Study population and subpopulations

Eligible participants are male or female patients with SA as defined in Table 1^2,29, ≥18 years of age at the time of informed consent signature, with documented physician-diagnosed asthma for at least 12 months. During the 12 months prior to enrolment, patients need to have at least two BEC, one immunoglobulin E (IgE), and one FeNO measurements available; alternatively, such measurements need to be planned by the physician to be performed at the study visit, as per their routine practice and independently of their decision to include the patient in the current study, and thus be available at enrolment. For patients who are treated with omalizumab on the day of enrolment, at least one IgE measurement must be available before omalizumab initiation. All participants must provide written informed consent prior to inclusion in the study. Key eligibility criteria are listed in Figure 1.

Figure 1

Overview of study design and data collection variables

Patients will be classified by SA phenotype using two definitions, based on a published consensus-driven algorithm previously used to categorize patients enrolled in the International Severe Asthma Registry³⁰, with differentiations newly proposed herein according to expert’s input in the study team. These two definitions, namely BASE and STRICT, will classify patients into ‘T2-low (definite or possible)’, ‘definite T2-low’, and ‘possible T2-low’; and ‘T2-high (definite or possible)’, ‘definite T2-high’, and ‘possible T2-high’. Given that T2 low is characterized on the basis of non-detection of markers of T2 inflammation, and in order to be more aligned with international terminology³⁰, ‘definite T2-low’ could also be referred to as ‘unlikely T2-high’, ‘possible T2-low’ as ‘least likely T2-high’, ‘possible T2-high’ as ‘likely T2-high’, and ‘definite T2-high’ as ‘most likely T2-high’. Both definitions will be based on the same composite measure of airway inflammation markers (i.e. BEC and FeNO), allergic/atopic status and age-of-onset phenotype (as defined in Table 1), but will employ different airway inflammation marker thresholds, also taking into consideration the current receipt of biologic and/or maintenance oral corticosteroid (mOCS) treatment and response to this treatment (as applicable) as presented in Figures 2 and 3. Patients within each T2-low subpopulation (‘definite T2-low/unlikely T2-high’, and ‘possible T2-low/least likely T2-high’) will be further stratified into subgroups based on their asthma symptom control level at the study visit using the Asthma Control Test^TM (ACT) into ‘controlled’ (ACT score ≥20) and ‘uncontrolled’ (ACT score <20).

Figure 2

Criteria and scoring system for classification of SA phenotypes based on study definitions

Figure 3

Composite scoring system

Study objectives

The primary objective of this study is to assess the prevalence of T2-low asthma phenotype using the BASE definition among patients with SA treated in routine care settings in Greece. Secondary objectives include assessment of the prevalence of T2-low asthma phenotype using the STRICT definition, assessment of the level of asthma symptom control in the T2-low subpopulation [‘overall (definite or possible)’, ‘definite’, and ‘possible’, as per BASE and STRICT definitions separately], as well as characterization of the patient demographic and clinical profile, current and past therapeutic management strategies, clinically significant asthma exacerbation (CSE) burden (as defined in Table 1) over the past 12 months, and patients’ lung function, as assessed by spirometry at the study visit, in the T2-low subpopulation (using the BASE and STRICT definitions separately) and its subgroups by asthma symptom control level (‘controlled’ and ‘uncontrolled’). Furthermore, the study will assess the asthma-specific QoL using the Mini – Asthma Quality of Life Questionnaire (Mini-AQLQ), the anxiety and depression levels using the hospital anxiety and depression scale (HADS), the work productivity loss and activity impairment using the Work Productivity and Activity Impairment: Respiratory Symptoms (WPAI:RS) questionnaire at the study visit, the asthma-related HCRU in the past 12 months before the study visit, and the frequency of OCS-dependent patients, in the T2-low SA subpopulation and its subgroups by asthma symptom control level.

Exploratory objectives of this study include the assessment of the influence of selected patient and disease characteristics on the prevalence of T2-low SA (according to the BASE definition); the identification of potential patient, treatment and disease characteristics influencing ACT-assessed asthma symptom control in the T2-low SA subpopulations as per the BASE and STRICT definitions separately; and the examination of the differences in patient, disease, and treatment characteristics between the subgroups of patients with definite and possible T2-low SA in the T2-low SA subpopulations as per the BASE and STRICT definitions separately. The study objectives and the (sub)populations to which they refer are summarized in Table 2.

Study design

PHOLLOW (a real-world, multicenter, cross-sectional and retrospective chart review study to assess the prevalence of asthma inflammatory PHenotypes and to characterize the patient prOfiLe, clinical features and disease burden of type-2 LOW severe asthma in routine care settings in Greece) is an epidemiological, single-country, multicenter, cross-sectional with retrospective chart review two-part study, with a single-visit data collection schedule. The study will include a representative sample of outpatients with SA treated by asthma specialists (pulmonologists and allergists) in routine care settings in Greece. The planned recruitment period, and, consequently, the expected duration of the study is 12 months, but will actually be determined by achievement of the target sample size. Each patient’s look-back period for data collection from medical charts extends from the date of asthma symptom onset to the date of obtainment of informed consent. The study design including data collection variables at the study visit is summarized in Figure 1.

In part A of the study, appropriate top-level patient data are collected from all enrolled eligible SA patients to assess the prevalence and identify potential predictors of T2-low SA. The most recent BEC/FeNO/IgE measurements recorded in the 12-month period prior to and including the study visit, along with allergic/atopic status, current receipt of biologic and/or mOCS treatment, and response to biologic and/or mOCS treatment are used for stratification of patients into SA phenotypes; an exception applies for patients treated with omalizumab at the study visit, for whom the most recent IgE measurement before omalizumab initiation is used for determining the allergic/atopic status, in case that the patient meets one of the other two qualifying conditions [i.e. Skin Prick Test (SPT) positivity or Early-Onset Asthma (EOA)]. SPT results are collected, if needed, prior to omalizumab initiation. For patients classified as T2-high SA according to the BASE definition, study participation ends in part A, and no further data are collected about them.

Part B of the study only concerns patients classified as T2-low as per the BASE definition and involves further data collection and patient-reported outcome (PRO) assessments, in order to address the applicable study secondary, and exploratory objectives.

The study is designed and will be reported in accordance with the ethical principles laid down in the Declaration of Helsinki, the Guidelines for Good Pharmacoepidemiology Practice of the International Society for Pharmacoepidemiology, the STROBE guidelines where applicable, the EU General Data Protection Regulation, and the local rules and regulations.

Site selection and patient enrolment

Candidate investigators originate from a pool of specialists, both pulmonologists and allergists, across Greece, with efforts made to recruit sites with representative medical practices and to provide data describing the many real-life aspects of asthma management and its burden on patients and the healthcare system.

Site selection has been performed in a non-random manner, through a documented and constructed feasibility assessment process accounting, among other factors, for physicians’ qualifications, number of potentially eligible patients, and representativeness based on hospital’s patient care activity (number of cared SA patients during the 12-month period before start of recruitment), hospital location (geographical region), and type of healthcare site/institution (publicly owned non-academic site/institution, university hospital, privately-owned hospital, private practice). Based on preliminary feasibility projections, between 20 and 40 patients are to be recruited by each site, with the planned number of sites being between 20 and 30.

To minimize patient selection bias, physicians are requested to consecutively enroll the first eligible patients attending their clinic/office over the pre-specified study recruitment period. To enable assessment of potential selective enrolment, physicians are asked to complete a screening log of all potentially eligible patients, i.e. of patients who meet study eligibility criteria regardless of whether they agree or are able to participate or not, including the minimum required information of date seen, sex, age range on the date seen, and reason for non-consenting.

Data collection

The study involves both primary and secondary data collected by means of a web-based data capture system [electronic Case Report Form (eCRF)] adhering to all applicable data protection regulations and requirements with regard to electronic records and database validation.

Data are collected directly by the participating physicians as generated within the normal clinical practice setting, through routine clinical assessments performed in the context of the single study visit as well as patient self-report (where applicable) and use of PROs. Data collection includes the variables and measurements that are described in Figure 1 and Table 3. Patient source data pertaining to medical and asthma-related history, exacerbation history, hospital-based HCRU over the past 12 months, results of airway inflammation biomarkers of interest, and asthma management practices are abstracted from patients’ medicals records and patient self-report. PROs are completed by the patients themselves via self-administered paper questionnaires of the validated Greek versions (Table 4)^31-34. The scores/answers are subsequently transcribed into the eCRF directly by the participating physicians.

In line with the purely observational and non-interventional nature of the study, no further clinical, laboratory, imaging, and lung function assessments are required apart from those performed as per the treating physician’s routine medical practice.

Internal validity of the outcomes is safeguarded to the extent feasible with the implementation of appropriate source data verification and quality assurance measures.

Sample size

This study’s primary aim is purely descriptive in nature. A sample size of 600 patients, offers a margin of error of 4.1% [with 95% confidence interval (CI) using the Clopper-Pearson exact method ranging between 45.9% and 54.1%] for the estimation of a percentage of 50% where the width of the CI is largest. Assuming that a proportion of at least 20% in the overall eligible population will be classified as T2-low according to the BASE definition, the study size that will be used for analysis of the applicable secondary objectives is expected to exceed 100 patients. This means that regarding any qualitative variable observed at any frequency >0%, the margin of error would range from >3.6% to 10.2%.

Statistical analysis

All analyses will be performed in the set of eligible enrolled patients among each subpopulation of interest provided that the number of patients included in each subgroup is sufficient and allows for meaningful estimates.

Continuous variables will be expressed using descriptive statistical measures, while categorical variables will be expressed using frequency tables. The 95% Clopper-Pearson CI for binomial proportions and the 95% Poisson CI for incidence rates will be estimated. The normality of distribution of continuous variables will be examined using the Shapiro-Wilk test. Comparisons of continuous variables will be performed using the t-test (parametric) or Mann-Whitney U (non-parametric) test for two independent samples, while comparisons of categorical variables will be performed using either the Pearson’s chi-squared test or Fisher’s exact test.

The association of patient, treatment and disease characteristics with binary and continuous outcomes will be evaluated through univariable and multivariable binary logistic regression and linear regression models, respectively. In regard to multivariable regression analysis, the stepwise procedure based on the minimization of the Akaike’s information criterion will be applied for the final model selection. All models will be fitted, provided that the number of patients included is sufficient to allow for meaningful inferences. In addition, the type of institution (academic/non-academic) and the administrative geographical region of the study site (Attica, outside Attica) will be considered as potential confounders on the association of patient and disease characteristics with the prevalence of T2-low SA and asthma symptom control level.

No imputation methods will be applied for missing data, with the exception of partial dates. With regard to the scoring and handling rules of any missing PRO items, the relevant scoring instructions, where available, will be followed. All statistical tests will be two-sided and will be performed at a 0.05 significance level.

Limitations

There are several limitations inherent to the observational cross-sectional and retrospective design. Mechanisms have been put in place to mitigate confounding due to potential patient selection bias, as well as information/recall and response bias in patient report through use of a consecutive sampling process, validated and standardized instruments involving short recall periods, and specific timing for PRO completion before study-related assessments, respectively. To account for potential confounding on the primary study outcome, the possible influence of confounding factors will be examined through multivariable analyses, as an exploratory outcome. It should also be noted that although efforts have been made to recruit sites with representative medical practices, the sampling frame was based on a non-probability technique, with physicians being selected in a non-random manner; thus, the generalizability of the study outcomes is indeterminate. Another foreseen limitation of PHOLLOW pertains to the sample size that will be eventually used for assessment of the secondary and exploratory objectives, since the actual prevalence of T2-low is unknown. Nevertheless, the number of T2-low patients in the study population is expected to exceed 100, which is considered sufficient to characterize with acceptable precision this subpopulation. Systematic error arising from inaccurate measurements or missing data is expected to be mitigated by having as a prerequisite for study eligibility the availability of sufficient relevant medical records. Nonetheless, inaccuracies in the assessment of asthma-related HCRU in the last 12 months before enrollment may still be present, as visits performed in an outpatient setting may not be fully recorded in the patients’ records.

Despite these limitations, studies of such design essentially, through collecting data about real-world management practices and their outcomes, help to bridge clinical research in strictly controlled randomized settings with daily clinical practice. As real-world studies follow less restrictive methodological standards than controlled trials in terms of patient selection, treatment, and other design aspects, their results are better generalizable, especially for populations with diseases of complex and heterogeneous biology, such as asthma. Moreover, enrolment of patients from diverse locations throughout Greece and different types of healthcare institutions, will further enhance the generalizability of the findings.