Grant: Catalyst Grant: Analysis of CLSA Data
McMaster University
$70,000.00
How to Handle Missing Spirometry Data in Population-Based Studies?
FY 2019-20
Summary
Spirometry has increasingly been utilized in population-based studies including the Canadian Longitudinal Study on Aging (CLSA), to assess the presence of lung disease at the population level. According to the literature, about 10 to 30% of participants in population-based studies do not have valid spirometry data due to either contraindication from spirometry tests or poor quality spirometry. Of the 30,097 participants within the CLSA comprehensive cohort, 29.1% have no spirometry data available at baseline due to low quality spirometry or contraindication for spirometry and are considered as "missing" in the statistical analysis. The substantial amount of missing spirometry data may lead to efficiency reduction and serious bias, which undermine the validity of research findings.
Two measures are commonly collected from the spirometry for evaluation of lung function: forced expiratory volume in one second (FEV1) and forced vital capacity (FVC), which measure flow rate and lung volume in one maximal breath. The ratio of FEV1 to FVC together with FEV1 and FVC are used to identify obstructive and restrictive lung defects. The FEV1 and FVC are found to have a very strong, age-independent linear relationship in healthy individuals; however, this relationship does not hold among those with lung impairment. Traditional imputation techniques are inappropriate to handle the missing spirometry data since (1) FEV1 and FVC are paired measures from one spirometry test and they are observed or missing simultaneously; and (2) the relationship between FEV1 and FVC depends on whether an individual has lung impairment.
This investigation will help researchers appreciate the complexity of missing spirometry data and adopt appropriate imputation methods to handle them. Moreover, methods developed in this investigation can be generalized to handle the missingness of any paired data, such as systolic and diastolic blood pressure, a common phenomenon in health research.
Department
Program
Catalyst Grant: Analysis of CLSA Data
Recipient
Location
Ontario, Canada
Recipient Type
Academia
Type
Grant
Source
CIHR Funding Decisions Database
As published
Id
410748
Name
Ma Jinhui
Region
Ontario
Country
Canada
Orgtype
Institution Paid; Research Institution
Instname2
Circulatory and Respiratory Health
Theme2
Social / Cultural / Environmental / Population Health
Prcname2
Catalyst Grant: Analysis of CLSA Data
Projecttitle
How to Handle Missing Spirometry Data in Population-Based Studies?
Programtype2
Operating Grants
Orgnameinp2
McMaster University
Cihramount2
$70,000
Cihrequipment2
$0
Programname2
Catalyst Grant: Analysis of CLSA Data
Keyworddelim
Missing Data; Population-Based Study; Spirometry
Abstract
Spirometry has increasingly been utilized in population-based studies including the Canadian Longitudinal Study on Aging (CLSA), to assess the presence of lung disease at the population level. According to the literature, about 10 to 30% of participants in population-based studies do not have valid spirometry data due to either contraindication from spirometry tests or poor quality spirometry. Of the 30,097 participants within the CLSA comprehensive cohort, 29.1% have no spirometry data available at baseline due to low quality spirometry or contraindication for spirometry and are considered as "missing" in the statistical analysis. The substantial amount of missing spirometry data may lead to efficiency reduction and serious bias, which undermine the validity of research findings.
Two measures are commonly collected from the spirometry for evaluation of lung function: forced expiratory volume in one second (FEV1) and forced vital capacity (FVC), which measure flow rate and lung volume in one maximal breath. The ratio of FEV1 to FVC together with FEV1 and FVC are used to identify obstructive and restrictive lung defects. The FEV1 and FVC are found to have a very strong, age-independent linear relationship in healthy individuals; however, this relationship does not hold among those with lung impairment. Traditional imputation techniques are inappropriate to handle the missing spirometry data since (1) FEV1 and FVC are paired measures from one spirometry test and they are observed or missing simultaneously; and (2) the relationship between FEV1 and FVC depends on whether an individual has lung impairment.
This investigation will help researchers appreciate the complexity of missing spirometry data and adopt appropriate imputation methods to handle them. Moreover, methods developed in this investigation can be generalized to handle the missingness of any paired data, such as systolic and diastolic blood pressure, a common phenomenon in health research.
Competitiondate
201909
Approvedterm2
1 yr 0 mth
Pinamesdelim
Ma, Jinhui
Approvedterm
1 an 0 mois
Cihrcontribution2
Inst of Population &Publ Hlth~ $70,000~ $0~