT and ES were all involved in drafting the manuscript and/or revising it critically for important intellectual content. All the authors have given final approval of the version to be published.