Health-Related Quality of Life Varies in Different Respiratory Disorders

A Multi-case Control Population Based Study

Veronica Cappa; Alessandro Marcon; Gianfranco Di Gennaro; Liliya Chamitava; Lucia Cazzoletti; Cristina Bombieri; Morena Nicolis; Luigi Perbellini; Silvia Sembeni; Roberto de Marco; Francesco Spelta; Marcello Ferrari; Maria Elisabetta Zanolin


BMC Pulm Med. 2019;19(32) 

In This Article


The GEIRD project is an ongoing multicase-control study, coordinated by the Verona centre, involving seven Italian centres.[23] In the first stage of the study, new random samples and pre-existing cohorts (the Italian Study on Asthma in Young Adults (ISAYA)[24] and the Italian arm of the European Community Respiratory Health Survey (ECRHS)[25]) from the general population (20–64 years, male/female = 1/1) were mailed a screening questionnaire on respiratory symptoms (Figure 1). In particular, for the new random sample, 3000 subjects aged 20–44 and 1000 subjects aged 45–64, male/females =1/1, from the general population were selected. In the second stage of GEIRD, on the basis of answers to the screening questionnaire (Additional file 1 (a)), all probable cases of asthma and COPD/CB, a sample of probable cases of AR (44%), other condition (39%), and controls (61%), were invited to clinics to be phenotyped.[26] During the clinical visit, each subject underwent a computer-assisted clinical interview, lung function[27,28] and allergological tests[29] (Additional file 1 (b)), and the 36-item Short Form (SF-36) questionnaire. All the protocols were in agreement with the international guidelines and can be found on the GEIRD website ([30] Ethical approval was obtained in each centre from its appropriate ethics committee, and written consent was obtained from each participant.

Figure 1.

Timeline for the new random sample and the cohorts of the GEIRD study

In this paper, only the data of the Verona centre were considered, because the data collection was at an initial stage in the other centres.

Quality of Life Questionnaire – the SF-36

As a part of the GEIRD core protocol, all the participants' HRQL was assessed using SF-36 questionnaire version 1.6, which is a generic quality of life, self-administered measure containing 36 items.[27,31] Physical Component Summary (PCS) and Mental Component Summary (MCS) measures were calculated. Missing data were treated as recommended in the SF-36 manual and interpretation guide.[28]

Identification of Cases and Controls in Clinics

The subjects were hierarchically classified as follows:

  • 28 cases of COPD (he/she had post-bronchodilator FEV1/FVC < LLN or < 70%);

  • 224 cases of current asthma ("CA", he/she reported lifetime asthma; or he/she reported asthma-like symptoms/medicines in the last 12 months and had (1): a positive methacholine challenge test with PD20 < 1 mg or (2) pre-bronchodilator FEV1/FVC < 70% or < LLN with a positive reversibility test (i.e. FEV1 > 12% and > 200 mL after the administration of 400 μg of salbutamol); all cases already defined as COPD cases could not be defined as CA cases;

  • 126 cases of past asthma ("PA", he/she reported lifetime asthma but did not fulfil the criteria for CA); all cases already defined as COPD cases could not be defined as PA cases;

  • 48 cases of CB (he/she was not a COPD or CA/PA case and he/she reported chronic cough or phlegm (> 3 months/year for at least 2 years));

  • 163 cases of allergic rhinitis (AR) and 95 non-allergic rhinitis ("NAR") (he/she was not a COPD, CA/PA, BC case and he/she had nasal allergies or nasal problems in the presence of animal(s), pollens, dust plus negative (NAR) or positive (AR) skin prick test);

  • 328 controls (subjects without any nasal/respiratory symptoms/conditions reported in the clinical questionnaire neither in the clinic nor in the screening questionnaire).

The Additional file 2 shows the details of this hierarchical classification and the overlapping among cases.

Potential Determinants of HRQL and Covariates

The relationship between case-control status (COPD, CA, PA, AR, NAR, CB and controls, the independent variable) and HRQL (the dependent variable) was investigated. We took into account the following potential confounders: gender, age (years), body mass index (BMI, kg/m2), education level (low if a subject had completed full-time education before the age of 16, high otherwise), smoking habits (never smoker, past smoker = ever smoker who did not smoke in the last month, current smoker = subject who smoked in the last month), presence/absence of at least one non-respiratory comorbidity (gastritis, stomach ulcer, gastroesophageal reflux, hiatal hernia, esophagitis, osteoporosis, gout, arthritis, osteoarthritis, pulmonary embolism, diabetes, stroke, cancer), presence/absence of at least one heart disease (heart attack (coronary thrombosis), angina, arrhythmia, hypertension and other heart problems).

The analyses were also adjusted for study sample/cohort (ECRHS Italy, ISAYA, new random sample) and calendar period, which divided the study period into semesters, starting from the beginning of stage 2 (April 2008). The latter covariate took into account both seasonality (April to September, the "warm/hot" season in Italy, compared to October to March, the "cool/cold" season) and potential temporal trends.

Statistical Analysis

Data were summarized as counts and percentages, means (standard deviation (SD)) and medians (interquartile difference (IQD)). The Chi-square and the Kruskal-Wallis tests were used to investigate differences in variables among respiratory conditions, where appropriate. As scores were not normally distributed, differences in the crude medians of PCS and MCS across cases and controls were tested using the Kruskal-Wallis test.

Quantile regression models[32] were applied to study the relationship between HRQL and respiratory diseases, controlling for the other potential determinants, estimating conditional medians of the response variables (PCS and MCS). The quantile regression coefficients are interpreted as the ordinary regression ones and they indicate the change in the dependent variable for every one unit change (or category for nominal variables) in each covariate.[33] As higher PCS and MCS scores are index of better HRQL, a negative regression coefficient indicates that an increase in the independent variable worsens HRQL and vice versa.

Statistical analyses were performed using Stata Statistical Software: Release 14.0 (