- Email: [email protected]
Predictive Value of Different Expressions of Forced Expiratory Volume in 1 Second (FEV1) for Adverse Outcomes in a Cohort of Adults Aged 80 and Older
JAMDA xxx (2016) 1.e1e1.e8
JAMDA journal homepage: www.jamda.com
Original Study
Predictive Value of Different Expressions of Forced Expiratory Volume in 1 Second (FEV1) for Adverse Outcomes in a Cohort of Adults Aged 80 and Older Eralda Hegendörfer MD a, b, *, Bert Vaes MD, PhD a, b, Elena Andreeva MD a, Catharina Matheï MD, PhD b, Gijs Van Pottelbergh MD, PhD b, Jean-Marie Degryse MD, PhD a, b a b
Institute of Health and Society, Université Catholique de Louvain, Brussels, Belgium Department of Public Health and Primary Care, Katholieke Universiteit Leuven, Leuven, Belgium
a b s t r a c t Keywords: FEV1 expressions adverse health outcomes adults aged 80 and older
Objectives: Forced expiratory volume in 1 second (FEV1) is proposed as a marker of healthy ageing and FEV1 expressions that are independent of reference values have been reported to be better at predicting mortality in older adults. We assess and compare the predictive value of different FEV1 expressions for mortality, hospitalization, and physical and mental decline in adults aged 80 and older. Design: Population-based, prospective, cohort study. Setting: The BELFRAIL study, Belgium. Participants: A total of 501 community-dwelling adults aged 80 and older (mean age 84.7 years). Measurements: Baseline FEV1 expressed as percent predicted (FEV1PP) and z-score (FEV1Z) using the Global Lung Function Initiative 2012 reference values; over lowest sex-specific percentile (FEV1Q), and height squared (FEV1/Ht2) and cubed (FEV1/Ht3). Mortality data until 5.1 0.2 years from baseline; hospitalization data until 3.0 0.25 years. Activities of daily living, battery of physical performance tests, Mini-Mental State Examination, and 15-item Geriatric Depression Scale at baseline and after 1.7 0.2 years. Results: Individuals in the lowest quartile of FEV1 expressions had higher adjusted risk than the rest of study population for all-cause mortality (highest hazard ratio 2.05 [95% Confidence Interval 1.50e2.80] for FEV1Q and 2.01 [1.47e2.76] for FEV1/Ht3), first hospitalization (highest hazard ratio 1.63 [1.21e2.16] for FEV1/Ht2 and 1.61[1.20e2.16] for FEV1/Ht3), mental decline (highest odds ratio 2.80 [1.61e4.89] for FEV1Q) and physical decline (only FEV1/Ht3 with odds ratio 1.93 [1.13e3.30]). Based on risk classification improvement measures, FEV1/Ht3 and FEV1Q performed better than FEV1PP. Conclusion: In a cohort of adults aged 80 and older, FEV1 expressions that are independent of reference values (FEV1/Ht3 and FEV1Q) were better at predicting adverse health outcomes than traditional expressions that depend on reference values, and should be used in further research on FEV1 and aging. Ó 2016 AMDA e The Society for Post-Acute and Long-Term Care Medicine.
Forced expiratory volume in 1 second (FEV1) has been proposed as a marker of overall health and healthy aging due to its associations with adverse health outcomes beyond respiratory morbidity and mortality, The authors have no conflicts of interest to declare. The BELFRAIL study was funded by an unconditional grant from Fondation Louvain, Brussels, Belgium (grant B40320084685). EH has received a scholarship from ERAWEB 2, Erasmus Mundus program at Katholieke Universiteit Leuven. * Address correspondence to Eralda Hegendörfer, MD, Institute of Health and Society, Université Catholique de Louvain, Clos Chapelle-aux-Champs 30, bte B1.30.15, 1200 Brussels, Belgium. E-mail address: [email protected] (E. Hegendörfer). http://dx.doi.org/10.1016/j.jamda.2016.08.012 1525-8610/Ó 2016 AMDA e The Society for Post-Acute and Long-Term Care Medicine.
such as all-cause mortality, cardiovascular mortality and morbidity, disability, and reduced physical and cognitive performance.1e11 As FEV1 varies with age, height, sex, and ethnicity, its measured value is reported in relation to a predicted value derived from reference spirometry data of healthy, never-smokers with equivalent age, sex, height, and ethnicity.12 It is commonly expressed as a percentage of measured over predicted values and referred to as FEV1 “percent predicted” (FEV1PP). Yet this expression of FEV1 does not account for the “normal” age-related decline in FEV1 and the variability of reference values that is even higher in older adults.13 This has led to the expression of measured FEV1 as z-scores (FEV1Z) that
1.e2
E. Hegendörfer et al. / JAMDA xxx (2016) 1.e1e1.e8
Fig. 1. Flowchart of the data collection in the BELFRAIL cohort. ADL, activities of daily living; GDS-15, 15-item geriatric depression scale; MMSE, mini mental state examination; PPT, battery of physical performance tests.
E. Hegendörfer et al. / JAMDA xxx (2016) 1.e1e1.e8
1.e3
Table 1 Baseline Characteristics and Outcomes of the Study Population in Total and by the Lowest Quartile of FEV1
Age, y Sex, male Height, m FEV1, L Smoker/Ex-smoker Asthma/COPD Nonrespiratory morbidities ADL PPT MMSE GDS-15 hsCRP, mg/dL NTproBNP, pg/mL eGFR, mL/min/1.73 m2 Death at 5 y First hospitalization Mental decline Physical decline
Total Population, n ¼ 501
FEV1 Lowest Quartile, n ¼ 126
FEV1 Rest of Quartiles, n ¼ 375
P
84.7 3.7 186 (37.1) 1.59 0.09 1.7 0.6 158 (31.6) 70 (14) 5 [3; 7] 25 [21; 27] 9 [5; 11] 28 [26; 29] 2 [1; 4] 0.18 [0.08; 0.40] 186.9 [93.8; 510.1] 62.8 [49.5; 77.8] 208 (41.5) 249 (50.4) 90 (23.8) 146 (38.5)
85.9 4.2 20 (15.9) 1.55 0.08 1.0 0.2 29 (23.2) 32 (25.6) 5 [3; 7] 22 [18; 26] 6 [4; 9] 27 [25; 28] 2.5 [2; 5] 0.20 [0.10; 0.54] 239.3 [109.2; 615.1] 58.8 [45.6; 78.0] 82 (65.1) 73 (67.8) 32 (36.8) 41 (47.7)
84.3 3.4 166 (44.3) 1.61 0.09 1.9 0.5 129 (34.4) 38 (10.1) 4 [3; 7] 26 [22; 29] 9 [6; 12] 28 [26; 29] 2 [1; 4] 0.16 [0.07; 0.36] 174.7 [90.0; 423.4] 63.4 [50.4; 77.5] 126 (33.6) 176 (47.1) 58 (19.9) 105 (35.8)
<.001* <.001y <.001* <.001* .02y <.001y .04z <.001z <.001z <.001z <.001z .02z .01z .4z <.001y .009y .002y .06y
ADL, activities of daily living; COPD, chronic obstructive pulmonary disease; eGFR, estimated glomerular filtration rate; FEV1, forced expiratory volume in one second (lowest quartile < 1.27 L); GDS-15, 15-item geriatric depression scale; hs-CRP, high sensitivity C-reactive protein; MMSE, mini-mental state examination; NT-proBNP, N-terminal probrain natriuretic peptide; PPT, battery of physical performance tests. Data are presented as n, mean SD, n (%), or median [interquartile range], unless otherwise stated. *P value based on the Student t test. y P value based on Pearson c2 test. z P value based on the Mann-Whitney U test.
take into consideration the variability of the reference values and the “normal” age-related decline.13,14 Although the z-score approach is statistically robust and has been reported as clinically valid,13,15 it relies on the availability of reliable age-specific reference values that have been relatively limited for adults aged 80 and older, until recently when the Global Lung Function Initiative all-ages reference equations have been made available for populations 3 to 95 years old.16 Yet, these reference equations need more data from those aged 80 and older and there is an ongoing discussion on the best way to derive reference values for interpretation of spirometry parameters in this age group.16,17 The use of FEV1 expressions that are independent of reference values, such as FEV1 divided by height cubed (FEV1/Ht3) or squared (FEV1/Ht2) and as a function of the sex-specific first percentile (FEV1 quotient [FEV1Q]), has been previously reported in studies on the association of FEV1 to all-cause mortality in older adults, leading to an increased interest on the use of these FEV1 expressions.18e22 Yet, studies on the predictive ability of these alternative FEV1 expressions not only for all-cause mortality, but other relevant adverse health outcomes and focusing on the adults aged 80 and older are still very limited.21,23,24 This study aims to assess and compare the predictive value of the different FEV1 expressions for all-cause mortality, first unplanned hospitalization, as well as declines in physical and mental function in a cohort of adults aged 80 and older. Method Study Design and Population The BELFRAIL study is a prospective, observational, populationbased cohort study of people aged 80 years and older living in Belgium, aiming to acquire a better understanding of the epidemiology and pathophysiology of chronic diseases in this age group. The study protocol and sampling methods have been already described in detail elsewhere.25 Briefly, between November 2008 and September 2009, in 29 general practice centers, 567 individuals aged 80 years and older were recruited, excluding only those with severe dementia, in palliative care or medical emergencies. At baseline, the general practitioners recorded sociodemographic data and medical history. A
clinical research assistant performed a standardized assessment at the participants’ home including performance tests, spirometry, and blood sample collection at baseline and after 1.7 0.2 years. Hospitalization data were collected until 3.0 0.25 years and mortality data until 5.1 0.2 years from baseline (Figure 1). The study protocol was approved by the Biomedical Ethics Committee of the Medical School of the Université Catholique de Louvain in Belgium. All participants gave informed consent. Baseline Spirometry and FEV1 Expressions Spirometry data were gathered by 2 trained clinical research assistants using a Spirobank spirometer (Medical International Research, Rome, Italy). Reversibility testing was not performed. Standardized measurement of height was performed during the clinical research assistant’s visit. Two independent researchers evaluated all spirograms based on the acceptability and repeatability criteria of the American Thoracic Society/European Respiratory Society.26 Only individuals with usable spirograms were included in this study.23,26 FEV1 was expressed as FEV1PP and FEV1Z based on the measured FEV1 and the sex-, age-, and height-specific mean predicted values (for FEV1PP) and z-scores (for FEV1Z) derived from the Global Lung Function Initiative 2012 reference equations.27 FEV1Q was calculated as a ratio of measured FEV1 over the sex-specific first percentile of a large heterogeneous population (0.5 L for men and 0.4 L for women).20 FEV1/Ht2 was calculated as the ratio of measured FEV1 over height squared, whereas FEV1/Ht3 as ratio over height cubed. Outcomes Time to all-cause mortality at 5 years and first unplanned hospitalization at 3-year follow-up were used as outcome measurements. Activities of daily living and a battery of physical performance tests were used as measures of physical functioning.25 Physical decline was defined as a relevant decline in any of these tests between baseline and follow-up assessment (see Appendix). The Mini-Mental State Examination and the 15-item Geriatric Depression Scale were used to assess the cognitive and mood/affective components of the mental
1.e4
E. Hegendörfer et al. / JAMDA xxx (2016) 1.e1e1.e8
Quartile 3
Quartile 3
Quartile 4
Quartile 4
Quartile 3
Q uartile 3
Quartile 4
Quartile 3
Quartile 4
Quartile 4
Fig. 2. Kaplan-Meier survival curves of 5-year all-cause mortality for quartiles of FEV1 expressions.
status, respectively.25 Mental decline was defined as a relevant decline in any of these tests between baseline and follow-up assessment (see Appendix). Other Variables In addition to age and sex, the following potential confounding variables were used in the statistical analysis: smoking status (never, previous, or current smoker), presence of respiratory diseases (asthma or chronic obstructive pulmonary disease), number of nonrespiratory morbidities (unweighted disease count; see Appendix), highsensitivity C-reactive protein, N-terminal pro-brain natriuretic peptide, and estimated glomerular filtration rate28e30 (estimated with the Modification of Diet in Renal Disease Study Group equation31). The general practitioners reported the recorded status of smoking and morbidities at baseline. Morning blood samples were collected and serum samples were stored frozen at 80 C until analysis.25
Statistical Analysis FEV1 expressions were ranked into quartiles of their distribution. Comparisons of baseline and outcome variables between the lowest quartile and the rest of the study population were tested with the independent Student t test for parametric variables, and MannWhitney U test for nonparametric and Pearson’s c2 test for categorical variables. Kaplan-Meier curves for all-cause mortality and hospitalization were plotted for the quartiles of each FEV1 expression, using the logrank test for comparison. Cox proportional hazards regression models were used to estimate the hazard ratio (HR) for mortality and first hospitalization of the lowest quartile of each FEV1 expression after adjusting for potential confounders. Models were checked for the proportional hazard assumption. The logistic regression model was used to estimate the odds ratios (OR) of the lowest quartile of FEV1 expressions for physical and mental decline. The rest of the study
E. Hegendörfer et al. / JAMDA xxx (2016) 1.e1e1.e8
Quartile 3
Quartile 4
Quartile 3
Proportion without hospitalization
Q uartile 3
1.e5
Q uartile 4
FEV1Ht3 1.00 0.75 0.50
Proportion without hospitalization
log-rank test p< 0.001
0.25 0.00 0
Number at risk Quartile 1 Quartile 2 Quartile 3 Quartile 4
Number at risk Quartile 1 Quartile 2 Quartile 3 Quartile 4
Quartile 4
Quartile 1 (lowest)
Quartile 2
Quartile 3
Quartile 4
1 2 Time to first hospitalization (years)
119 124 126 125
75 95 98 113
50 67 82 89
3 14 15 12 13
FEV1Q
0 121 124 124 125
Quartile 1 (lowest)
Quartile 2
Quartile 3
Quartile 4
1 2 Time to first hospitalization (years) 79 93 102 107
51 67 83 87
3 14 10 21 9
Fig. 3. Kaplan-Meier survival curves for first unplanned hospitalization at 3-year follow-up for quartiles of FEV1 expressions.
population was used as reference category. Variables were checked for multicollinearity. A 2-tailed probability value of P < .05 was considered statistically significant. Harrell’s concordance index, area under the receiver operating curve, continuous net reclassification improvement (NRI), and integrated discrimination improvement (IDI) were used to compare the predictive value of the FEV1 expressions using the FEV1PP as reference. The continuous NRI is the sum of proportion of correctly reclassified events (NRI events) and nonevents (NRI nonevents) considering all changes in predicted risk between 2 models for events and nonevents, without a defined risk categorization, whereas IDI is the difference in discrimination slopes between 2 models or this difference over the slope of the reference model (relative IDI).32 Statistical analysis was
performed with SPSS 23.0 (SPSS Inc., Chicago, IL), Stata 13.0 (StataCorp, College Station, TX), and SAS University Edition (SAS Institute, Inc, Cary, NC). Results Of the 567 participants of the BELFRAIL cohort, 522 (92.1%) performed spirometry at baseline; 501 of them had usable spirograms and were included in this study (Figure 1). They had a mean age of 84.7 years and consisted of 37% men. The excluded participants (n ¼ 66) did not have statistically significant differences compared with this study population for all baseline and outcome variables (except for a higher median baseline score of the 15-Item Geriatric Depression Scale). The
1.e6
E. Hegendörfer et al. / JAMDA xxx (2016) 1.e1e1.e8
Table 2 Multivariable Cox Regression Analysis and Discrimination Statistics of the FEV1 Expressions for All-Cause Mortality and Hospitalization
Mortality FEV1PP FEV1Z FEV1Q FEV1/Ht2 FEV1/Ht3 Hospitalization FEV1PP FEV1Z FEV1Q FEV1/Ht2 FEV1/Ht3
Nr
Crude HR (95% CI)
Adjusted HR (95% CI)
Harrell’s C (95% CI)
NRI (95% CI)
480 480 486 486 486
2.36 2.22 2.62 2.40 2.45
(1.77e3.12) (1.67e2.96) (1.98e3.45) (1.81e3.17) (1.85e3.25)
1.94 1.92 2.05 1.84 2.01
(1.38e2.74) (1.37e2.70) (1.50e2.80) (1.33e2.54) (1.47e2.76)
0.72 0.72 0.72 0.72 0.72
(0.68e0.75) (0.68e0.75) (0.69e0.76) (0.69e0.75) (0.69e0.76)
0.49 0.31 0.04 0.39
(0.32; 0.66) (0.12; 0.50) (0.22; 0.16) (0.21; 0.59)
474 474 480 480 480
1.66 1.64 1.74 1.82 1.78
(1.26e2.19) (1.25e2.16) (1.33e2.28) (1.39e2.38) (1.35e2.33)
1.51 1.49 1.46 1.63 1.61
(1.11e2.06) (1.10e2.03) (1.09e1.96) (1.21e2.19) (1.20e2.16)
0.67 0.67 0.67 0.68 0.68
(0.64e0.71) (0.64e0.71) (0.64e0.71) (0.65e0.71) (0.65e0.71)
0.03 0.13 0.32 0.29
(0.23; 0.17) (0.33; 0.10) (0.12; 0.50) (0.08; 0.49)
NRI Events, %
NRI Nonevents, %
IDI (95% CI)
22.6 17.9 11.6 21.9
26.1 12.6 15.2 17.2
0.004 0.011 0.001 0.012
(0.0002; 0.007) (0.002; 0.021) (0.004; 0.007) (0.004; 0.020)
4.6 2.9 18.1 3.5
2.0 9.7 49.8 32.2
0.001 0.001 0.005 0.005
(0.003; 0.001) (0.003; 0.005) (0.001; 0.009) (0.001; 0.010)
Relative IDI
0.025 0.073 0.008 0.078
0.005 0.006 0.035 0.036
FEV1PP, forced expiratory volume in one second percent of predicted; FEV1Z, FEV1 z-score; FEV1Q, FEV1 over the first sex-specific percentile; FEV1Ht2, FEV1 divided by height squared. FEV1Ht3, FEV1 divided by height cubed; Nr, number of participants with all data for the analysis; HR, hazard ratio; CI, confidence interval; Adjusted HR, adjusted for age, sex, smoking status, respiratory disease, number of other co-morbidities, hs-CRP >0.3 mg/dl, highest sex-specific tertile of NTproBNP, eGFR < 45 ml/min/1.73m2; Harrell’s C, Harrell’s concordance index; NRI, continuous net reclassification index; IDI, integrative discrimination index; Reference group, lowest quartile versus rest of study population (lowest quartile cut-offs: 65.29% FEV1PP; 1.78 FEV1Z; 3.05 FEV1Q; 0.34 L/m3 FEV1Ht3; 0.53 L/m2 FEV1Ht2).
main baseline demographic and clinical characteristics of the study population are shown in Table 1. The participants in the lowest quartile of the FEV1 as raw measurement were older and more often women (Table 1). Current/previous smoking was present in 31.6% of the population, with fewer smokers/ex-smokers in the lowest FEV1 quartile. The prevalence of respiratory diseases was 14%, with a higher prevalence in the lowest FEV1 quartile. Participants in the lowest FEV1 quartile had worse scores on all components of physical and mental functioning at baseline compared with the rest of the population (Table 1). Data on allcause mortality at 5-year follow-up were available for all the participants, whereas data on first unplanned hospitalization at 3-year followup were available for 494 participants. During 3.0 0.25 years of followup, 249 (50.4%) had at least 1 unplanned hospitalization reported and at 5.0 0.25 years, 208 (41.5%) had died (Table 1). Participants in the lowest quartile of all FEV1 expressions had significantly higher all-cause mortality at 5 years and first hospitalization at 3-year follow-up compared with the rest of the study population (Figures 2 and 3). Even after adjustment, participants in the lowest quartile of all FEV1 expressions had a higher risk of all-cause mortality and first hospitalization compared with the rest of the study population (Table 2). The highest adjusted HRs for mortality were 2.05 (95% confidence interval [CI] 1.50e2.80) for FEV1Q and 2.01 (1.47e2.76) for FEV1/Ht3, whereas for first hospitalization were 1.63 (1.21e2.19) for FEV1/Ht2 and 1.61 (1.20e2.16) for FEV1/Ht3 (Table 2). There were no significant differences
in Harrell’s concordance index between the FEV1 expressions for both mortality and hospitalization. Based on the NRI, FEV1/Ht3 improved the risk classification for 21.9% of dead and 17.2% of alive study participants compared with FEV1PP, whereas FEV1Q improved it for 17.9% of dead and 12.6% of alive participants (Table 2). For those without hospitalization, risk classification was improved for 49.8% by FEV1/Ht2 and 32.2% by FEV1/Ht3, whereas for those with hospitalization it worsened for 3.5% by FEV1/Ht3 and 18% by FEV1/Ht2 (Table 2). Overall, FEV1/Ht3 had the highest relative IDI for both mortality (0.078) and first hospitalization (0.036), increasing the difference of the mean predicted risk of events (death/hospitalization) and nonevents (alive/no hospitalization) with nearly 8% for mortality and 4% for first hospitalization compared with FEV1PP (Table 2). Complete data regarding physical and mental decline at 1.7 0.2 years of follow-up were available for 378 participants. Physical decline was identified in 146 (38.5%) participants, whereas mental decline in 90 (23.8%) (Table 1). Only FEV1/Ht3 was independently associated with physical decline with adjusted OR 1.93 (1.13e3.30) (Table 3). Participants in the lowest quartile of all FEV1 expressions, except FEV1PP, had a statistically significant increased risk for mental decline with the highest adjusted OR of 2.80 (1.61e4.89) for FEV1Q (Table 3). Based on the event and nonevent NRIs, FEV1Q improved risk reclassification for 49.7% of those without mental decline and worsened it for 15.6% of those with decline. It also had the highest relative IDI (0.348), increasing with 34.8% the difference of mean predicted
Table 3 Multivariable Logistic Regression Analysis Discrimination Statistics of the FEV1 Expressions for Physical and Mental Decline
Mental decline FEV1PP FEV1Z FEV1Q FEV1/Ht2 FEV1/Ht3 Physical decline FEV1PP FEV1Z FEV1Q FEV1/Ht2 FEV1/Ht3
Nr
Crude OR (95% CI)
Adjusted OR (95% CI)
AUC (95% CI)
NRI (95% CI)
NRI Events, %
NRI Nonevents, %
IDI (95% CI)
376 376 377 377 377
1.92 1.92 3.03 2.45 2.44
(1.13e3.25) (1.13e3.25) (1.80e5.08) (1.45e4.13) (1.44e4.15)
1.77 1.83 2.80 2.27 2.29
(0.98e3.18) (1.02e3.30) (1.61e4.89) (1.29e4.00) (1.30e4.06)
0.64 0.64 0.66 0.65 0.65
(0.57e0.71) (0.57e0.71) (0.59e0.73) (0.58e0.72) (0.58e0.72)
0.34 0.34 0.14 0.20
(0.11; 0.58) (0.11; 0.57) (0.09; 0.38) (0.04; 0.43)
0 15.6 17.8 2.2
34.3 49.7 32.2 21.7
0.002 0.023 0.007 0.008
377 377 378 378 378
1.65 1.55 1.59 1.73 1.89
(1.01e2.68) (0.95e2.52) (0.98e2.58) (1.06e2.82) (1.16e3.09)
1.68 1.60 1.54 1.71 1.93
(0.98e2.87) (0.94e2.74) (0.91e2.59) (1.00e2.90) (1.13e3.30)
0.63 0.63 0.63 0.64 0.64
(0.58e0.69) (0.57e0.68) (0.58e0.69) (0.58e0.69) (0.58e0.70)
0.03 0.02 0.21 0.30
(0.17; 0.23) (0.19; 0.23) (0.01; 0.40) (0.10; 0.50)
29.7 0.7 40.7 10.4
26.7 2.6 19.8 40.5
0.002 0.001 0.002 0.008
(0.001; 0.005) (0.010; 0.036) (0.0002; 0.015) (0.001; 0.017)
(0.004; 0.001) (0.007; 0.004) (0.002; 0.007) (0.001; 0.015)
Relative IDI
0.027 0.348 0.112 0.120
0.037 0.027 0.051 0.177
FEV1PP, forced expiratory volume in one second percent of predicted; FEV1Z, FEV1 z-score; FEV1Q, FEV1 over the first sex-specific percentile; FEV1Ht2, FEV1 divided by height squared; FEV1Ht3, FEV1 divided by height cubed; Nr, number of participants with all data for the analysis; OR, odds ratio; CI, confidence interval; Adjusted OR, adjusted for age, sex, smoking status, respiratory disease and number of other co-morbidities; AUC, area under the receiver operating curve; NRI, continuous net reclassification index; IDI, integrative discrimination index; Reference group, lowest quartile versus rest of study population (lowest quartile cut-offs: 65.29% FEV1PP; 1.78 FEV1Z; 3.05 FEV1Q; 0.34 L/ m3 FEV1Ht3; 0.53 L/m2 FEV1Ht2).
E. Hegendörfer et al. / JAMDA xxx (2016) 1.e1e1.e8
probabilities of those with and without mental decline, compared with FEV1PP (Table 3). Discussion In a large, representative population-based prospective cohort of adults aged 80 and older, we found low FEV1 to be an independent predictor of all-cause mortality, unplanned hospitalization, decline in mental and physical functioning, and its expressions that are independent of reference values, such as FEV1/Ht3 or FEV1/Q, were better at predicting these adverse outcomes than the traditional expression as FEV1PP. FEV1/Ht3 or FEV1/Ht2 and FEV1Q have already been reported to perform better than FEV1PP at predicting all-cause mortality in clinical setting studies that included older adults.18,20,21 Our study confirmed these findings in a cohort of community-dwelling adults aged 80 and older who had not been the primary focus of previous studies. Additionally, we examined the association of FEV1 expressions with other relevant adverse outcomes for this age group, such as hospitalization, and physical and mental decline. In line with other studies, FEV1Q and FEV1/Ht3 were the best predictors for 5-year allcause mortality in our cohort of community-dwelling very old adults, independently of a wider variety of potential confounders that had not been adjusted for in previous studies, including respiratory and nonrespiratory comorbidities, and high levels of C-reactive protein and N-terminal pro-brain natriuretic peptide. N-terminal probrain natriuretic peptide is valuable in ruling out heart failure, a common comorbidity that can influence spirometry parameters,30 and as a predictor of mortality in adults aged 80 and older.28 Although both FEV1/Ht2 and FEV1/Ht3 had better predictive performance for first hospitalization, only FEV1/Ht3 had a significant risk reclassification improvement for both time to all-cause mortality and first unplanned hospitalization. Lower FEV1 expressed as raw measurement, FEV1PP, or height adjusted has already been found to be associated with lower scores of different measures of physical and cognitive functioning across different adult age groups.6,9,10 We investigated for the first time the predictive value of different FEV1 expressions for mental and physical decline over 2 years of follow-up in adults aged 80 and older. FEV1Q had the best performance in predicting mental decline compared with FEV1PP, followed closely by FEV1/Ht3 and FEV1/Ht2. Only the lowest quartile of FEV1/Ht3 had a statistically significant higher adjusted risk for physical decline. Our findings confirm and extend the current status of research that suggests FEV1 as a predictor of important adverse outcomes and an indicator of overall health in geriatric assessments.2,23 Over all 4 adverse health outcomes in our cohort of adults aged 80 and older, FEV1/Ht3 seems to have the best predictive performance followed closely by FEV1Q (only for all-cause mortality and mental decline). These FEV1 expressions are easier to use, as they do not depend on reference values, while still allowing adjustment for body size and sex.20,21 Although FEV1Z derived with the lambda-mu-sigma method accounts for age-related changes in pulmonary function and has been recently reported to be better than FEV1PP and FEV1/Ht3 for association over time with cardiopulmonary predictors in a cohort of people older than 40 years,13,15,24 in our study of adults aged 80 and older it did not show better predictive performance for adverse health outcomes. Considering the difficulties of generating reliable spirometry reference values for the worldwide growing group of very old adults,25 future research on the role of FEV1 as a predictor of adverse outcomes in older adults should consider the use of and further exploration of the predictive value of these alternative expressions. This study has several strengths, including a large heterogeneous population representative of the adults aged 80 and older in Belgium who had a standardized comprehensive geriatric
1.e7
assessment at baseline and at approximately 2 years’ follow-up, as well as data on 5-year mortality and time to first unplanned hospitalization at 3 years.25 The exclusion criteria of the BELFRAIL study (dementia or severe cognitive impairment and being in palliative or emergency care) are one limitation of this study. The physical and mental decline definitions were not based on a validated method, so decline for some components may have been missed or overestimated. We also used the measured standing height that might be an issue in the older adults, as height reduction is frequent due to age- and disease-related osteoporotic vertebral changes, introducing bias with possible overestimation of predictive values.21 Although this effect has been found to be small for predicted FEV1, height calculated as a function of arm span and age could be better and needs to be explored in future studies.33 In conclusion, in a representative sample of adults aged 80 and older, FEV1 expressions that are independent of reference values, such as FEV1/Ht3 and FEV1Q, performed better at predicting adverse health outcomes. These findings support the use of FEV1/Ht3 and FEV1Q as alternative expressions of FEV1 and further exploration of its role as a risk marker for adverse health outcomes in older adults than traditional expressions that depend on reference values. Acknowledgments The authors thank the general practitioners in Belgium who recruited and followed their patients in the BELFRAIL cohort. The research protocol of BELFRAIL was approved by the Biomedical Ethics Committee of the Medical School of the Université Catholique de Louvain in Belgium and written consent was obtained from all participants. Supplementary Data Supplementary data related to this article can be found online at http://dx.doi.org/10.1016/j.jamda.2016.08.012. References 1. Lange P. Spirometric findings as predictors of survival. Thorax 2011;66:1e2. 2. Lara J, Cooper R, Nissan J, et al. A proposed panel of biomarkers of healthy ageing. BMC Med 2015;13:222. 3. Anstey KJ, Lord SR, Smith GA. Measuring human functional age: A review of empirical findings. Exp Aging Res 1996;22:245e266. 4. Hole DJ, Watt GC, Davey-Smith G, et al. Impaired lung function and mortality risk in men and women: Findings from the Renfrew and Paisley prospective population study. BMJ 1996;313:711e715. 5. Schunemann HJ, Dorn J, Grant BJ, et al. Pulmonary function is a long-term predictor of mortality in the general population: 29-year follow-up of the Buffalo Health Study. Chest 2000;118:656e664. 6. Simpson CF, Punjabi NM, Wolfenden L, et al. Relationship between lung function and physical performance in disabled older women. J Gerontol A Biol Sci Med Sci 2005;60:350e354. 7. Sin DD, Wu L, Man SF. The relationship between reduced lung function and cardiovascular mortality: A population-based study and a systematic review of the literature. Chest 2005;127:1952e1959. 8. Jacobsen PK, Sigsgaard T, Pedersen OF, et al. Lung function as a predictor of survival in very elderly people: The Danish 1905 cohort study. J Am Geriatr Soc 2008;56:2150e2152. 9. Singh-Manoux A, Dugravot A, Kauffmann F, et al. Association of lung function with physical, mental and cognitive function in early old age. Age (Dordr) 2011;33:385e392. 10. Vidal JS, Aspelund T, Jonsdottir MK, et al. Pulmonary function impairment may be an early risk factor for late-life cognitive impairment. J Am Geriatr Soc 2013; 61:79e83. 11. Lee HM, Liu MA, Barrett-Connor E, Wong ND. Association of lung function with coronary heart disease and cardiovascular disease outcomes in elderly: The Rancho Bernardo study. Respir Med 2014;108:1779e1785. 12. Pellegrino R, Viegi G, Brusasco V, et al. Interpretative strategies for lung function tests. Eur Respir J 2005;26:948e968. 13. Stanojevic S, Wade A, Stocks J, et al. Reference ranges for spirometry across all ages: A new approach. Am J Respir Crit Care Med 2008;177:253e260. 14. Miller MR, Pincock AC. Predicted values: How should we use them? Thorax 1988;43:265e267.
1.e8
E. Hegendörfer et al. / JAMDA xxx (2016) 1.e1e1.e8
15. Vaz Fragoso CA, Gill TM, McAvay G, et al. Respiratory impairment in older persons: When less means more. Am J Med 2013;126:49e57. 16. Quanjer PH, Stanojevic S, Cole TJ, et al. Multi-ethnic reference values for spirometry for the 3e95-yr age range: The global lung function 2012 equations. Eur Respir J 2012;40:1324e1343. 17. Miller MR, Thinggaard M, Christensen K, et al. Best lung function equations for the very elderly selected by survival analysis. Eur Respir J 2014;43:1338e1346. 18. Chinn S, Gislason T, Aspelund T, Gudnason V. Optimum expression of adult lung function based on all-cause mortality: Results from the Reykjavik study. Respir Med 2007;101:601e609. 19. Miller MR, Pedersen OF, Lange P, Vestbo J. Improved survival prediction from lung function data in a large population sample. Respir Med 2009;103:442e448. 20. Miller MR, Pedersen OF. New concepts for expressing forced expiratory volume in 1 s arising from survival analysis. Eur Respir J 2010;35:873e882. 21. Pedone C, Scarlata S, Scichilone N, et al. Alternative ways of expressing FEV1 and mortality in elderly people with and without COPD. Eur Respir J 2013;41:800e805. 22. Sorlie PD, Kannel WB, O’Connor G. Mortality associated with respiratory function and symptoms in advanced age. The Framingham Study. Am Rev Respir Dis 1989;140:379e384. 23. Turkeshi E, Vaes B, Andreeva E, et al. Short-term prognostic value of forced expiratory volume in 1 second divided by height cubed in a prospective cohort of people 80 years and older. BMC Geriatr 2015;15:15. 24. Vaz Fragoso CA, McAvay G, Van Ness PH, et al. Aging-related considerations when evaluating the forced expiratory volume in 1 second (FEV1) over time. J Gerontol A Biol Sci Med Sci 2016;71:929e934.
25. Vaes B, Pasquet A, Wallemacq P, et al. The BELFRAIL (BFC80þ) study: A population-based prospective cohort study of the very elderly in Belgium. BMC Geriatr 2010;10:39. 26. Miller MR, Hankinson J, Brusasco V, et al. Standardisation of spirometry. Eur Respir J 2005;26:319e338. 27. Stanojevic S. GLI-2012 Excel Sheet Calculator. Available at: http://www. ers-education.org/guidelines/global-lung-function-initiative/tools/excel-sheetcalculator.aspx; 2013. Accessed June 18, 2015. 28. Vaes B, de Ruijter W, Degryse J, et al. Clinical relevance of a raised plasma NTproBNP level in a population-based cohort of nonagenarians. J Am Geriatr Soc 2009;57:823e829. 29. Sabia S, Shipley M, Elbaz A, et al. Why does lung function predict mortality? Results from the Whitehall II Cohort Study. Am J Epidemiol 2010;172: 1415e1423. 30. Guder G, Rutten FH, Brenner S, et al. The impact of heart failure on the classification of COPD severity. J Card Fail 2012;18:637e644. 31. Levey AS, Bosch JP, Lewis JB, et al. A more accurate method to estimate glomerular filtration rate from serum creatinine: A new prediction equation. Modification of Diet in Renal Disease Study Group. Ann Intern Med 1999;130:461e470. 32. Pencina MJ, D’Agostino RB Sr, Demler OV. Novel metrics for evaluating improvement in discrimination: Net reclassification and integrated discrimination improvement for normal variables and nested models. Stat Med 2012; 31:101e113. 33. Quanjer PH, Capderou A, Mazicioglu MM, et al. All-age relationship between arm span and height in different ethnic groups. Eur Respir J 2014;44:905e912.
JAMDA journal homepage: www.jamda.com
Original Study
Predictive Value of Different Expressions of Forced Expiratory Volume in 1 Second (FEV1) for Adverse Outcomes in a Cohort of Adults Aged 80 and Older Eralda Hegendörfer MD a, b, *, Bert Vaes MD, PhD a, b, Elena Andreeva MD a, Catharina Matheï MD, PhD b, Gijs Van Pottelbergh MD, PhD b, Jean-Marie Degryse MD, PhD a, b a b
Institute of Health and Society, Université Catholique de Louvain, Brussels, Belgium Department of Public Health and Primary Care, Katholieke Universiteit Leuven, Leuven, Belgium
a b s t r a c t Keywords: FEV1 expressions adverse health outcomes adults aged 80 and older
Objectives: Forced expiratory volume in 1 second (FEV1) is proposed as a marker of healthy ageing and FEV1 expressions that are independent of reference values have been reported to be better at predicting mortality in older adults. We assess and compare the predictive value of different FEV1 expressions for mortality, hospitalization, and physical and mental decline in adults aged 80 and older. Design: Population-based, prospective, cohort study. Setting: The BELFRAIL study, Belgium. Participants: A total of 501 community-dwelling adults aged 80 and older (mean age 84.7 years). Measurements: Baseline FEV1 expressed as percent predicted (FEV1PP) and z-score (FEV1Z) using the Global Lung Function Initiative 2012 reference values; over lowest sex-specific percentile (FEV1Q), and height squared (FEV1/Ht2) and cubed (FEV1/Ht3). Mortality data until 5.1 0.2 years from baseline; hospitalization data until 3.0 0.25 years. Activities of daily living, battery of physical performance tests, Mini-Mental State Examination, and 15-item Geriatric Depression Scale at baseline and after 1.7 0.2 years. Results: Individuals in the lowest quartile of FEV1 expressions had higher adjusted risk than the rest of study population for all-cause mortality (highest hazard ratio 2.05 [95% Confidence Interval 1.50e2.80] for FEV1Q and 2.01 [1.47e2.76] for FEV1/Ht3), first hospitalization (highest hazard ratio 1.63 [1.21e2.16] for FEV1/Ht2 and 1.61[1.20e2.16] for FEV1/Ht3), mental decline (highest odds ratio 2.80 [1.61e4.89] for FEV1Q) and physical decline (only FEV1/Ht3 with odds ratio 1.93 [1.13e3.30]). Based on risk classification improvement measures, FEV1/Ht3 and FEV1Q performed better than FEV1PP. Conclusion: In a cohort of adults aged 80 and older, FEV1 expressions that are independent of reference values (FEV1/Ht3 and FEV1Q) were better at predicting adverse health outcomes than traditional expressions that depend on reference values, and should be used in further research on FEV1 and aging. Ó 2016 AMDA e The Society for Post-Acute and Long-Term Care Medicine.
Forced expiratory volume in 1 second (FEV1) has been proposed as a marker of overall health and healthy aging due to its associations with adverse health outcomes beyond respiratory morbidity and mortality, The authors have no conflicts of interest to declare. The BELFRAIL study was funded by an unconditional grant from Fondation Louvain, Brussels, Belgium (grant B40320084685). EH has received a scholarship from ERAWEB 2, Erasmus Mundus program at Katholieke Universiteit Leuven. * Address correspondence to Eralda Hegendörfer, MD, Institute of Health and Society, Université Catholique de Louvain, Clos Chapelle-aux-Champs 30, bte B1.30.15, 1200 Brussels, Belgium. E-mail address: [email protected] (E. Hegendörfer). http://dx.doi.org/10.1016/j.jamda.2016.08.012 1525-8610/Ó 2016 AMDA e The Society for Post-Acute and Long-Term Care Medicine.
such as all-cause mortality, cardiovascular mortality and morbidity, disability, and reduced physical and cognitive performance.1e11 As FEV1 varies with age, height, sex, and ethnicity, its measured value is reported in relation to a predicted value derived from reference spirometry data of healthy, never-smokers with equivalent age, sex, height, and ethnicity.12 It is commonly expressed as a percentage of measured over predicted values and referred to as FEV1 “percent predicted” (FEV1PP). Yet this expression of FEV1 does not account for the “normal” age-related decline in FEV1 and the variability of reference values that is even higher in older adults.13 This has led to the expression of measured FEV1 as z-scores (FEV1Z) that
1.e2
E. Hegendörfer et al. / JAMDA xxx (2016) 1.e1e1.e8
Fig. 1. Flowchart of the data collection in the BELFRAIL cohort. ADL, activities of daily living; GDS-15, 15-item geriatric depression scale; MMSE, mini mental state examination; PPT, battery of physical performance tests.
E. Hegendörfer et al. / JAMDA xxx (2016) 1.e1e1.e8
1.e3
Table 1 Baseline Characteristics and Outcomes of the Study Population in Total and by the Lowest Quartile of FEV1
Age, y Sex, male Height, m FEV1, L Smoker/Ex-smoker Asthma/COPD Nonrespiratory morbidities ADL PPT MMSE GDS-15 hsCRP, mg/dL NTproBNP, pg/mL eGFR, mL/min/1.73 m2 Death at 5 y First hospitalization Mental decline Physical decline
Total Population, n ¼ 501
FEV1 Lowest Quartile, n ¼ 126
FEV1 Rest of Quartiles, n ¼ 375
P
84.7 3.7 186 (37.1) 1.59 0.09 1.7 0.6 158 (31.6) 70 (14) 5 [3; 7] 25 [21; 27] 9 [5; 11] 28 [26; 29] 2 [1; 4] 0.18 [0.08; 0.40] 186.9 [93.8; 510.1] 62.8 [49.5; 77.8] 208 (41.5) 249 (50.4) 90 (23.8) 146 (38.5)
85.9 4.2 20 (15.9) 1.55 0.08 1.0 0.2 29 (23.2) 32 (25.6) 5 [3; 7] 22 [18; 26] 6 [4; 9] 27 [25; 28] 2.5 [2; 5] 0.20 [0.10; 0.54] 239.3 [109.2; 615.1] 58.8 [45.6; 78.0] 82 (65.1) 73 (67.8) 32 (36.8) 41 (47.7)
84.3 3.4 166 (44.3) 1.61 0.09 1.9 0.5 129 (34.4) 38 (10.1) 4 [3; 7] 26 [22; 29] 9 [6; 12] 28 [26; 29] 2 [1; 4] 0.16 [0.07; 0.36] 174.7 [90.0; 423.4] 63.4 [50.4; 77.5] 126 (33.6) 176 (47.1) 58 (19.9) 105 (35.8)
<.001* <.001y <.001* <.001* .02y <.001y .04z <.001z <.001z <.001z <.001z .02z .01z .4z <.001y .009y .002y .06y
ADL, activities of daily living; COPD, chronic obstructive pulmonary disease; eGFR, estimated glomerular filtration rate; FEV1, forced expiratory volume in one second (lowest quartile < 1.27 L); GDS-15, 15-item geriatric depression scale; hs-CRP, high sensitivity C-reactive protein; MMSE, mini-mental state examination; NT-proBNP, N-terminal probrain natriuretic peptide; PPT, battery of physical performance tests. Data are presented as n, mean SD, n (%), or median [interquartile range], unless otherwise stated. *P value based on the Student t test. y P value based on Pearson c2 test. z P value based on the Mann-Whitney U test.
take into consideration the variability of the reference values and the “normal” age-related decline.13,14 Although the z-score approach is statistically robust and has been reported as clinically valid,13,15 it relies on the availability of reliable age-specific reference values that have been relatively limited for adults aged 80 and older, until recently when the Global Lung Function Initiative all-ages reference equations have been made available for populations 3 to 95 years old.16 Yet, these reference equations need more data from those aged 80 and older and there is an ongoing discussion on the best way to derive reference values for interpretation of spirometry parameters in this age group.16,17 The use of FEV1 expressions that are independent of reference values, such as FEV1 divided by height cubed (FEV1/Ht3) or squared (FEV1/Ht2) and as a function of the sex-specific first percentile (FEV1 quotient [FEV1Q]), has been previously reported in studies on the association of FEV1 to all-cause mortality in older adults, leading to an increased interest on the use of these FEV1 expressions.18e22 Yet, studies on the predictive ability of these alternative FEV1 expressions not only for all-cause mortality, but other relevant adverse health outcomes and focusing on the adults aged 80 and older are still very limited.21,23,24 This study aims to assess and compare the predictive value of the different FEV1 expressions for all-cause mortality, first unplanned hospitalization, as well as declines in physical and mental function in a cohort of adults aged 80 and older. Method Study Design and Population The BELFRAIL study is a prospective, observational, populationbased cohort study of people aged 80 years and older living in Belgium, aiming to acquire a better understanding of the epidemiology and pathophysiology of chronic diseases in this age group. The study protocol and sampling methods have been already described in detail elsewhere.25 Briefly, between November 2008 and September 2009, in 29 general practice centers, 567 individuals aged 80 years and older were recruited, excluding only those with severe dementia, in palliative care or medical emergencies. At baseline, the general practitioners recorded sociodemographic data and medical history. A
clinical research assistant performed a standardized assessment at the participants’ home including performance tests, spirometry, and blood sample collection at baseline and after 1.7 0.2 years. Hospitalization data were collected until 3.0 0.25 years and mortality data until 5.1 0.2 years from baseline (Figure 1). The study protocol was approved by the Biomedical Ethics Committee of the Medical School of the Université Catholique de Louvain in Belgium. All participants gave informed consent. Baseline Spirometry and FEV1 Expressions Spirometry data were gathered by 2 trained clinical research assistants using a Spirobank spirometer (Medical International Research, Rome, Italy). Reversibility testing was not performed. Standardized measurement of height was performed during the clinical research assistant’s visit. Two independent researchers evaluated all spirograms based on the acceptability and repeatability criteria of the American Thoracic Society/European Respiratory Society.26 Only individuals with usable spirograms were included in this study.23,26 FEV1 was expressed as FEV1PP and FEV1Z based on the measured FEV1 and the sex-, age-, and height-specific mean predicted values (for FEV1PP) and z-scores (for FEV1Z) derived from the Global Lung Function Initiative 2012 reference equations.27 FEV1Q was calculated as a ratio of measured FEV1 over the sex-specific first percentile of a large heterogeneous population (0.5 L for men and 0.4 L for women).20 FEV1/Ht2 was calculated as the ratio of measured FEV1 over height squared, whereas FEV1/Ht3 as ratio over height cubed. Outcomes Time to all-cause mortality at 5 years and first unplanned hospitalization at 3-year follow-up were used as outcome measurements. Activities of daily living and a battery of physical performance tests were used as measures of physical functioning.25 Physical decline was defined as a relevant decline in any of these tests between baseline and follow-up assessment (see Appendix). The Mini-Mental State Examination and the 15-item Geriatric Depression Scale were used to assess the cognitive and mood/affective components of the mental
1.e4
E. Hegendörfer et al. / JAMDA xxx (2016) 1.e1e1.e8
Quartile 3
Quartile 3
Quartile 4
Quartile 4
Quartile 3
Q uartile 3
Quartile 4
Quartile 3
Quartile 4
Quartile 4
Fig. 2. Kaplan-Meier survival curves of 5-year all-cause mortality for quartiles of FEV1 expressions.
status, respectively.25 Mental decline was defined as a relevant decline in any of these tests between baseline and follow-up assessment (see Appendix). Other Variables In addition to age and sex, the following potential confounding variables were used in the statistical analysis: smoking status (never, previous, or current smoker), presence of respiratory diseases (asthma or chronic obstructive pulmonary disease), number of nonrespiratory morbidities (unweighted disease count; see Appendix), highsensitivity C-reactive protein, N-terminal pro-brain natriuretic peptide, and estimated glomerular filtration rate28e30 (estimated with the Modification of Diet in Renal Disease Study Group equation31). The general practitioners reported the recorded status of smoking and morbidities at baseline. Morning blood samples were collected and serum samples were stored frozen at 80 C until analysis.25
Statistical Analysis FEV1 expressions were ranked into quartiles of their distribution. Comparisons of baseline and outcome variables between the lowest quartile and the rest of the study population were tested with the independent Student t test for parametric variables, and MannWhitney U test for nonparametric and Pearson’s c2 test for categorical variables. Kaplan-Meier curves for all-cause mortality and hospitalization were plotted for the quartiles of each FEV1 expression, using the logrank test for comparison. Cox proportional hazards regression models were used to estimate the hazard ratio (HR) for mortality and first hospitalization of the lowest quartile of each FEV1 expression after adjusting for potential confounders. Models were checked for the proportional hazard assumption. The logistic regression model was used to estimate the odds ratios (OR) of the lowest quartile of FEV1 expressions for physical and mental decline. The rest of the study
E. Hegendörfer et al. / JAMDA xxx (2016) 1.e1e1.e8
Quartile 3
Quartile 4
Quartile 3
Proportion without hospitalization
Q uartile 3
1.e5
Q uartile 4
FEV1Ht3 1.00 0.75 0.50
Proportion without hospitalization
log-rank test p< 0.001
0.25 0.00 0
Number at risk Quartile 1 Quartile 2 Quartile 3 Quartile 4
Number at risk Quartile 1 Quartile 2 Quartile 3 Quartile 4
Quartile 4
Quartile 1 (lowest)
Quartile 2
Quartile 3
Quartile 4
1 2 Time to first hospitalization (years)
119 124 126 125
75 95 98 113
50 67 82 89
3 14 15 12 13
FEV1Q
0 121 124 124 125
Quartile 1 (lowest)
Quartile 2
Quartile 3
Quartile 4
1 2 Time to first hospitalization (years) 79 93 102 107
51 67 83 87
3 14 10 21 9
Fig. 3. Kaplan-Meier survival curves for first unplanned hospitalization at 3-year follow-up for quartiles of FEV1 expressions.
population was used as reference category. Variables were checked for multicollinearity. A 2-tailed probability value of P < .05 was considered statistically significant. Harrell’s concordance index, area under the receiver operating curve, continuous net reclassification improvement (NRI), and integrated discrimination improvement (IDI) were used to compare the predictive value of the FEV1 expressions using the FEV1PP as reference. The continuous NRI is the sum of proportion of correctly reclassified events (NRI events) and nonevents (NRI nonevents) considering all changes in predicted risk between 2 models for events and nonevents, without a defined risk categorization, whereas IDI is the difference in discrimination slopes between 2 models or this difference over the slope of the reference model (relative IDI).32 Statistical analysis was
performed with SPSS 23.0 (SPSS Inc., Chicago, IL), Stata 13.0 (StataCorp, College Station, TX), and SAS University Edition (SAS Institute, Inc, Cary, NC). Results Of the 567 participants of the BELFRAIL cohort, 522 (92.1%) performed spirometry at baseline; 501 of them had usable spirograms and were included in this study (Figure 1). They had a mean age of 84.7 years and consisted of 37% men. The excluded participants (n ¼ 66) did not have statistically significant differences compared with this study population for all baseline and outcome variables (except for a higher median baseline score of the 15-Item Geriatric Depression Scale). The
1.e6
E. Hegendörfer et al. / JAMDA xxx (2016) 1.e1e1.e8
Table 2 Multivariable Cox Regression Analysis and Discrimination Statistics of the FEV1 Expressions for All-Cause Mortality and Hospitalization
Mortality FEV1PP FEV1Z FEV1Q FEV1/Ht2 FEV1/Ht3 Hospitalization FEV1PP FEV1Z FEV1Q FEV1/Ht2 FEV1/Ht3
Nr
Crude HR (95% CI)
Adjusted HR (95% CI)
Harrell’s C (95% CI)
NRI (95% CI)
480 480 486 486 486
2.36 2.22 2.62 2.40 2.45
(1.77e3.12) (1.67e2.96) (1.98e3.45) (1.81e3.17) (1.85e3.25)
1.94 1.92 2.05 1.84 2.01
(1.38e2.74) (1.37e2.70) (1.50e2.80) (1.33e2.54) (1.47e2.76)
0.72 0.72 0.72 0.72 0.72
(0.68e0.75) (0.68e0.75) (0.69e0.76) (0.69e0.75) (0.69e0.76)
0.49 0.31 0.04 0.39
(0.32; 0.66) (0.12; 0.50) (0.22; 0.16) (0.21; 0.59)
474 474 480 480 480
1.66 1.64 1.74 1.82 1.78
(1.26e2.19) (1.25e2.16) (1.33e2.28) (1.39e2.38) (1.35e2.33)
1.51 1.49 1.46 1.63 1.61
(1.11e2.06) (1.10e2.03) (1.09e1.96) (1.21e2.19) (1.20e2.16)
0.67 0.67 0.67 0.68 0.68
(0.64e0.71) (0.64e0.71) (0.64e0.71) (0.65e0.71) (0.65e0.71)
0.03 0.13 0.32 0.29
(0.23; 0.17) (0.33; 0.10) (0.12; 0.50) (0.08; 0.49)
NRI Events, %
NRI Nonevents, %
IDI (95% CI)
22.6 17.9 11.6 21.9
26.1 12.6 15.2 17.2
0.004 0.011 0.001 0.012
(0.0002; 0.007) (0.002; 0.021) (0.004; 0.007) (0.004; 0.020)
4.6 2.9 18.1 3.5
2.0 9.7 49.8 32.2
0.001 0.001 0.005 0.005
(0.003; 0.001) (0.003; 0.005) (0.001; 0.009) (0.001; 0.010)
Relative IDI
0.025 0.073 0.008 0.078
0.005 0.006 0.035 0.036
FEV1PP, forced expiratory volume in one second percent of predicted; FEV1Z, FEV1 z-score; FEV1Q, FEV1 over the first sex-specific percentile; FEV1Ht2, FEV1 divided by height squared. FEV1Ht3, FEV1 divided by height cubed; Nr, number of participants with all data for the analysis; HR, hazard ratio; CI, confidence interval; Adjusted HR, adjusted for age, sex, smoking status, respiratory disease, number of other co-morbidities, hs-CRP >0.3 mg/dl, highest sex-specific tertile of NTproBNP, eGFR < 45 ml/min/1.73m2; Harrell’s C, Harrell’s concordance index; NRI, continuous net reclassification index; IDI, integrative discrimination index; Reference group, lowest quartile versus rest of study population (lowest quartile cut-offs: 65.29% FEV1PP; 1.78 FEV1Z; 3.05 FEV1Q; 0.34 L/m3 FEV1Ht3; 0.53 L/m2 FEV1Ht2).
main baseline demographic and clinical characteristics of the study population are shown in Table 1. The participants in the lowest quartile of the FEV1 as raw measurement were older and more often women (Table 1). Current/previous smoking was present in 31.6% of the population, with fewer smokers/ex-smokers in the lowest FEV1 quartile. The prevalence of respiratory diseases was 14%, with a higher prevalence in the lowest FEV1 quartile. Participants in the lowest FEV1 quartile had worse scores on all components of physical and mental functioning at baseline compared with the rest of the population (Table 1). Data on allcause mortality at 5-year follow-up were available for all the participants, whereas data on first unplanned hospitalization at 3-year followup were available for 494 participants. During 3.0 0.25 years of followup, 249 (50.4%) had at least 1 unplanned hospitalization reported and at 5.0 0.25 years, 208 (41.5%) had died (Table 1). Participants in the lowest quartile of all FEV1 expressions had significantly higher all-cause mortality at 5 years and first hospitalization at 3-year follow-up compared with the rest of the study population (Figures 2 and 3). Even after adjustment, participants in the lowest quartile of all FEV1 expressions had a higher risk of all-cause mortality and first hospitalization compared with the rest of the study population (Table 2). The highest adjusted HRs for mortality were 2.05 (95% confidence interval [CI] 1.50e2.80) for FEV1Q and 2.01 (1.47e2.76) for FEV1/Ht3, whereas for first hospitalization were 1.63 (1.21e2.19) for FEV1/Ht2 and 1.61 (1.20e2.16) for FEV1/Ht3 (Table 2). There were no significant differences
in Harrell’s concordance index between the FEV1 expressions for both mortality and hospitalization. Based on the NRI, FEV1/Ht3 improved the risk classification for 21.9% of dead and 17.2% of alive study participants compared with FEV1PP, whereas FEV1Q improved it for 17.9% of dead and 12.6% of alive participants (Table 2). For those without hospitalization, risk classification was improved for 49.8% by FEV1/Ht2 and 32.2% by FEV1/Ht3, whereas for those with hospitalization it worsened for 3.5% by FEV1/Ht3 and 18% by FEV1/Ht2 (Table 2). Overall, FEV1/Ht3 had the highest relative IDI for both mortality (0.078) and first hospitalization (0.036), increasing the difference of the mean predicted risk of events (death/hospitalization) and nonevents (alive/no hospitalization) with nearly 8% for mortality and 4% for first hospitalization compared with FEV1PP (Table 2). Complete data regarding physical and mental decline at 1.7 0.2 years of follow-up were available for 378 participants. Physical decline was identified in 146 (38.5%) participants, whereas mental decline in 90 (23.8%) (Table 1). Only FEV1/Ht3 was independently associated with physical decline with adjusted OR 1.93 (1.13e3.30) (Table 3). Participants in the lowest quartile of all FEV1 expressions, except FEV1PP, had a statistically significant increased risk for mental decline with the highest adjusted OR of 2.80 (1.61e4.89) for FEV1Q (Table 3). Based on the event and nonevent NRIs, FEV1Q improved risk reclassification for 49.7% of those without mental decline and worsened it for 15.6% of those with decline. It also had the highest relative IDI (0.348), increasing with 34.8% the difference of mean predicted
Table 3 Multivariable Logistic Regression Analysis Discrimination Statistics of the FEV1 Expressions for Physical and Mental Decline
Mental decline FEV1PP FEV1Z FEV1Q FEV1/Ht2 FEV1/Ht3 Physical decline FEV1PP FEV1Z FEV1Q FEV1/Ht2 FEV1/Ht3
Nr
Crude OR (95% CI)
Adjusted OR (95% CI)
AUC (95% CI)
NRI (95% CI)
NRI Events, %
NRI Nonevents, %
IDI (95% CI)
376 376 377 377 377
1.92 1.92 3.03 2.45 2.44
(1.13e3.25) (1.13e3.25) (1.80e5.08) (1.45e4.13) (1.44e4.15)
1.77 1.83 2.80 2.27 2.29
(0.98e3.18) (1.02e3.30) (1.61e4.89) (1.29e4.00) (1.30e4.06)
0.64 0.64 0.66 0.65 0.65
(0.57e0.71) (0.57e0.71) (0.59e0.73) (0.58e0.72) (0.58e0.72)
0.34 0.34 0.14 0.20
(0.11; 0.58) (0.11; 0.57) (0.09; 0.38) (0.04; 0.43)
0 15.6 17.8 2.2
34.3 49.7 32.2 21.7
0.002 0.023 0.007 0.008
377 377 378 378 378
1.65 1.55 1.59 1.73 1.89
(1.01e2.68) (0.95e2.52) (0.98e2.58) (1.06e2.82) (1.16e3.09)
1.68 1.60 1.54 1.71 1.93
(0.98e2.87) (0.94e2.74) (0.91e2.59) (1.00e2.90) (1.13e3.30)
0.63 0.63 0.63 0.64 0.64
(0.58e0.69) (0.57e0.68) (0.58e0.69) (0.58e0.69) (0.58e0.70)
0.03 0.02 0.21 0.30
(0.17; 0.23) (0.19; 0.23) (0.01; 0.40) (0.10; 0.50)
29.7 0.7 40.7 10.4
26.7 2.6 19.8 40.5
0.002 0.001 0.002 0.008
(0.001; 0.005) (0.010; 0.036) (0.0002; 0.015) (0.001; 0.017)
(0.004; 0.001) (0.007; 0.004) (0.002; 0.007) (0.001; 0.015)
Relative IDI
0.027 0.348 0.112 0.120
0.037 0.027 0.051 0.177
FEV1PP, forced expiratory volume in one second percent of predicted; FEV1Z, FEV1 z-score; FEV1Q, FEV1 over the first sex-specific percentile; FEV1Ht2, FEV1 divided by height squared; FEV1Ht3, FEV1 divided by height cubed; Nr, number of participants with all data for the analysis; OR, odds ratio; CI, confidence interval; Adjusted OR, adjusted for age, sex, smoking status, respiratory disease and number of other co-morbidities; AUC, area under the receiver operating curve; NRI, continuous net reclassification index; IDI, integrative discrimination index; Reference group, lowest quartile versus rest of study population (lowest quartile cut-offs: 65.29% FEV1PP; 1.78 FEV1Z; 3.05 FEV1Q; 0.34 L/ m3 FEV1Ht3; 0.53 L/m2 FEV1Ht2).
E. Hegendörfer et al. / JAMDA xxx (2016) 1.e1e1.e8
probabilities of those with and without mental decline, compared with FEV1PP (Table 3). Discussion In a large, representative population-based prospective cohort of adults aged 80 and older, we found low FEV1 to be an independent predictor of all-cause mortality, unplanned hospitalization, decline in mental and physical functioning, and its expressions that are independent of reference values, such as FEV1/Ht3 or FEV1/Q, were better at predicting these adverse outcomes than the traditional expression as FEV1PP. FEV1/Ht3 or FEV1/Ht2 and FEV1Q have already been reported to perform better than FEV1PP at predicting all-cause mortality in clinical setting studies that included older adults.18,20,21 Our study confirmed these findings in a cohort of community-dwelling adults aged 80 and older who had not been the primary focus of previous studies. Additionally, we examined the association of FEV1 expressions with other relevant adverse outcomes for this age group, such as hospitalization, and physical and mental decline. In line with other studies, FEV1Q and FEV1/Ht3 were the best predictors for 5-year allcause mortality in our cohort of community-dwelling very old adults, independently of a wider variety of potential confounders that had not been adjusted for in previous studies, including respiratory and nonrespiratory comorbidities, and high levels of C-reactive protein and N-terminal pro-brain natriuretic peptide. N-terminal probrain natriuretic peptide is valuable in ruling out heart failure, a common comorbidity that can influence spirometry parameters,30 and as a predictor of mortality in adults aged 80 and older.28 Although both FEV1/Ht2 and FEV1/Ht3 had better predictive performance for first hospitalization, only FEV1/Ht3 had a significant risk reclassification improvement for both time to all-cause mortality and first unplanned hospitalization. Lower FEV1 expressed as raw measurement, FEV1PP, or height adjusted has already been found to be associated with lower scores of different measures of physical and cognitive functioning across different adult age groups.6,9,10 We investigated for the first time the predictive value of different FEV1 expressions for mental and physical decline over 2 years of follow-up in adults aged 80 and older. FEV1Q had the best performance in predicting mental decline compared with FEV1PP, followed closely by FEV1/Ht3 and FEV1/Ht2. Only the lowest quartile of FEV1/Ht3 had a statistically significant higher adjusted risk for physical decline. Our findings confirm and extend the current status of research that suggests FEV1 as a predictor of important adverse outcomes and an indicator of overall health in geriatric assessments.2,23 Over all 4 adverse health outcomes in our cohort of adults aged 80 and older, FEV1/Ht3 seems to have the best predictive performance followed closely by FEV1Q (only for all-cause mortality and mental decline). These FEV1 expressions are easier to use, as they do not depend on reference values, while still allowing adjustment for body size and sex.20,21 Although FEV1Z derived with the lambda-mu-sigma method accounts for age-related changes in pulmonary function and has been recently reported to be better than FEV1PP and FEV1/Ht3 for association over time with cardiopulmonary predictors in a cohort of people older than 40 years,13,15,24 in our study of adults aged 80 and older it did not show better predictive performance for adverse health outcomes. Considering the difficulties of generating reliable spirometry reference values for the worldwide growing group of very old adults,25 future research on the role of FEV1 as a predictor of adverse outcomes in older adults should consider the use of and further exploration of the predictive value of these alternative expressions. This study has several strengths, including a large heterogeneous population representative of the adults aged 80 and older in Belgium who had a standardized comprehensive geriatric
1.e7
assessment at baseline and at approximately 2 years’ follow-up, as well as data on 5-year mortality and time to first unplanned hospitalization at 3 years.25 The exclusion criteria of the BELFRAIL study (dementia or severe cognitive impairment and being in palliative or emergency care) are one limitation of this study. The physical and mental decline definitions were not based on a validated method, so decline for some components may have been missed or overestimated. We also used the measured standing height that might be an issue in the older adults, as height reduction is frequent due to age- and disease-related osteoporotic vertebral changes, introducing bias with possible overestimation of predictive values.21 Although this effect has been found to be small for predicted FEV1, height calculated as a function of arm span and age could be better and needs to be explored in future studies.33 In conclusion, in a representative sample of adults aged 80 and older, FEV1 expressions that are independent of reference values, such as FEV1/Ht3 and FEV1Q, performed better at predicting adverse health outcomes. These findings support the use of FEV1/Ht3 and FEV1Q as alternative expressions of FEV1 and further exploration of its role as a risk marker for adverse health outcomes in older adults than traditional expressions that depend on reference values. Acknowledgments The authors thank the general practitioners in Belgium who recruited and followed their patients in the BELFRAIL cohort. The research protocol of BELFRAIL was approved by the Biomedical Ethics Committee of the Medical School of the Université Catholique de Louvain in Belgium and written consent was obtained from all participants. Supplementary Data Supplementary data related to this article can be found online at http://dx.doi.org/10.1016/j.jamda.2016.08.012. References 1. Lange P. Spirometric findings as predictors of survival. Thorax 2011;66:1e2. 2. Lara J, Cooper R, Nissan J, et al. A proposed panel of biomarkers of healthy ageing. BMC Med 2015;13:222. 3. Anstey KJ, Lord SR, Smith GA. Measuring human functional age: A review of empirical findings. Exp Aging Res 1996;22:245e266. 4. Hole DJ, Watt GC, Davey-Smith G, et al. Impaired lung function and mortality risk in men and women: Findings from the Renfrew and Paisley prospective population study. BMJ 1996;313:711e715. 5. Schunemann HJ, Dorn J, Grant BJ, et al. Pulmonary function is a long-term predictor of mortality in the general population: 29-year follow-up of the Buffalo Health Study. Chest 2000;118:656e664. 6. Simpson CF, Punjabi NM, Wolfenden L, et al. Relationship between lung function and physical performance in disabled older women. J Gerontol A Biol Sci Med Sci 2005;60:350e354. 7. Sin DD, Wu L, Man SF. The relationship between reduced lung function and cardiovascular mortality: A population-based study and a systematic review of the literature. Chest 2005;127:1952e1959. 8. Jacobsen PK, Sigsgaard T, Pedersen OF, et al. Lung function as a predictor of survival in very elderly people: The Danish 1905 cohort study. J Am Geriatr Soc 2008;56:2150e2152. 9. Singh-Manoux A, Dugravot A, Kauffmann F, et al. Association of lung function with physical, mental and cognitive function in early old age. Age (Dordr) 2011;33:385e392. 10. Vidal JS, Aspelund T, Jonsdottir MK, et al. Pulmonary function impairment may be an early risk factor for late-life cognitive impairment. J Am Geriatr Soc 2013; 61:79e83. 11. Lee HM, Liu MA, Barrett-Connor E, Wong ND. Association of lung function with coronary heart disease and cardiovascular disease outcomes in elderly: The Rancho Bernardo study. Respir Med 2014;108:1779e1785. 12. Pellegrino R, Viegi G, Brusasco V, et al. Interpretative strategies for lung function tests. Eur Respir J 2005;26:948e968. 13. Stanojevic S, Wade A, Stocks J, et al. Reference ranges for spirometry across all ages: A new approach. Am J Respir Crit Care Med 2008;177:253e260. 14. Miller MR, Pincock AC. Predicted values: How should we use them? Thorax 1988;43:265e267.
1.e8
E. Hegendörfer et al. / JAMDA xxx (2016) 1.e1e1.e8
15. Vaz Fragoso CA, Gill TM, McAvay G, et al. Respiratory impairment in older persons: When less means more. Am J Med 2013;126:49e57. 16. Quanjer PH, Stanojevic S, Cole TJ, et al. Multi-ethnic reference values for spirometry for the 3e95-yr age range: The global lung function 2012 equations. Eur Respir J 2012;40:1324e1343. 17. Miller MR, Thinggaard M, Christensen K, et al. Best lung function equations for the very elderly selected by survival analysis. Eur Respir J 2014;43:1338e1346. 18. Chinn S, Gislason T, Aspelund T, Gudnason V. Optimum expression of adult lung function based on all-cause mortality: Results from the Reykjavik study. Respir Med 2007;101:601e609. 19. Miller MR, Pedersen OF, Lange P, Vestbo J. Improved survival prediction from lung function data in a large population sample. Respir Med 2009;103:442e448. 20. Miller MR, Pedersen OF. New concepts for expressing forced expiratory volume in 1 s arising from survival analysis. Eur Respir J 2010;35:873e882. 21. Pedone C, Scarlata S, Scichilone N, et al. Alternative ways of expressing FEV1 and mortality in elderly people with and without COPD. Eur Respir J 2013;41:800e805. 22. Sorlie PD, Kannel WB, O’Connor G. Mortality associated with respiratory function and symptoms in advanced age. The Framingham Study. Am Rev Respir Dis 1989;140:379e384. 23. Turkeshi E, Vaes B, Andreeva E, et al. Short-term prognostic value of forced expiratory volume in 1 second divided by height cubed in a prospective cohort of people 80 years and older. BMC Geriatr 2015;15:15. 24. Vaz Fragoso CA, McAvay G, Van Ness PH, et al. Aging-related considerations when evaluating the forced expiratory volume in 1 second (FEV1) over time. J Gerontol A Biol Sci Med Sci 2016;71:929e934.
25. Vaes B, Pasquet A, Wallemacq P, et al. The BELFRAIL (BFC80þ) study: A population-based prospective cohort study of the very elderly in Belgium. BMC Geriatr 2010;10:39. 26. Miller MR, Hankinson J, Brusasco V, et al. Standardisation of spirometry. Eur Respir J 2005;26:319e338. 27. Stanojevic S. GLI-2012 Excel Sheet Calculator. Available at: http://www. ers-education.org/guidelines/global-lung-function-initiative/tools/excel-sheetcalculator.aspx; 2013. Accessed June 18, 2015. 28. Vaes B, de Ruijter W, Degryse J, et al. Clinical relevance of a raised plasma NTproBNP level in a population-based cohort of nonagenarians. J Am Geriatr Soc 2009;57:823e829. 29. Sabia S, Shipley M, Elbaz A, et al. Why does lung function predict mortality? Results from the Whitehall II Cohort Study. Am J Epidemiol 2010;172: 1415e1423. 30. Guder G, Rutten FH, Brenner S, et al. The impact of heart failure on the classification of COPD severity. J Card Fail 2012;18:637e644. 31. Levey AS, Bosch JP, Lewis JB, et al. A more accurate method to estimate glomerular filtration rate from serum creatinine: A new prediction equation. Modification of Diet in Renal Disease Study Group. Ann Intern Med 1999;130:461e470. 32. Pencina MJ, D’Agostino RB Sr, Demler OV. Novel metrics for evaluating improvement in discrimination: Net reclassification and integrated discrimination improvement for normal variables and nested models. Stat Med 2012; 31:101e113. 33. Quanjer PH, Capderou A, Mazicioglu MM, et al. All-age relationship between arm span and height in different ethnic groups. Eur Respir J 2014;44:905e912.