PHENOTYPE CHARACTERIZATION OF COPD



ONLINE SUPPLEMENT

IDENTIFICATION AND PROSPECTIVE VALIDATION OF CLINICALLY RELEVANT CHRONIC OBSTRUCTIVE PULMONARY DISEASE (COPD) SUBTYPES

Judith Garcia-Aymerich, Federico P Gómez, Marta Benet, Eva Farrero, Xavier Basagaña, Àngel Gayete, Carles Paré, Xavier Freixa, Jaume Ferrer, Antoni Ferrer, Josep Roca, Juan B Gáldiz, Jaume Sauleda, Eduard Monsó, Joaquim Gea, Joan A Barberà, Àlvar Agustí, and Josep M Antó, on behalf of the PAC-COPD Study Group.

This supplement includes:

- METHODS (COMPLETE VERSION)

- Table 1. Number of subjects in the total sample and subsamples.

- Table 2. Number of missing values for each variable.

- Figure 1. Flow chart of the PAC-COPD study.

- Table 3. Complete description (non-standardized values of variables) of the three COPD Groups identified by cluster analysis, for each of the six cluster analyses (total sample and the five subsamples ).

- Table 4. Influence of severity in the association between the three COPD Groups identified by cluster analysis and longitudinal outcomes (Cox regression models).

- SUPPLEMENT REFERENCES

METHODS (COMPLETE VERSION)

Design

The present study includes a cross-sectional analysis to identify COPD groups and a 4-years prospective assessment of their relationship with cause-specific admissions and all-cause mortality.

Subjects

We recruited all subjects hospitalized for the first time because of an exacerbation of COPD in nine teaching hospitals in Spain between January 2004 and March 2006. Any hospital stay or time spent in the emergency room for at least 18 hours, with a clinical diagnosis of COPD exacerbation was considered as admission. A COPD exacerbation was defined as “a sustained worsening of the patient's condition, from the stable state and beyond normal day-to-day variations, that is acute in onset and necessitates a change in regular medication in a patient with underlying COPD”.[[i]] The criteria for a first admission definition was established by means of a questionnaire, the patient’s clinical record and a search of the hospital records. The exclusion criteria were: 1) patients under 45 years of age; 2) severe comorbidity, i.e. tuberculosis with residual lesion affecting more than 1/3 of parenchyma, pneumectomy or diagnosed pneumoconiosis, advanced cancer, psychiatric disorder, severe cardiovascular or neurological disease, and other; 3) mental incapacity; 4) frail or elderly patients with any disability that would hinder participation in the study; 5) not living in the healthcare area of that particular hospital; and 6) not understanding the language. The diagnosis of COPD was confirmed by spirometry when the patient had reached clinical stability (at least three months after discharge), according to the criteria of the American Thoracic Society (ATS) and the European Respiratory Society (ERS) (postbronchodilator forced expiratory volume in one second to forced vital capacity ratio ≤0.7).[[ii]] The protocol was approved by the Ethics Committees of all the participating hospitals, and written informed consent was obtained from all subjects. Detailed information about recruitment methods has been previously published.[[iii]]

Measurements

Detailed information about the methods used, sources of questionnaires and standardization of the tests used in the study have been published elsewhere.[[iv]] In brief, when patients were considered clinically stable, at least three months after discharge, they underwent extensive characterization that required from two to four visits, in separate days, to the hospital, and included the following: (1) a computerized epidemiologic questionnaire, covering sociodemographic and environmental exposure data, smoking habits and cumulative exposure, dietary habits, self-reported comorbidities, previous treatments and diagnoses, respiratory symptoms, health-related quality of life (St George’s Respiratory Questionnaire),[[v]] activities of daily living (Barthel index),[[vi]] Epworth Sleepiness Scale,[[vii]] and the Hospital Anxiety and Depression Scale[[viii]]; (2) detailed physical examination and Charlson comorbidity index[[ix]] obtained by a respiratory physician participating in the study; (3) body composition determination by bioelectric impedance;[[x]] (4) complete lung function tests including forced spirometry, bronchodilator test, body plethysmography, carbon monoxide diffusing capacity, resting arterial blood gases, respiratory and peripheral muscle strength, and night-time pulse oximetry;[[xi],[xii],[xiii],[xiv]] (5) exercise capacity assessment by the 6-minute walk test (6MWT);[[xv],[xvi]] (6) chest radiographs;[[xvii]] (7) skin prick tests; (8) induced sputum;[[xviii]] (9) routine laboratory analyses in each of the participating hospitals, including peripheral blood cell counts, cholesterol, triglycerides and total IgE; (10) measurements of serum inflammatory and oxidative stress markers, to reflect systemic inflammation, centralized in a single laboratory; and, (11) Doppler echocardiography evaluation[[xix],[xx]] for centralized readings (after testing for reliability in a random sample). All these previous measurements were done in the total sample of patients (n=342). Additionally, due to their technical and logistical requirements, the following measurements could be performed only in some of the participating hospitals: (12) lung density and emphysema quantification from computed tomographies (CT)[[xxi]] by a centralized evaluation using PulmoCT (Siemens, Munich, Germany) (Subsample 1: three hospitals, 102 subjects); (13) semiquantitative evaluation of bronchial wall thickness[[xxii]] on CT scans by four trained radiologists (Subsample 2: four hospitals, 148 subjects); (14) microbiological culture of sputum performed in the microbiological laboratories of the hospitals (Subsample 3: 8 hospitals, 224 subjects); (15) sputum samples processed following the selection plug method[18] with centralized measurement of inflammatory markers and differential cell counts (Subsample 4: 7 hospitals, 181 subjects); and, (16) a cardiopulmonary incremental exercise test (CPET) with cycloergometer[[xxiii],[xxiv]] to measure exercise capacity (Subsample 5: 6 hospitals, 200 subjects). The overlap between patients in Subsamples is shown in Table 1 of the Supplement.

Quality control consisted in centralized training sessions, rapid support and supervision of all fieldworkers, recording and checking of questionnaires administration, double verification of case report forms, and at least one visit to each of the participating centers during data collection. Further details of quality control have also been previously published.[4]

Information on re-hospitalizations up to December 31st 2007 was obtained for all patients in the cohort from the Minimum Basic Dataset (CMBD), a national administrative database. According to the International Classification of Diseases, 9th revision, an admission for a COPD exacerbation was defined as any admission with codes 466, 480-486, 490-496, or 518.81 as the main diagnosis. Cardiovascular admissions were defined by codes 390 to 459. Survival status until December 31st 2008 was obtained from direct interview to the patients or their relatives. In case of death, both hospital and primary care registries were checked to ensure exact date. At the stage of the current analyses only all-cause mortality was available.

Statistical analysis

Sample size calculations

There are no sample size calculation formulas for cluster analysis. However, our study included a sample size equivalent to or greater than previous studies that were able to identify heterogeneous groups in COPD populations even when they used a much smaller set of variables.[[xxv],[xxvi]] The use of clustering methods has become widespread in the context of high-dimension low-sample size data, such as in microarray data analysis, where the ratio of number of subjects to number of variables is often as low as 0.01.[[xxvii]] In our study, the number of subjects was greater than the number of variables with the exception of Subsamples 1 and 2. In addition, our a priori sample size calculations, as published in the original protocol,[4] predicted that with a recruitment of between 300 and 400 patients with COPD, readmission rates of 50% and 60% after one and two years,[[xxviii]] and mortality rates of 15% and 20% after one and two years,[28] we would have a statistical power greater than 80% to detect relative risks of 1.5 for readmission and 1.8 for mortality at the 0.05 significance level when comparing two hypothetical groups (phenotypes) derived from the cluster analysis.

Variables selection and treatment of missing data

A total of 536 variables were obtained from the different tests and questionnaires. Among them, a total of 224 were considered after excluding: (i) variables with additive relations (e.g., % fat free mass was kept, instead of keeping both % fat mass and % fat free mass); and, (ii) those resulting from categorizations (e.g., continuous FEV1 was kept, instead of keeping both continuous FEV1 and COPD severity according to stratification of FEV1).

The complexity of data collection produced missing values in almost all variables (Supplement Table 2). Most of them were found in a small proportion of subjects of the corresponding (sub)sample ( ................
................

In order to avoid copyright disputes, this page is only a partial summary.

Google Online Preview   Download