Respiratory disease phenotypes in a general population sample: latent transition analysis

Risultato della ricerca: Meeting Abstractpeer review


Background: Few data are available on the temporal pattern of respiratory disease phenotypes in generalpopulation.Aim: To detect longitudinal patterns of disease phenotypes related to risk factors and physician visits.Methods: Pisan general population sample from 2 cross-sectional studies (PI1: 1991-93; PI2: 2009-11;n=1107), questionnaire-based data. Latent transition analysis (LTA) was performed to assess respiratorydisease phenotypes at PI1 and PI2, labelled according to disease/symptom occurrence. Possible patternswere persistence, worsening and improvement of the phenotype. Multiple logistic regression models wereestimated to assess the association among phenotype patterns, risk factors (smoking habits-SH andoccupational exposure-OE) and physician visits, adjusting for age, sex and educational level.Results: 4 cross-sectional phenotypes were ranked by severity in PI1 and in PI2: healthy (H) (59 and 55%),allergic rhinitis (AR) (28 and 27%), cough/phlegm (CP) (11 and 15%), asthma (A) (3 and 4%). Thelongitudinal patterns were: persistent H (53%), persistent AR (27%), persistent CP (9%), improvement(2%), worsening (10%).Significant associations: persistent CP with persistent SH (OR 5.9), persistent OE (OR 2.9) and incident OE(OR 2.2); persistent CP and worsening of phenotype with current visits of family physician (OR 5.4 and OR4.7, respectively) and of specialist (OR 2.7 and OR 3.9, respectively).Conclusions: LTA allowed to identify four different phenotypes based on respiratory symptoms/diseasesand their longitudinal patterns over 18 years. Such analysis brings new perspectives in the analyses ofpopulation-based data.
Lingua originaleEnglish
Numero di pagine2
RivistaEuropean Respiratory Journal
Stato di pubblicazionePublished - 2018


Entra nei temi di ricerca di 'Respiratory disease phenotypes in a general population sample: latent transition analysis'. Insieme formano una fingerprint unica.

Cita questo