Expert Review of Respiratory Medicine
ISSN: 1747-6348 (Print) 1747-6356 (Online) Journal homepage: http://www.tandfonline.com/loi/ierx20
Changes in lung function in older people from the English Longitudinal Study of Ageing Abebaw M Yohannes & Gindo Tampubolon To cite this article: Abebaw M Yohannes & Gindo Tampubolon (2014) Changes in lung function in older people from the English Longitudinal Study of Ageing, Expert Review of Respiratory Medicine, 8:4, 515-521 To link to this article: http://dx.doi.org/10.1586/17476348.2014.919226
Published online: 16 May 2014.
Submit your article to this journal
Article views: 25
View related articles
View Crossmark data
Full Terms & Conditions of access and use can be found at http://www.tandfonline.com/action/journalInformation?journalCode=ierx20 Download by: [UNSW Library]
Date: 14 September 2015, At: 18:05
Original Research
Changes in lung function in older people from the English Longitudinal Study of Ageing Expert Review of Respiratory Medicine 2014.8:515-521.
Expert Rev. Respir. Med. 8(4), 515–521 (2014)
Abebaw M Yohannes*1 and Gindo Tampubolon2 1 Department of Health Professions, Manchester Metropolitan University, Elizabeth Gaskell Campus, Hathersage Road, Manchester, M13 0JA, UK 2 The University of Manchester, The Brooks World Poverty Institute, Manchester, UK *Author for correspondence: Tel.: +44 161 247 2943 Fax: +44 161 247 6571 [email protected]
Background: Decline in lung function with increasing age is common in older people. However, the rate of decline using the forced expiratory volume in 1 s (FEV1) in a longitudinal study in the elderly community dwellers is unknown. Methods: We analyzed data from the English Longitudinal Study of Ageing on people 50 years and older who had FEV1 measurement at Wave 2 and Wave 4 of 4 years of follow-up, respectively. A random coefficient model was employed to examine the changes in FEV1 and predict differences in the levels of FEV1 in older people. Results: A total of 4224 participants were included in the study. The mean (± standard deviation [SD]) rate of change in FEV1 was a decline of 32.92 ± 0.96 ml/year. The absolute difference in mean FEV1 level between female and male participants was 767.07 ± 16.6 ml. It was 253.91 ± 22.7 ml lower in current smokers than in nonsmokers, 73.67 ± 18.67 ml lower in participants with a history of sputum in winter months than in participants without sputum, 63.32 ± 7.07 ml lower in participants with a higher dyspnea score than in participants with a lower dyspnea score and 67.77 ± 15.87 ml higher in participants with good health compared to participants with fair/poor health status. One microgram increase in C-reactive protein level lowered FEV1 by 4.66 ± 0.86 ml and one Hb of hemoglobin increased the FEV1 level by 4.78 ± 0.77 ml. All were statistically significant at p < 0.001. Conclusions: The average rate of FEV1 decline in older people without respiratory diseases was found to be high. Lower level of FEV1 was also observed in current smokers, females, those with a history of sputum in winter months and in participants with higher dyspnea score or in those with poorer health status. KEYWORDS: COPD • ELSA • FEV1 • longitudinal study • lung function • older people
Decline in lung function with increasing age is common in older people. Untreated or underrecognized progressive decline in lung function in older people has major detrimental effects on the physical functioning and quality of life, and increases the risk of developing chronic obstructive pulmonary disease (COPD) and may contribute to premature mortality [1–3]. Previous longitudinal studies [4–6] focused on patients with established diagnosis of COPD and their findings indicate that the change in forced expiratory volume in 1 s (FEV1) is highly variable. They reported that increased decline in FEV1 in COPD patients was associated with current smoking [4], with increased risk of hospital readmission [7]. However, very little data are available from longitudinal informahealthcare.com
10.1586/17476348.2014.919226
studies on the decline in lung function in older people living in communities, without a confirmed diagnosis of respiratory disease (respiratory impairment) using FEV1. They are sparse and lacking robust data or predominantly on men [8]. Therefore, longitudinal studies are beneficial to healthcare providers and policy makers as a public health agenda for smoking-related diseases and to devise appropriate preventative and treatment strategies. They may also provide some insight into the impact of undiagnosed lung function impairment in older people who reside in the community. In addition, very little data are available from longitudinal studies that examined gender differences and subsequent changes in lung function [9,10]. These study findings are inconsistent [9,10], partly due to the
2014 Informa UK Ltd
ISSN 1747-6348
515
Original Research
Yohannes & Tampubolon
Expert Review of Respiratory Medicine 2014.8:515-521.
design of the methodologies and use of different statistical tests that are difficult to compare. Undiagnosed chronic cough and respiratory symptoms tend to contribute to higher level of lung function decline in women compared to men [11,12]. Thus, we took the opportunity to examine potential gender differences in the levels of FEV1 using the English Longitudinal Study of Ageing (ELSA) from the population-based data. We hypothesized that women would have substantially higher lung function decline compared to men in this longitudinal study. In order to test this hypothesis, we used data from the ELSA, a prospective observational study with 4 years of followup, to explore changes in lung function using FEV1 in community-dwelling subjects aged 50 years and older. We also examined the factors predicting differences in the level of FEV1 in older people. Methods Study population
The data were extracted from Wave 2 and Wave 4 of ELSA on individuals who underwent lung function tests on both occasions. The same individuals are re-interviewed and followed up every 2 years. Details of the study methodology have been published elsewhere [13,14]. Briefly, the sample was drawn from the Health Survey of England conducted in 1998, 1999 and 2001 for those born before March 1952. Detailed data are available on the ELSA participants at the website [15]. The sample drawn was representative of the people living in private households in England. A total of 8688 respondents participated in Wave 2 (2004–2005, response rate 82%); 7114 participated in Wave 3 (2006–2007, response rate 73%); and 6623 participated in Wave 4 (2008–2009, response rate 60%). The response rate is defined as [9] ‘total individual respondents to a given wave divided by total individuals eligible for that wave’. The inclusion of the participant in either numerator or denominator was not based on the response in any previous wave. Procedures
In all the waves, data were collected by face-to-face interviews and a self-completion questionnaire. At Wave 2 and Wave 4, a trained nurse visited the subjects at their home to take the lung function measurements. Our analysis is based on the ELSA participants who underwent lung function testing at the baseline assessment and at 4 years and for whom 4-year data are available (n = 4226 for both Wave 2 and Wave 4). All participants gave written informed consent and the study was approved by the relevant ethics committee. Lung function
The research nurse carried out two lung function tests (FEV1 and forced vital capacity [FVC]) at baseline in 2004 and during follow-up in 2008. Subjects underwent lung function test using a hand-held Vitalograph spirometer on both occasions. Three consecutive readings were taken, and as per the convention for the epidemiological study [8], the maximum of the three lung function maneuvers was recorded. 516
Serum and plasma samples were collected for measurement of biomarkers at baseline (Wave 2) and stored at -80˚C until the analysis was carried out. Association between the changes in FEV1 and the circulating levels of C-reactive protein (CRP) and fibrinogen was also examined. Socio-demographic characteristics were obtained at baseline. Educational status was categorized into three groups: higher (including university degree, other higher post or postsecondary education and ‘A level’ education or equivalent), secondary (certificate of secondary education or equivalent) and lower (including all people with lower than secondary education or no educational qualifications). Smoking status was classified as current smoker and nonsmoker. General health status was rated as poor, fair and good using self-reported measurement . The amount of alcohol consumed per day or per week over the last 12 months was recorded. Data on the presence of comorbid condition(s) for cardiovascular diseases (CVDs) at baseline (myocardial infarction, stroke, heart failure, angina or transient ischemic attacks) were extracted using the dichotomous variables ‘Yes’ or ‘No’. The severity of breathlessness was measured using the Medical Research Council dyspnea scale [9]. The Medical Research Council scale is categorized into five grades: Grade 1, not troubled by breathlessness except during strenuous exercise; Grade 2, short of breath when hurrying or walking up a slight hill; Grade 3, walks slower than contemporaries on level ground because of breathlessness, or has to stop for breath when walking at own pace; Grade 4, stops for breath after walking about 100 m or after a few minutes on level ground; and Grade 5, too breathless to leave the house or breathless when dressing or undressing [9]. Statistical analysis
Continuous variables are presented as means + standard deviation, and differences between the groups were analyzed using t-test and analysis of variance. Dichotomized variables are presented in percentages. Differences in categorical variables were assessed using the Chi-square test. Random coefficients model was fitted to the data to derive the rate of change in lung function (FEV1) and to estimate the association between lung function and individual characteristics. In particular, random slopes of time (age in years) were included, and fixed coefficient (not random coefficient) of square of age was also included to allow for nonlinear change in lung function. The virtue of this model stems from its ability to estimate each individual rate of change rather than one average rate of change for all individuals. This model is a special case of linear mixed model where random coefficients of variables other than time are allowed. This particular model is often fit to explain development over time of repeatedly observed individuals. The model has this form:
Yit = b 0 + b1 age it + b 2 X it + g i age i + e i + e it , where b0 is the usual constant term and b1 is the rate of change with g i being the random slopes or the rate of change Expert Rev. Respir. Med. 8(4), (2014)
Expert Review of Respiratory Medicine 2014.8:515-521.
Lung function decline in older people
Results
A total of 4224 participants were included in this study. Of these, 25% were aged 50–59 years, 40% were 60–69 years old, 27% were 70–79 years old and 8% were aged 80 years and above. Women comprised 55% of the participants. Of the 4224 participants, 3661 (86.6%) were nonsmokers or exsmokers and 561 (13.3) were current smokers. The baseline characteristics and changes in FEV1 during follow-up are reported in TABLE 1. The mean FVC at baseline was 3.90 l and 4 years later, it was 3.80 1, respectively, for the entire group. The mean percentage of FEV1/FVC at 4 years was 71%. Rate of change & difference in FEV1
shows the effects of predictors on the levels of FEV1 and changes in FEV1 during the 4-year follow-up period. The mean rate of change in FEV1 was -32 ± 0.95 ml/year (95% CI: -34 to -31). The absolute mean difference in FEV1 was -767 ml (95% CI: -799 to -734) for women in comparison to men (p < 0.001). Those with intermediate education had a difference in FEV1 by -119 ml (95% CI: -153 to -85; p < 0.001), but there was no difference in those having a degree compared with those with no education (p = 0.82). Selfreported good health had a positive association with preservation of lung function: those reporting good health had a larger FEV1 by 67 ml (95% CI: 36–98). Current smoking had a deleterious effect on lung function; compared to nonsmokers (including ex-smokers), current smokers had lower FEV1 of 253 ml (95% CI: -298 to -209). Daily or frequent consumption of alcohol had a positive association with better lung function wherein the FEV1 was 30 ml (95% CI: 2–58; p < 0.001). Presence of comorbid CVDs had a negative association with lung function which was lowered by -37 ml (95% CI: -64 to -10; p < 0.001). Breathlessness was also associated with lower level of lung function by -63 ml (95% CI: -77 to -49; p < 0.001). TABLE 2
informahealthcare.com
0.15
0.1 Density
for each individual i. The sum of these two terms gives, for each individual, a distinct individual rate of change; all these rates of change form a distribution of rates of change of lung function. b2 is the effect of age square and contains all (vector of) the effects of individual characteristics such as sex, education, occupational class, self-rated health, smoking, alcohol consumption, breathlessness, sputum, CVD indicators and biomarkers. ei is the random intercept, one for each individual i. eit is the usual residual random error term. The model assumes all the random coefficients to be normally distributed, but allows the random intercepts to be correlated with the random slopes. The virtue of this model is shown by focusing on g i, the individual rate of change; the sum of this individual rate with the average rate is summarized and depicted in the kernel density estimate. The empirical Bayes estimate of the rate of change for each individual in FEV1 was calculated and the resulting distribution is presented in a smoothed density function or a kernel density estimate in FIGURE 1. Analyses were performed using Stata version 12 (StataCorp LP, College station, USA, 2012). Significance was set at p < 0.05.
Original Research
0.05
0 -50
-40 -30 Rate of change, ml/year
-20
Figure 1. Distribution of estimated annual rates of change in forced expiratory volume in 1 s over the 4-year period from the English Longitudinal Study of Ageing on subjects 50 years and older (n = 4224).
Likewise, sputum production during winter months was associated with lower level of lung function by -73 ml (95% CI: -110 to -37; p < 0.001). Of the biomarkers, higher CRP had an association with lower level of FEV1 by -4.6 ml (95% CI: -6 to -3). Higher hemoglobin level had a positive association with better lung function in FEV1 by 4.6 ml (95% CI: 3–6). There was no statistically significant difference in the rate of change with the stratification of age. For example, in subjects of age 60 years and above, the rate and 95% CI were 32.28 ml/year and -34.63 to -29.93, respectively, in comparison to those of age 70 years and above, for whom the rate and 95% CI were -30.56 ml/year and -35.19 to -25.93, respectively (t = 1.8, p = 0. 09). Comparing the baseline level of FEV1 in those who completed the follow-up assessment versus who did not complete (2409 ml vs. 2162 ml) (t = -12.1, p < 0.001). Participants with low lung function may have been lost to follow-up. FIGURE 1 shows the distribution of the rate of change for each individual in the sample, and his or her empirical Bayes estimate. More than 97.5% of the distribution had a decline in FEV1 in excess of 32 ml/year. Discussion
This prospective observational study demonstrates that the rate of decline in FEV1 over a 4-year period was very high (32 ml/ year) in older people, and that the difference in lung function was associated with current smoking, sputum production in winter months, higher level of dyspnea, female gender, presence of CVD and higher level of CRP. In this study, the rate of decline in FEV1 was far higher in unselected elderly community dwellers than that previously reported in large cohort studies of 3 and 4 years of follow-up with moderate-to-severe COPD, in which the rate of decline in FEV1 was 41.0 ml/year [5], but was not significantly different from zero. Factors that contribute to the lower decline in these carefully selected sample studies are unclear; it is probably related to the 517
Original Research
Yohannes & Tampubolon
Table 1. Demographic characteristics of participants at baseline and 4 years of follow-up, and measurements of Forced expiratory volume in 1 s. Variable
Number (%)
Baseline FEV1 mean in ml
4-years FEV1 mean in ml
Age 50–59 60–69 70–79 80+
1486 (35) 1546 (37) 976 (23) 216 (5)
2732.46 2405.66 2066.21 1755.41 F = 205.21
2701.13 2460.42 2101.86 1711.01 F = 225.40
ISSN: 1747-6348 (Print) 1747-6356 (Online) Journal homepage: http://www.tandfonline.com/loi/ierx20
Changes in lung function in older people from the English Longitudinal Study of Ageing Abebaw M Yohannes & Gindo Tampubolon To cite this article: Abebaw M Yohannes & Gindo Tampubolon (2014) Changes in lung function in older people from the English Longitudinal Study of Ageing, Expert Review of Respiratory Medicine, 8:4, 515-521 To link to this article: http://dx.doi.org/10.1586/17476348.2014.919226
Published online: 16 May 2014.
Submit your article to this journal
Article views: 25
View related articles
View Crossmark data
Full Terms & Conditions of access and use can be found at http://www.tandfonline.com/action/journalInformation?journalCode=ierx20 Download by: [UNSW Library]
Date: 14 September 2015, At: 18:05
Original Research
Changes in lung function in older people from the English Longitudinal Study of Ageing Expert Review of Respiratory Medicine 2014.8:515-521.
Expert Rev. Respir. Med. 8(4), 515–521 (2014)
Abebaw M Yohannes*1 and Gindo Tampubolon2 1 Department of Health Professions, Manchester Metropolitan University, Elizabeth Gaskell Campus, Hathersage Road, Manchester, M13 0JA, UK 2 The University of Manchester, The Brooks World Poverty Institute, Manchester, UK *Author for correspondence: Tel.: +44 161 247 2943 Fax: +44 161 247 6571 [email protected]
Background: Decline in lung function with increasing age is common in older people. However, the rate of decline using the forced expiratory volume in 1 s (FEV1) in a longitudinal study in the elderly community dwellers is unknown. Methods: We analyzed data from the English Longitudinal Study of Ageing on people 50 years and older who had FEV1 measurement at Wave 2 and Wave 4 of 4 years of follow-up, respectively. A random coefficient model was employed to examine the changes in FEV1 and predict differences in the levels of FEV1 in older people. Results: A total of 4224 participants were included in the study. The mean (± standard deviation [SD]) rate of change in FEV1 was a decline of 32.92 ± 0.96 ml/year. The absolute difference in mean FEV1 level between female and male participants was 767.07 ± 16.6 ml. It was 253.91 ± 22.7 ml lower in current smokers than in nonsmokers, 73.67 ± 18.67 ml lower in participants with a history of sputum in winter months than in participants without sputum, 63.32 ± 7.07 ml lower in participants with a higher dyspnea score than in participants with a lower dyspnea score and 67.77 ± 15.87 ml higher in participants with good health compared to participants with fair/poor health status. One microgram increase in C-reactive protein level lowered FEV1 by 4.66 ± 0.86 ml and one Hb of hemoglobin increased the FEV1 level by 4.78 ± 0.77 ml. All were statistically significant at p < 0.001. Conclusions: The average rate of FEV1 decline in older people without respiratory diseases was found to be high. Lower level of FEV1 was also observed in current smokers, females, those with a history of sputum in winter months and in participants with higher dyspnea score or in those with poorer health status. KEYWORDS: COPD • ELSA • FEV1 • longitudinal study • lung function • older people
Decline in lung function with increasing age is common in older people. Untreated or underrecognized progressive decline in lung function in older people has major detrimental effects on the physical functioning and quality of life, and increases the risk of developing chronic obstructive pulmonary disease (COPD) and may contribute to premature mortality [1–3]. Previous longitudinal studies [4–6] focused on patients with established diagnosis of COPD and their findings indicate that the change in forced expiratory volume in 1 s (FEV1) is highly variable. They reported that increased decline in FEV1 in COPD patients was associated with current smoking [4], with increased risk of hospital readmission [7]. However, very little data are available from longitudinal informahealthcare.com
10.1586/17476348.2014.919226
studies on the decline in lung function in older people living in communities, without a confirmed diagnosis of respiratory disease (respiratory impairment) using FEV1. They are sparse and lacking robust data or predominantly on men [8]. Therefore, longitudinal studies are beneficial to healthcare providers and policy makers as a public health agenda for smoking-related diseases and to devise appropriate preventative and treatment strategies. They may also provide some insight into the impact of undiagnosed lung function impairment in older people who reside in the community. In addition, very little data are available from longitudinal studies that examined gender differences and subsequent changes in lung function [9,10]. These study findings are inconsistent [9,10], partly due to the
2014 Informa UK Ltd
ISSN 1747-6348
515
Original Research
Yohannes & Tampubolon
Expert Review of Respiratory Medicine 2014.8:515-521.
design of the methodologies and use of different statistical tests that are difficult to compare. Undiagnosed chronic cough and respiratory symptoms tend to contribute to higher level of lung function decline in women compared to men [11,12]. Thus, we took the opportunity to examine potential gender differences in the levels of FEV1 using the English Longitudinal Study of Ageing (ELSA) from the population-based data. We hypothesized that women would have substantially higher lung function decline compared to men in this longitudinal study. In order to test this hypothesis, we used data from the ELSA, a prospective observational study with 4 years of followup, to explore changes in lung function using FEV1 in community-dwelling subjects aged 50 years and older. We also examined the factors predicting differences in the level of FEV1 in older people. Methods Study population
The data were extracted from Wave 2 and Wave 4 of ELSA on individuals who underwent lung function tests on both occasions. The same individuals are re-interviewed and followed up every 2 years. Details of the study methodology have been published elsewhere [13,14]. Briefly, the sample was drawn from the Health Survey of England conducted in 1998, 1999 and 2001 for those born before March 1952. Detailed data are available on the ELSA participants at the website [15]. The sample drawn was representative of the people living in private households in England. A total of 8688 respondents participated in Wave 2 (2004–2005, response rate 82%); 7114 participated in Wave 3 (2006–2007, response rate 73%); and 6623 participated in Wave 4 (2008–2009, response rate 60%). The response rate is defined as [9] ‘total individual respondents to a given wave divided by total individuals eligible for that wave’. The inclusion of the participant in either numerator or denominator was not based on the response in any previous wave. Procedures
In all the waves, data were collected by face-to-face interviews and a self-completion questionnaire. At Wave 2 and Wave 4, a trained nurse visited the subjects at their home to take the lung function measurements. Our analysis is based on the ELSA participants who underwent lung function testing at the baseline assessment and at 4 years and for whom 4-year data are available (n = 4226 for both Wave 2 and Wave 4). All participants gave written informed consent and the study was approved by the relevant ethics committee. Lung function
The research nurse carried out two lung function tests (FEV1 and forced vital capacity [FVC]) at baseline in 2004 and during follow-up in 2008. Subjects underwent lung function test using a hand-held Vitalograph spirometer on both occasions. Three consecutive readings were taken, and as per the convention for the epidemiological study [8], the maximum of the three lung function maneuvers was recorded. 516
Serum and plasma samples were collected for measurement of biomarkers at baseline (Wave 2) and stored at -80˚C until the analysis was carried out. Association between the changes in FEV1 and the circulating levels of C-reactive protein (CRP) and fibrinogen was also examined. Socio-demographic characteristics were obtained at baseline. Educational status was categorized into three groups: higher (including university degree, other higher post or postsecondary education and ‘A level’ education or equivalent), secondary (certificate of secondary education or equivalent) and lower (including all people with lower than secondary education or no educational qualifications). Smoking status was classified as current smoker and nonsmoker. General health status was rated as poor, fair and good using self-reported measurement . The amount of alcohol consumed per day or per week over the last 12 months was recorded. Data on the presence of comorbid condition(s) for cardiovascular diseases (CVDs) at baseline (myocardial infarction, stroke, heart failure, angina or transient ischemic attacks) were extracted using the dichotomous variables ‘Yes’ or ‘No’. The severity of breathlessness was measured using the Medical Research Council dyspnea scale [9]. The Medical Research Council scale is categorized into five grades: Grade 1, not troubled by breathlessness except during strenuous exercise; Grade 2, short of breath when hurrying or walking up a slight hill; Grade 3, walks slower than contemporaries on level ground because of breathlessness, or has to stop for breath when walking at own pace; Grade 4, stops for breath after walking about 100 m or after a few minutes on level ground; and Grade 5, too breathless to leave the house or breathless when dressing or undressing [9]. Statistical analysis
Continuous variables are presented as means + standard deviation, and differences between the groups were analyzed using t-test and analysis of variance. Dichotomized variables are presented in percentages. Differences in categorical variables were assessed using the Chi-square test. Random coefficients model was fitted to the data to derive the rate of change in lung function (FEV1) and to estimate the association between lung function and individual characteristics. In particular, random slopes of time (age in years) were included, and fixed coefficient (not random coefficient) of square of age was also included to allow for nonlinear change in lung function. The virtue of this model stems from its ability to estimate each individual rate of change rather than one average rate of change for all individuals. This model is a special case of linear mixed model where random coefficients of variables other than time are allowed. This particular model is often fit to explain development over time of repeatedly observed individuals. The model has this form:
Yit = b 0 + b1 age it + b 2 X it + g i age i + e i + e it , where b0 is the usual constant term and b1 is the rate of change with g i being the random slopes or the rate of change Expert Rev. Respir. Med. 8(4), (2014)
Expert Review of Respiratory Medicine 2014.8:515-521.
Lung function decline in older people
Results
A total of 4224 participants were included in this study. Of these, 25% were aged 50–59 years, 40% were 60–69 years old, 27% were 70–79 years old and 8% were aged 80 years and above. Women comprised 55% of the participants. Of the 4224 participants, 3661 (86.6%) were nonsmokers or exsmokers and 561 (13.3) were current smokers. The baseline characteristics and changes in FEV1 during follow-up are reported in TABLE 1. The mean FVC at baseline was 3.90 l and 4 years later, it was 3.80 1, respectively, for the entire group. The mean percentage of FEV1/FVC at 4 years was 71%. Rate of change & difference in FEV1
shows the effects of predictors on the levels of FEV1 and changes in FEV1 during the 4-year follow-up period. The mean rate of change in FEV1 was -32 ± 0.95 ml/year (95% CI: -34 to -31). The absolute mean difference in FEV1 was -767 ml (95% CI: -799 to -734) for women in comparison to men (p < 0.001). Those with intermediate education had a difference in FEV1 by -119 ml (95% CI: -153 to -85; p < 0.001), but there was no difference in those having a degree compared with those with no education (p = 0.82). Selfreported good health had a positive association with preservation of lung function: those reporting good health had a larger FEV1 by 67 ml (95% CI: 36–98). Current smoking had a deleterious effect on lung function; compared to nonsmokers (including ex-smokers), current smokers had lower FEV1 of 253 ml (95% CI: -298 to -209). Daily or frequent consumption of alcohol had a positive association with better lung function wherein the FEV1 was 30 ml (95% CI: 2–58; p < 0.001). Presence of comorbid CVDs had a negative association with lung function which was lowered by -37 ml (95% CI: -64 to -10; p < 0.001). Breathlessness was also associated with lower level of lung function by -63 ml (95% CI: -77 to -49; p < 0.001). TABLE 2
informahealthcare.com
0.15
0.1 Density
for each individual i. The sum of these two terms gives, for each individual, a distinct individual rate of change; all these rates of change form a distribution of rates of change of lung function. b2 is the effect of age square and contains all (vector of) the effects of individual characteristics such as sex, education, occupational class, self-rated health, smoking, alcohol consumption, breathlessness, sputum, CVD indicators and biomarkers. ei is the random intercept, one for each individual i. eit is the usual residual random error term. The model assumes all the random coefficients to be normally distributed, but allows the random intercepts to be correlated with the random slopes. The virtue of this model is shown by focusing on g i, the individual rate of change; the sum of this individual rate with the average rate is summarized and depicted in the kernel density estimate. The empirical Bayes estimate of the rate of change for each individual in FEV1 was calculated and the resulting distribution is presented in a smoothed density function or a kernel density estimate in FIGURE 1. Analyses were performed using Stata version 12 (StataCorp LP, College station, USA, 2012). Significance was set at p < 0.05.
Original Research
0.05
0 -50
-40 -30 Rate of change, ml/year
-20
Figure 1. Distribution of estimated annual rates of change in forced expiratory volume in 1 s over the 4-year period from the English Longitudinal Study of Ageing on subjects 50 years and older (n = 4224).
Likewise, sputum production during winter months was associated with lower level of lung function by -73 ml (95% CI: -110 to -37; p < 0.001). Of the biomarkers, higher CRP had an association with lower level of FEV1 by -4.6 ml (95% CI: -6 to -3). Higher hemoglobin level had a positive association with better lung function in FEV1 by 4.6 ml (95% CI: 3–6). There was no statistically significant difference in the rate of change with the stratification of age. For example, in subjects of age 60 years and above, the rate and 95% CI were 32.28 ml/year and -34.63 to -29.93, respectively, in comparison to those of age 70 years and above, for whom the rate and 95% CI were -30.56 ml/year and -35.19 to -25.93, respectively (t = 1.8, p = 0. 09). Comparing the baseline level of FEV1 in those who completed the follow-up assessment versus who did not complete (2409 ml vs. 2162 ml) (t = -12.1, p < 0.001). Participants with low lung function may have been lost to follow-up. FIGURE 1 shows the distribution of the rate of change for each individual in the sample, and his or her empirical Bayes estimate. More than 97.5% of the distribution had a decline in FEV1 in excess of 32 ml/year. Discussion
This prospective observational study demonstrates that the rate of decline in FEV1 over a 4-year period was very high (32 ml/ year) in older people, and that the difference in lung function was associated with current smoking, sputum production in winter months, higher level of dyspnea, female gender, presence of CVD and higher level of CRP. In this study, the rate of decline in FEV1 was far higher in unselected elderly community dwellers than that previously reported in large cohort studies of 3 and 4 years of follow-up with moderate-to-severe COPD, in which the rate of decline in FEV1 was 41.0 ml/year [5], but was not significantly different from zero. Factors that contribute to the lower decline in these carefully selected sample studies are unclear; it is probably related to the 517
Original Research
Yohannes & Tampubolon
Table 1. Demographic characteristics of participants at baseline and 4 years of follow-up, and measurements of Forced expiratory volume in 1 s. Variable
Number (%)
Baseline FEV1 mean in ml
4-years FEV1 mean in ml
Age 50–59 60–69 70–79 80+
1486 (35) 1546 (37) 976 (23) 216 (5)
2732.46 2405.66 2066.21 1755.41 F = 205.21
2701.13 2460.42 2101.86 1711.01 F = 225.40
