PROF.

PHILIPPE DEVILLIER (Orcid ID : 0000-0003-4107-8317)

Article type : Original Article: Airway Diseases

Accepted Article
Title: Immunotherapy with grass pollen tablets reduces medication dispensing for allergic

rhinitis and asthma: a retrospective database study in France

Short title: Grass pollen tablets and allergy progression

P. Devillier1, M. Molimard2, X. Ansolabehere3, I. Bardoulat3, N. Coulombel3, F. Maurel3, P.

Le Jeunne3, P. Demoly4.

1
UPRES EA 220, Department of Airway Diseases, Hospital Foch, University of Versailles

Saint Quentin, University Paris-Saclay, Suresnes, France.

2
Pharmacology Department, University of Bordeaux, INSERM unit CR1219, Bordeaux,

France.
3
Real-World Insights Department, IQVIA, La Défense, France.
4
Department of Pulmonology and Addictology, Arnaud de Villeneuve Hospital, Montpellier

University, Montpellier, and Sorbonne Universités, UPMC Paris 06, UMR-S 1136 INSERM,

IPLESP, Equipe EPAR, 75013, Paris, France

Author to whom correspondence and proofs should be sent and from whom offprints should

be requested:

This article has been accepted for publication and undergone full peer review but has not
been through the copyediting, typesetting, pagination and proofreading process, which may
lead to differences between this version and the Version of Record. Please cite this article as
doi: 10.1111/all.13705
This article is protected by copyright. All rights reserved.
 Philippe Devillier, UPRES EA 220, Université de Versailles Saint-Quentin, Foch Hospital,

Suresnes, France.
Accepted Article
Tel.: +33-146-252-791 /

Fax: +33-140-999-657

E-mail: p.devillier@hopital-foch.org

Abstract

Background: Moderate-to-severe allergic rhinitis (AR) may increase the risk of developing or

worsening asthma, whereas treatment of AR with subcutaneously or sublingual allergen

immunotherapy (SLIT) may slow this progression.

Methods: In a retrospective real-world analysis, prescription fulfilment data were gathered

from French retail pharmacies between March 1st, 2012, and December 31st, 2016. Using

linear regression analyses, patients having received at least two prescriptions of grass pollen

SLIT tablets over at least two successive years were compared with control patients having

received symptomatic medications only.

Results: 1,099 SLIT patients and 27,475 control patients were included in the main analysis.

With regard to symptomatic AR medication dispensing, we observed a 50% decrease in the

pre-index/follow-up ratio in the SLIT group, a 30% increase in the control group without age

matching (p<0.0001 vs. SLIT), and a 20% increase in the control group with age matching

(p<0.0001 vs. SLIT). During the follow-up, 11 (1.8%) and 782 (5.3%) patients initiated

asthma treatment in the SLIT and control groups, respectively. The relative risk of

medication dispensing for new asthma was lower in the SLIT group (by 62.5%

[29.1%;80.1%] without age matching (p=0.0025), and by 63.7% [31.5%;80.7%] with age

This article is protected by copyright. All rights reserved.

 matching (p=0.0018)). SLIT was also associated with slower progression of asthma

medication dispensing during the follow-up period, relative to the control group (regression
Accepted Article
coefficient: -0.58[-0.74;-0.42] without age matching (p<0.0001) and -0.61 [-0.76;-0.46] with

age matching (p<0.0001)).

Conclusion: Prescription of grass pollen SLIT tablets reduced the dispensing of AR and

asthma medications in real life.

KEYWORDS: allergic, asthma, database analysis, grass pollen tablet, progression

INTRODUCTION

Allergic rhinitis (AR) is one of the most common chronic diseases (1-3); it affects about 500

million people worldwide (4), including 100 million people in Europe (3). In a two-step,

cross-sectional, population-based survey with a clinical assessment (1), the estimated

prevalence [95% confidence interval (CI)] of AR in France was 24.5% [21.0–28.0]. Another

cross-sectional survey of a French population sample (based on questionnaires alone)

reported an overall prevalence of AR of 31% (5). Grass pollen is the main disease-inducing

allergen in seasonal AR; in the above-mentioned survey, specific IgEs against grass pollen

were detected in 52% of the subjects with a clinical diagnosis of AR.

In addition to the disease burden associated with AR per se (4, 6), a body of evidence further

suggests that moderate-to-severe AR (7)) increases the risk of developing or worsening

allergic asthma as part of the “allergic march” from atopic dermatitis to AR and then asthma

(8, 9). Indeed, the prevalence of asthma is higher among individuals with AR than among

This article is protected by copyright. All rights reserved.

 AR-free individuals (10, 11). In France, asthma is the allergic condition most frequently

associated with AR (5).

Accepted Article
A large number of double-blind, placebo-controlled, randomized clinical trials (DBPC

RCTs), meta-analyses of trials, systematic reviews, and guidelines from learned societies and

medical associations have emphasized that allergen immunotherapy (AIT) significantly

decreases the occurrence of severe AR symptoms, reduces the use of symptomatic

medications, and improves disease control in AR sufferers with allergic asthma (12-22). In

contrast, symptomatic medications (e.g. antihistamines and corticoids) provide only short-

term relief. It has notably been shown that grass pollen SLIT tablets have a greater mean

relative clinical impact than second-generation antihistamines and montelukast, and much the

same mean relative clinical impact as nasal corticosteroids (23).

Furthermore, it has also been suggested that AIT may slow the “allergic march” in AR

sufferers (24). Data from the five-year grass sublingual immunotherapy tablet asthma

prevention (GAP) trial in children with grass pollen allergy indicated that a grass pollen SLIT

tablet reduced the risk of experiencing asthma symptoms and using asthma medication but

did not show an effect on the time to onset of asthma (24). In a recently published guideline,

the European Academy of Allergy and Clinical Immunology (EAACI) recommended

“subcutaneous or sublingual AIT for children and adolescents with moderate-to-severe

allergic rhinitis (AR) triggered by grass/birch pollen allergy to prevent asthma for up to 2

years post-AIT in addition to its sustained effect on AR symptoms and medication” (25).

This article is protected by copyright. All rights reserved.

 The results of “real-world” studies have also demonstrated AIT’s ability to slow the allergic

march. Although DBPC RCTs and meta-analyses of DBPC RCTs constitute the “gold
Accepted Article
standard” for evaluating clinical safety and efficacy, real-world (post-marketing) evidence

can (i) produce evidence of clinical effectiveness, (ii) compensate for a paucity of clinical

trial evidence on the efficacy of AIT beyond 2 years post-treatment, and (iii) substantiate

long-term preventive effects of AIT on asthma progression and new asthma onset with a

broader patient base than that found in clinical trials (26-28). Treatment decisions can be

guided by “real-world” data that take into account the relationships between patient

characteristics, lifestyle factors, treatment compliance and regimens, comorbidities, quality of

life, and clinical outcomes. Recently, a retrospective, real-world analysis of a German

longitudinal prescription database highlighted the long-term clinical benefits of sublingual

immunotherapy (SLIT) grass pollen tablets for (i) slowing the progression of AR, and (ii)

slowing the onset and progression of asthma in patients with grass-pollen-induced AR for up

to 6 years after treatment cessation (29). Zielen et al. found that AR medication use was

lower in patients treated with grass pollen SLIT tablets than in control patients (i.e. treated

with symptomatic medications but not SLIT). Furthermore, the risk of asthma onset in

previously asthma-free individuals and the use of asthma medication after treatment cessation

were lower in SLIT-treated patients than in control patients (29). With a view to (i)

establishing whether these findings could be generalized to countries with different

prescription profiles, and (ii) increasing the amount of real-world data on the progression of

AR and/or asthma in patients receiving AIT, we conducted a prescription-based study of

retail pharmacy data in France. The French study was similar to the one performed by Zielen

et al. in Germany but included a broader panel of reimbursed drugs.

This article is protected by copyright. All rights reserved.

 METHODS

Source data
Accepted Article
We performed a real-world, retrospective, longitudinal analysis of prescription records from

the IMS Lifelink™ Treatment Dynamics (LTD) database (QuintilesIMS, La Défense, France)

created in March 2012 (29). This database is updated monthly with data from a panel of

French retail pharmacies (accounting for 34% of the nationwide total). These data include the

patient’s year of birth and gender, the prescriber’s medical specialty, and data on medications

prescribed and reimbursed - including the exact dispensing date, the medication’s name and

dosage, and the number of packs (Table S1). A fully anonymised patient ID enables

individual patients to be monitored over time. Over-the-counter drugs are not recorded in the

database, as they are not reimbursed by the French national health insurance system.

The study population

We performed a real-world, retrospective study of patients with AR over the age of 5 years

(corresponding to the minimum age for the prescription of grass pollen SLIT tablets). Given

the absence of diagnostic information in the IMS LTD database, AR was defined by

reference to a non-clinical proxy marker: a prescription dispensed during the pre-index period

for any of three classes of symptomatic medication (defined according to the Anatomical

Therapeutic Chemical (ATC)/European Pharmaceutical Market Research Association

(EphMRA) classification) typically used to treat AR: nasal corticosteroids, oral or systemic

antihistamine drugs, and nasal preparations (excluding anti-allergic drugs and ICSs). The

study population comprised a SLIT group and a control group. The SLIT group corresponded

to AR patients having received at least two prescriptions of grass pollen SLIT tablets

(Oralair® from Stallergenes Greer, Antony, France, or Grazax® from AL , rsholm,

Denmark) per course of treatment over at least two successive years. Members of the SLIT

This article is protected by copyright. All rights reserved.

 group could also receive prescriptions of symptomatic medication. The control group

comprised AR patients with moderate-to-severe disease (defined as at least two dispensed

Accepted Article
prescriptions of nasal corticosteroids during two successive pollen seasons) not having

received SLIT.

Study dates and periods

Our comparative analysis required the definition of several key dates and periods. Firstly, the

pollen season was defined as the period from April 1st to July 31st (30). Secondly, we defined

an index date. For the SLIT group, the index date was defined as the date of the first

prescription of grass pollen SLIT tablets during the pre-treatment or treatment period for the

2013 or 2014 grass pollen season (between December 2012 and July 2013, or between

December 2013 and July 2014). For the control group, the definition of the index date was

more complex (see the Supplementary Material – Methods) but was designed for

comparability with the SLIT group.

Next, the pre-index period was defined as the 365-day period before each patient’s index

date. The treatment period extended from the index date to the expiry date of the last

dispensed prescription of SLIT (for the SLIT group) or the last dispensed prescription of

symptomatic AR treatment in the following pollen season (for the control group). Lastly, the

follow-up period was defined as the period from the expiry date of the last prescription of

SLIT or symptomatic AR treatment to the end of the study (database lock: December 31st,

2016).

This article is protected by copyright. All rights reserved.

 Inclusion and exclusion criteria

The inclusion criteria were as follows: a valid index date; age over 5 years at the index date;
Accepted Article
moderate-to-severe AR in three successive pollen seasons (including the season before the

index date); and (for patients in the SLIT group) at least one year of follow-up after the last

dispensed prescription of SLIT tablets. Patients with severe asthma during the pre-index

period were excluded; severe asthma was defined as (i) at least a prescription of omalizumab

(Xolair®) or (ii) at least two prescriptions of oral corticosteroids recorded outside the pollen

season in patients receiving a combination of an ICS and a long-acting 2-agonist.

Study endpoints

The primary endpoint was the change over time in the number of prescriptions for

symptomatic AR medications (Table S1) during the follow-up period. The secondary

endpoints included the time to new asthma onset, i.e. cases of incident asthma during the

treatment period and during the follow-up period in patients who had not been treated for

asthma during the pre-index period. Incident asthma was defined as at least two dispensed

prescriptions of asthma medication (Table S1) in the same year or in two successive calendar

years, and was analysed as a binary variable (yes/no).

Another secondary endpoint was the change over time in the number of dispensed

prescriptions for asthma medication during the treatment period and during the follow-up

period in patients with AR and asthma at the index date. The outcome variables for AR and

asthma progression were adjusted for differences in treatment intensity during the pre-index

period and differences in the length of the study periods (relative to the one-year pre-index

period; see the Supplementary Material – Methods).

This article is protected by copyright. All rights reserved.

 Statistical analyses and matching procedures

In the main analysis, control patients were matched with patients in the SLIT group by index
Accepted Article
year, in order to minimize the confounding bias due to differences in the length and intensity

of the grass pollen seasons from one year to another. The SLIT control matching ratio was

1:25: each SLIT patient was matched with 25 distinct control patients, and each control

patient was matched with only one SLIT patient. In a secondary analysis, age (in three

classes: 5-17 years, 18-45 years, >45 years) was also included in the matching procedure, so

as to minimize confounding bias due to age differences between the SLIT tablet and control

groups (matching ratio: 1:10). Linear regression was used to analyse AR and asthma

progression, and logistic regression was used to analyse asthma onset. In multivariate

analyses of AR progression, asthma progression and asthma onset, covariates were included

(see Supplementary Material – Methods).

Ethics

Data collection and processing were approved by the French national data protection

commission (Commission Nationale de l'Informatique et des Libertés) on October 21st, 2011

(reference: DE-2011-097). In line with the French legislation on anonymised database

analyses, the patients’ informed consent was neither required nor sought.

RESULTS

Study population

Firstly, 44,963 patients receiving grass pollen SLIT tablets (the SLIT group) and 193,628

patients not receiving grass pollen SLIT tablets (the initial control group) were identified in

the IMS LTD database (Figure S1) during the period from March 1st, 2012, to December 31st,

2016. Secondly, 12,185 (27.1%) of the SLIT patients and 115,980 (59.9%) of the control

This article is protected by copyright. All rights reserved.

 patients had a valid index date, and respectively 1,099 (2.4%) and 103,404 (53.4%) patients

met all the eligibility criteria. After matching for the index year, 1,099 (2.4%) SLIT patients
Accepted Article
and 27,475 (14.2%) control patients were included in the main analysis. After additional

matching for age, 1,099 (2.4%) patients in the SLIT group and 10,990 (5.7%) patients in the

control group were included in the secondary analysis.

The baseline characteristics of the SLIT and control patients included in the main analysis are

summarized in Table S2. The patients in the SLIT group were younger than those in the

control group; for example, 90.5% of the SLIT patients vs. 29.5% of the control patients were

aged under 45 – thus justifying the age matching in the secondary analysis. On the basis of

the collected data, there was male predominance in both groups: 45.8%/36.8% males/females

in the SLIT group and 48.6%/46.3% in the control group. However, the percentage of

patients with missing data for sex was relatively high: 17.5% in the SLIT group and 5.1% in

the control group. This difference was no longer present after age matching (Table S2).

During the pre-index period, 37.6% of patients in the SLIT group and 39.2% of patients in

the control group had been treated for asthma. The symptomatic AR treatments were mainly

prescribed by a general practitioner (40.7%) or a private-practice pulmonologist (39.5%) in

the SLIT group and by a general practitioner (83.2%) in the control group.

Progression of AR medication intake after treatment cessation

During the pre-index period (with n=1,099 patients in the SLIT group and n=27,475 patients

in the control group), the mean number of prescriptions for symptomatic AR medication per

patient and per year was lower in the SLIT group (mean ± SD: 6.5 ± 5.6; median: 5) than in

the control group (mean ± SD: 11.5 ± 9.0 in the main analysis without age matching, and 9.6

± 7.7 in the secondary analysis with age matching; median: 9 and 7, respectively) (Table 1).

This article is protected by copyright. All rights reserved.

 Importantly, 37% of the patients in the SLIT group did not receive any prescriptions of

symptomatic AR medication during the follow-up period. The corresponding proportion in

Accepted Article
the control group without age matching was much lower (4.5%) (Figure 1a). For the follow-

up period (with n=688 patients in the SLIT group and n=26,241 patients in the control

group), the mean ± SD number of prescriptions for symptomatic AR medication per patient

and per year (i.e. the group average) was 2.6 ± 4.2 in the SLIT group (median: 0.8) but 11.0 ±

9.6 and 8.6 ± 8.1 in the control group with and without age matching, respectively (median:

8.5 and 6.2, respectively). When considering only patients receiving prescriptions in both the

pre-index period and the follow-up period, we observed a 50% decrease in the pre-

index/follow-up ratio in the SLIT group, a 20% increase in the control group with age

matching, and a 30% increase in the control group without age matching (Table 1, Figure 1b,

and Table S3).

In the main analysis and after adjustment for covariates (gender, age class [<18 or ≥18], the

main prescriber’s specialty, and asthma status at the index date), a linear regression indicated

that the reduction in symptomatic AR prescriptions was significantly greater in SLIT patients

than in the control group (regression coefficient [95% CI]: -0.75 [-0.86;-0.64]; p<0.0001). In

secondary analyses (after age matching), the reduction in the number of symptomatic AR

prescriptions was again significantly greater in SLIT patients than in the control group

(regression coefficient [95% CI]: -0.71 [-0.81;-0.60]; p<0.0001).

New asthma medication onset

When considering the unadjusted data, we found that 83 (12.1%) of the patients in the SLIT

group and 2,054 (12.3%) of the patients in the control group had received asthma medications

during the treatment period. During the follow-up period, these numbers were respectively 11

This article is protected by copyright. All rights reserved.

 (1.8%) and 782 (5.3%). Overall (i.e. considering the treatment and follow-up periods

together), 94 patients (13.7%) in the SLIT group and 2,836 patients (17.0%) in the control
Accepted Article
group had been treated for asthma.

After adjustment for covariates (gender, age class [<18 or ≥18], main prescriber specialty, the

number of symptomatic AR prescriptions during the pre-index period, and length of the

analytical time period), a logistic regression analysis indicated that the risk of new asthma

onset was significantly lower in the SLIT group vs. control group (with or without age-

matching, and for the treatment period and the follow-up period; Figure 2 and Table 2).

When compared with the control group, the risk [95%CI] of new asthma onset in the SLIT

group during the follow-up period was 62.5% [29.1%;80.1%] lower before age matching

(p=0.0025) and 63.7% [31.5%;80.7%] lower after age matching (p=0.0018) (Table 3 and

Figure S2).

Progression of asthma medication after treatment cessation

Data on asthma progression in patients with a history of treatment for asthma at the index

date are summarized in Table S4. During the pre-index period, the number of prescriptions of

symptomatic asthma medication per patient and per year was lower in the SLIT group (mean

± SD: 5.0 ± 4.7) than in the control group (9.9 ± 9.9 without age matching and 8.1 ± 7.9 with

age matching). During the treatment and follow-up periods, the number of prescriptions for

asthma medication per patient and per year decreased in the SLIT group and increased or

remain stable in the control group. The largest decrease in the SLIT group was observed

during the follow-up period. The receipt of asthma medication prescriptions among patients

with asthma in the pre-index period was markedly lower in the SLIT group than in the two

control groups (Table S4). The pre-index/follow-up ratio evidenced a 40% decrease in

This article is protected by copyright. All rights reserved.

 prescriptions for asthma medication in the SLIT group vs. 20% and 10% increases in the

control group (without and with age matching, respectively; Figure 3 and Table S4). After
Accepted Article
adjustment for covariates (gender, age class [<18 or ≥18], the main prescriber’s specialty, and

the number of AR prescriptions during the pre-index period), a linear regression analysis

showed that asthma progression was significantly slower in the SLIT tablets group than in the

control group - regardless of age matching and the study period (Table S5). The regression

coefficient [95% CI] in the SLIT group (versus the control group) during the follow-up

period was -0.58 [-0.74;-0.42] before age matching (p<0.0001) and -0.61 [-0.76;-0.46] after

age matching (p<0.0001) (Table S5). The total numbers of fulfilled prescriptions of asthma

medications are described by EphMRA class, study period and study group (after age

matching) in Table S6.

DISCUSSION

The results of the present real-world study of French patients receiving grass pollen SLIT

tablets (Oralair® and/or Grazax®) showed that the number of dispensed prescriptions for

symptomatic AR medication per patient and per year fell markedly after treatment cessation

in the SLIT group (a 50% decrease) but increased in the control group (a 30% increase with

age matching, and a 20% increase without age matching). Similarly, the number of patients

with a dispensed prescription for asthma medication (as a proportion of those who had not

been treated for asthma at index date) was lower in the SLIT group (13.7%) than in the

control group (17.0%). With regard to disease progression in patients considered to have

asthma at the index date, a large decrease (40%) in asthma medication prescriptions in the

SLIT group was observed during the follow-up period, relative to the pre-index period. In

contrast, increases of 20% and 10% (without and with age matching, respectively) were

observed in the control group.

This article is protected by copyright. All rights reserved.

 These differences in medication dispensing (considered as a proxy for disease diagnosis and

progression) indicate a potential effect of SLIT through a reduction in the burden of AR

Accepted Article
disease, less asthma onset, and the slower long-term progression of asthma.

A growing body of evidence suggests that AIT (which is primarily used to treat AR) can slow

the progression of the allergic march (i.e. progression to asthma) in some settings (12-14, 24,

31). Overall, the present results confirmed those obtained in Germany by Zielen et al., and

suggest that real-world data on SLIT’s effectiveness can be generalized to other countries

(29). However, the degree of long-term relief in AR and the extent of the reduction in asthma

onset and progression were apparently greater in the French study than in the German study

(see the Supplementary Material) (29). These disparities may have been due to differences in

prescribing habits and reimbursement between France and Germany or to methodological

differences between the studies.

Firstly, a larger number of medications for AR or asthma were recorded in the French study;

for example, nasal preparations (anti-allergy drugs and nasal corticosteroids, excluding

ATC/EphMRA R01A6) and ophthalmic anti-allergy drugs (ATC/EphMRA S01G1, S01G2

and S01G3) were recorded in the present French study but not in the German study (29). The

higher overall level of medication recorded in the French database influences the magnitude

of the decrease in symptomatic medication with SLIT, which was also greater in France than

in Germany. Secondly, antihistamines are only reimbursed in specific populations (i.e.

paediatric patients) in Germany and are thus not be recorded in the German database, which

only records prescriptions that cover the whole population. Thirdly, some data seem to

indicate that allergic disease was more severe in the French study than in the German study:

at the index date, the proportions of patients with asthma in the German study (21.2% in the

This article is protected by copyright. All rights reserved.

 SLIT group and 21.0% in the control group) were lower than in the present French study

(37.6% in the SLIT group and 39.2% in the control group) (29). Lastly, the follow-up period
Accepted Article
after treatment cessation was shorter in the present study (1 to 2 years) than in the German

study (2 to 6 years) because the French and German IMS LTD databases were created in

2012 and 2008, respectively; the longer the observation period, the more likely an event is to

occur.

The present study replicated many aspects of the German study (29), and thus had similar

limitations (29). The coverage of the French IMS LTD database was lower than that of the

German database; as of January 2017, it collected data on around 7,300 retail pharmacies in

mainland France and around 20 million patients (corresponding to about a third of the

national total) (32). Nevertheless, the IMS LTD database covers at least 20% of the retail

pharmacies in each administrative region of mainland France; this testifies to the database’s

robustness and the representativeness of the population sampled across France.

The main limitation of the present real-world study relates to the use of a proxy marker for

AR and asthma disease, i.e. the fulfilment of prescriptions for symptomatic medications. As

mentioned above, the study database did not include diagnostic information (e.g. allergen

sensitization) or clinical information (e.g. symptom severity or treatment regimens). For

example, our definition of “asthma onset” was not robust in the absence of a physician

diagnosis with an objective confirmation of airflow obstruction, peak flow reversibility

and/or asthma symptoms. Furthermore, the large control group constituted for our analysis

probably contained patients in whom the prime allergic trigger was not grass pollen (e.g.

another seasonal aeroallergen with an overlapping season, or a perennial allergen). However,

we sought to reduce this bias by focusing on prescriptions from December to July, i.e. during

This article is protected by copyright. All rights reserved.

 the pre-treatment period and the treatment period for the grass pollen season. The patients in

the control group were selected because they had fulfilled at least two prescriptions for nasal
Accepted Article
corticosteroids – a medication generally reserved for cases of moderate-to-severe AR – in

each of two successive pollen seasons. This increased the likelihood that the treated AR was

indeed due to seasonal exposure to grass pollen.

The control patients were matched with the patients in the SLIT group by index year, in order

to minimize the confounding bias due to differences in the length and intensity of the grass

pollen seasons from one year to another. There were much fewer patients in the SLIT group

than in the control group, although this reflected prescribing habits for grass pollen SLIT

tablets in France at the time of the study; we did not censor the inclusions, and included all

the SLIT patients and control patients meeting the study’s inclusion criteria. In the present

study, we also performed an age-matched analysis, with a view to minimizing confounding

bias caused by the large difference in age range between the SLIT tablet and control groups.

The results of the secondary (age-matched) analysis confirmed those of the primary analysis,

and strengthened our demonstration of SLIT’s effectiveness. However, after age matching,

we observed statistically significant differences between the SLIT tablet and control groups

with regard to gender, main prescriber’s medical specialty, and asthma status. We decided to

control for these for these potential residual confounders by adjusting the multivariate models

accordingly. Hence, gender, main prescriber, and age were included as covariates in all the

multivariate models. The covariate “asthma status” was included in the model of AR

progression, and the covariate “severity of AR” was included in the model for asthma onset

and progression (see the Supplementary Material for further details). Matching the SLIT

tablet group and the control group for more than the two selected criteria (year of the date

index, and age) would have made the analysis too complex, given the huge increase in the

This article is protected by copyright. All rights reserved.

 number of combinations (Year of the date index x Age class x Gender x Type of Prescriber x

Asthma status). Matching for so many criteria would have required the use of another
Accepted Article
methodology (such as a propensity score); this was not planned in the study protocol, and

would have constituted a major difference with regard to the study performed in Germany

(29). Indeed, the use of a propensity score would have reduced the number of matched

patients for analysis and thus the analysis’ power.

In summary, the present results (i) confirm the conclusions of the original German study (i.e.

the post-treatment effect of grass pollen SLIT tablets on AR may have a potential effect in

reducing asthma onset and asthma progression in routine clinical practice, using medication

dispensing as a proxy), (ii) suggest that the German results can be transferred to France, and

(iii) reinforce the transferability of these findings to other countries. Further research could

usefully confirm the longer-term maintenance of SLIT-associated benefits in patients treated

for three or more successive pollen seasons in clinical trials, and could also assess health

economic aspects of SLIT’s post-treatment effect.

ACKNOWLEDGMENTS

We thank David Fraser PhD (Biotech Communication SARL, Ploudalmézeau, France) for

copy-editing assistance, funded by Stallergenes Greer.

CONFLICT OF INTEREST

This study was funded by Stallergenes Greer (Antony, France). Copy-editing assistance was

also funded by Stallergenes Greer. X. Ansolabehere, I. Bardoulat, N. Coulombel, F. Maurel

and P. Le Jeunne are salaried employees of IQVIA (La Défense, France). P Devillier has

received fees for lectures and advisory boards from ALK, Astra Zeneca, Boehringer

This article is protected by copyright. All rights reserved.

 Ingelheim, Chiesi, GlaxoSmithKline, IMS Health GmbH & Co OHG, Meda Pharma,

Menarini, Novartis, Nycomed-Takeda, Sandoz , Stallergenes Greer and Teva outside the
Accepted Article
submitted work. M Molimard is employed by the University of Bordeaux France and reports

funding form ALK, Boehringer Ingelheim, IQVIA, GSK, Novartis Pharma, Stallergenes

Greer outside the submitted work. P. Demoly reports personal fees from ALK, Stallergenes

Greer, Chiesi, ThermoFisherScientific, Ménarini, Bausch&Lomb, and Mylan, outside the

submitted work.

AUTHOR CONTRIBUTIONS

All authors designed the study. I. Bardoulat and N. Coulombel, performed statistical

analyses. P. Devillier, P. Demoly, M. Molimard, X. Ansolabehere, I. Bardoulat and N.

Coulombel interpreted data and all authors contributed to the drafting of the manuscript

References

1. Bauchau V, Durham SR. Prevalence and rate of diagnosis of allergic rhinitis in

Europe. Eur Respir J 2004;24(5):758-764.
2. Sur DK, Scandale S. Treatment of allergic rhinitis. Am Fam Physician
2010;81(12):1440-1446.
3. Valovirta E. Respiratory Allergies. In: Associations EFoAaADP, editor.; 2011.
4. Ozdoganoglu T, Songu M. The burden of allergic rhinitis and asthma. Ther Adv
Respir Dis 2012;6(1):11-23.
5. Klossek JM, Annesi-Maesano I, Pribil C, Didier A. [INSTANT: national survey of
allergic rhinitis in a French adult population based-sample]. Presse Med 2009;38(9):1220-
1229.
6. Linneberg A, Dam Petersen K, Hahn-Pedersen J, Hammerby E, Serup-Hansen N,
Boxall N. Burden of allergic respiratory disease: a systematic review. Clin Mol Allergy
2016;14:12.
7. Bousquet J, Neukirch F, Bousquet PJ, Gehano P, Klossek JM, Le Gal M, et al.
Severity and impairment of allergic rhinitis in patients consulting in primary care. J Allergy
Clin Immunol 2006;117(1):158-162.
8. Shaker M. New insights into the allergic march. Curr Opin Pediatr 2014;26(4):516-
520.
9. Van Bever HP, Samuel ST, Lee BW. Halting the allergic march. World Allergy Organ
J 2008;1(4):57-62.

This article is protected by copyright. All rights reserved.

 10. Leynaert B, Neukirch C, Kony S, Guenegou A, Bousquet J, Aubier M, et al.
Association between asthma and rhinitis according to atopic sensitization in a population-
based study. J Allergy Clin Immunol 2004;113(1):86-93.
11. Guerra S, Sherrill DL, Martinez FD, Barbee RA. Rhinitis as an independent risk
Accepted Article
factor for adult-onset asthma. J Allergy Clin Immunol 2002;109(3):419-425.
12. Jacobsen L, Niggemann B, Dreborg S, Ferdousi HA, Halken S, Host A, et al. Specific
immunotherapy has long-term preventive effect of seasonal and perennial asthma: 10-year
follow-up on the PAT study. Allergy 2007;62(8):943-948.
13. Moller C, Dreborg S, Ferdousi HA, Halken S, Host A, Jacobsen L, et al. Pollen
immunotherapy reduces the development of asthma in children with seasonal
rhinoconjunctivitis (the PAT-study). J Allergy Clin Immunol 2002;109(2):251-256.
14. Niggemann B, Jacobsen L, Dreborg S, Ferdousi HA, Halken S, Host A, et al. Five-
year follow-up on the PAT study: specific immunotherapy and long-term prevention of
asthma in children. Allergy 2006;61(7):855-859.
15. Didier A, Wahn U, Horak F, Cox LS. Five-grass-pollen sublingual immunotherapy
tablet for the treatment of grass-pollen-induced allergic rhinoconjunctivitis: 5 years of
experience. Expert Rev Clin Immunol 2014;10(10):1309-1324.
16. Calamita Z, Saconato H, Pela AB, Atallah AN. Efficacy of sublingual immunotherapy
in asthma: systematic review of randomized-clinical trials using the Cochrane Collaboration
method. Allergy 2006;61(10):1162-1172.
17. Canonica GW, Passalacqua G. Sublingual immunotherapy in the treatment of adult
allergic rhinitis patients. Allergy 2006;61 Suppl 81:20-23.
18. Erny-Albrecht K, Valentine W, Christensen J, Vestenbaek U, AJ. P. Sublingual
immunotherapy in allergic rhinitis and asthma: a review of recent clinical evidence. J. Appl.
Res. 2007;7(1):17-31.
19. Incorvaia C, Di Rienzo A, Celani C, Makri E, Frati F. Treating allergic rhinitis by
sublingual immunotherapy: a review. Ann Ist Super Sanita 2012;48(2):172-176.
20. Dranitsaris G, Ellis AK. Sublingual or subcutaneous immunotherapy for seasonal
allergic rhinitis: an indirect analysis of efficacy, safety and cost. J Eval Clin Pract
2014;20(3):225-238.
21. Aboshady OA, Elghanam KM. Sublingual immunotherapy in allergic rhinitis:
efficacy, safety, adherence and guidelines. Clin Exp Otorhinolaryngol 2014;7(4):241-249.
22. Roberts G, Pfaar O, Akdis CA, Ansotegui IJ, Durham SR, Gerth van Wijk R, et al.
EAACI Guidelines on Allergen Immunotherapy: Allergic rhinoconjunctivitis. Allergy 2017:
Epub ahead of print.
23. Devillier P, Dreyfus JF, Demoly P, Calderon MA. A meta-analysis of sublingual
allergen immunotherapy and pharmacotherapy in pollen-induced seasonal allergic
rhinoconjunctivitis. BMC Med 2014;12:71.
24. Valovirta E, Petersen TH, Piotrowska T, Laursen MK, Andersen JS, Sorensen HF, et
al. Results from the 5-year SQ grass sublingual immunotherapy tablet asthma prevention
(GAP) trial in children with grass pollen allergy. J Allergy Clin Immunol 2018;141(2):529-
538 e513.
25. Halken S, Larenas-Linnemann D, Roberts G, Calderon MA, Angier E, Pfaar O, et al.
EAACI guidelines on allergen immunotherapy: Prevention of allergy. Pediatr Allergy
Immunol 2017;28(8):728-745.
26. Mahajan R. Real world data: Additional source for making clinical decisions. Int J
Appl Basic Med Res 2015;5(2):82.
27. Sherman RE, Anderson SA, Dal Pan GJ, Gray GW, Gross T, Hunter NL, et al. Real-
World Evidence - What Is It and What Can It Tell Us? N Engl J Med 2016;375(23):2293-
2297.

This article is protected by copyright. All rights reserved.

 28. Frieden TR. Evidence for Health Decision Making - Beyond Randomized, Controlled
Trials. N Engl J Med 2017;377(5):465-475.
29. Zielen S, Devillier P, Heinrich J, Richter H, Wahn U. Sublingual immunotherapy
provides long-term relief in allergic rhinitis and reduces the risk of asthma: A retrospective,
Accepted Article
real-world database analysis. Allergy 2018;73(1):165-177.
30. Jaeger S. Exposure to grass pollen in Europe. Clin Exp Allergy 2008;8:2-6.
31. Tsilochristou OA, Douladiris N, Makris M, Papadopoulos NG. Pediatric allergic
rhinitis and asthma: can the march be halted? Paediatr Drugs 2013;15(6):431-440.
32. Ordre National des Pharmaciens, Panorama au 1er Janvier 2017. Available at:
http://www.ordre.pharmacien.fr/Communications/Elements-demographiques/Les-
pharmaciens-Panorama-au-1er-janvier-2017 [In French]; 2017.

Figure 1a. Percentage of patients taking symptomatic AR medication during the follow-up
period in the SLIT tablet and control groups.

100%
4% 6%
90%

80% 37%

70%

60%

50%
96% 94%
40%

30% 63%

20%

10%

0%
SLIT group Control group Age-matched
n=1,099 n=27,475 control group
n=10,990

yes no

This article is protected by copyright. All rights reserved.

 Figure 1b. Relative change in symptomatic AR medication between the pre-index period (set
to 100 arbitrary units) and the follow-up period.
Accepted Article
140

120
+30% +20%
100

80
-50%
60

40

20

0
SLIT group Control group Age-matched control group

Pre-index period Follow-up period

To illustrate changes over time in symptomatic prescriptions, data from the pre-index period were
arbitrarily scaled to 100%

This article is protected by copyright. All rights reserved.

 Figure 2. Time to asthma onset, defined as time to the date of first prescriptions of short-
acting β-agonists or inhaled corticosteroids for sublingual immunotherapy (SLIT) and non-
AIT groups during the follow-up period, in patients without asthma at the end of treatment
Accepted Article
period (note the offset of the y-axis)

This article is protected by copyright. All rights reserved.

 Figure 3. Relative change in symptomatic asthma medication between the pre-index period

(set to 100 arbitrary units) and the follow-up period.

Accepted Article
140

120 +20 +10

% %
100

80 -40%

60

40

20

0
SLIT group Control group Age-matched control group

Pre-index period Follow-up period

This article is protected by copyright. All rights reserved.

 Table 1. Descriptive statistics: symptomatic allergic rhinitis (AR) prescriptions per
patient and per year in each group in pre-index and follow up periods (with and without
age matching)
Accepted Article
Number of symptomatic AR prescriptions
per patient and per year Mean ± SD Range Median IQR
Main analysis (no age matching)*
Pre-index period
SLIT group (n=1,099) 6.5 ± 5.6 1.0;49.0 5.0 [3.0;9.0]
Control group (n=27,475) 11.5 ± 9.0 1.0;97.0 9.0 [5.0;16.0]
p-value <0.001
Follow-up period
SLIT group (n=1,099) 2.6 ± 4.2 0.0;41.8 0.8 [0.0;3.2]
Control group (n=27,475) 11.0 ± 9.6 0.0;93.4 8.5 [4.0;15.6]
p-value <0.001
Pre-index/follow-up ratio
SLIT group (n=1,099) 0.5 ± 1.3 0.0;26.7 0.2 [0.0;0.7]
Control group (n=27,475) 1.3 ± 1.7 0.0;44.0 1.0 [0.5;1.5]
p-value <0.001
Secondary analysis (age matching) **
Pre-index period
SLIT group (n=1,099) 6.5 ± 5.6 1.0;49.0 5.0 [3.0;9.0]
Age-matched control group (n=10,990) 9.6 ± 7.7 1.0;72.0 7.0 [4.0;13.0]
p-value <0.001
Follow-up period
SLIT group (n=1,099) 2.6 ± 4.2 0.0;41.8 0.8 [0.0;3.2]
Age-matched control group (n=10,990) 8.6 ± 8.1 0.0;73.9 6.2 [2.8;11.9]
p-value <0.001
Pre-index/follow-up ratio
SLIT group (n=1,099) 0.5 ± 1.3 0.0;26.7 0.2 [0.0;0.7]
Age-matched control group (n=10,990) 1.2 ± 1.6 0.0;34.2 0.8 [0.4;1.4]
p-value <0.001

AR: allergic rhinitis; IQR: interquartile range; SD: standard deviation; SLIT: sublingual
immunotherapy tablet
*To avoid confounding bias due to differences in the length and intensity of the grass-pollen
season in different years, patients in the SLIT tablet and control groups were matched by
index year (the final SLIT:control matching ratio was 1:25).
**To avoid confounding bias due to differences in the patients’ age between the two groups,
patients in the SLIT tablet and control groups were matched by index year and age (the final
SLIT:control matching ratio was 1:10).

This article is protected by copyright. All rights reserved.

 Table 2. Risk of initiation of asthma medication* onset in patients without asthma
during the pre-index period, as a function of the study period (logistic regression, SLIT
group vs. control group, in the main and secondary analyses)
Accepted Article
Odds 95%CI p-value
ratio
Main analysis (no age matching)**
Treatment period 0.634 0.470‒0.857 0.0030
Follow-up period 0.375 0.199‒0.709 0.0025
Full analysis period (treatment + follow-up) 0.560 0.438‒0.717 <0.0001
Secondary analysis (age matching)***
Treatment period 0.564 0.410‒0.776 0.0004
Follow-up period 0.363 0.193‒0.685 0.0018
Full analysis period (treatment + follow-up) 0.549 0.431‒0.700 <0.0001

CI: confidence interval; SLIT: sublingual immunotherapy tablet

*Based on asthma medication fulfilment data.
**To avoid confounding bias due to differences in the length and intensity of the grass-pollen
season in different years, patients in the SLIT tablet and control groups were matched by
index year (the final SLIT:control matching ratio was 1:25).
***To avoid confounding bias due to differences in the patients’ age between the two groups,
patients in the SLIT tablet and control groups were matched by index year and age (the final
SLIT:control matching ratio was 1:10).

This article is protected by copyright. All rights reserved.

 Table 3: Description of the occurrence of asthma between different treatment groups

Patients with SLIT group Control group p-value

Presence of asthma asthma n=1,099 n=27,475
Accepted Article
medication
During the pre-index period No 686 (62.4%) 16699 (60.8%) 0.301
Yes 413 (37.6%) 10776 (39.2%)
Over the treatment period among patients No 603 (87.9%) 14645 (87.7%) 0.880
without asthma in the pre-index period Yes 83 (12.1%) 2054 (12.3%)
During the follow-up period among patients No 592 (86.3%) 13863 (83.0%) 0.005
without asthma during the pre-index period Yes 94 (13.7%) 2836 (17.0%)

SLIT: sublingual immunotherapy

This article is protected by copyright. All rights reserved.