- Email: [email protected]
Changes in Respiratory Symptoms and Health-Related Quality of Life
CHEST
Original Research HEALTH-RELATED QUALITY OF LIFE
Changes in Respiratory Symptoms and Health-Related Quality of Life* Marianne Voll-Aanerud, MD; Tomas M. L. Eagan, MD, PhD; Tore Wentzel-Larsen, MSc; Amund Gulsvik, MD, PhD; and Per S. Bakke, MD, PhD
Background: For a number of chronic diseases, health-related quality of life (HRQoL) has become an important outcome measure. Little data are available on how incidence, remission, or persistence of respiratory symptoms affect HRQoL. Methods: The Hordaland County Cohort Study was conducted between 1985 and 1997, and comprised 3,786 subjects, randomly selected, and aged 15 to 70 years in 1985. Respiratory symptoms were assessed both in 1985 and 1996/1997, and HRQoL was measured by the Short-Form 12 questionnaire in 1996/1997. Robust linear regression analysis was used to examine the relationship between changes in six respiratory symptoms and the physical component score (PCS) and mental component score (MCS). Results: Among subjects with incidence or persistence of any of the six examined respiratory symptoms, PCS and MCS were significantly lower than among subjects without symptoms. The PCS was more reduced than the MCS in symptomatic subjects; however, this trend was reduced after adjustment for the confounder’s gender, age, educational level, body mass index, and smoking status. Dyspnea attacks and dyspnea grade 2 had the largest negative impact on both PCS and MCS. Conclusions: This is the first longitudinal population study to show the negative impact of incidence and persistence of respiratory symptoms on HRQoL. (CHEST 2007; 131:1890 –1897) Key words: health-related quality of life; respiratory symptoms; Short-Form-12 Abbreviations: BMI ⫽ body mass index; EQ-5D ⫽ EuroQol five dimension; HRQoL ⫽ health-related quality of life; MCS ⫽ mental component score; PCS ⫽ physical component score; QoL ⫽ quality of life; SF ⫽ Short-Form
a number of chronic diseases, health-related F orquality of life (HRQoL) measured by standard questionnaires has become an important measure of the well-being of those affected. HRQoL is used to evaluate the magnitude of the impact of disease in observational studies, and increasingly as an outcome *From the Department of Thoracic Medicine (Drs. Voll-Aanerud and Eagan), Centre for Clinical Research (Mr. Wentzel-Larsen), and Institute of Medicine (Drs. Gulsvik and Bakke), Haukeland University Hospital, Bergen, Norway. None of the authors have any conflicts of interest related to the article or the research described. Manuscript received October 26, 2006; revision accepted February 17, 2006. Reproduction of this article is prohibited without written permission from the American College of Chest Physicians (www.chestjournal. org/misc/reprints.shtml). Correspondence to: Marianne Voll-Aanerud, MD, Department of Thoracic Medicine, Haukeland University Hospital, N-5021 Bergen, Norway; e-mail: [email protected] DOI: 10.1378/chest.06-2629 1890
measure in clinical trials.1,2 Further, HRQoL can be important as a predictor of the use of health-care resources.3–5 A disease often involves several different symptoms that may or may not have an impact on the patient well-being. Some symptoms may affect HRQoL more than others. Knowing the effect of individual symptoms on HRQoL may expand on the knowledge gained by studies of how disease affects HRQoL. Although several studies6 – 8 have found that patients with asthma and COPD have lower HRQoL scores than healthy subjects, very few studies have examined the role of respiratory symptoms. Chronic cough,9 dyspnea,9 –12 wheezing,10 and morning or night awakenings11 were associated with a worsened quality of life (QoL) in studies of patient populations with asthma. Only two population studies13,14 have Original Research
examined the impact of respiratory symptoms on HRQoL, both of which were cross-sectional in design. In South Wales, a random sample of 500 subjects aged ⱖ 70 years was examined, and the presence of dyspnea was found to significantly impair Short-Form (SF)-36 questionnaire scores.13 And, in a large patient register sample from England, the presence of wheeze, chest tightness, shortness of breath, and cough was associated with lower EuroQol five-dimension (EQ-5D) questionnaire scores.14 Respiratory symptoms are often transient, with high remission and recurrence rates. Whether HRQoL varies with the incidence or remission of respiratory symptoms has not been previously examined. The aim of the current study was to examine how the incidence, remission, and persistence of six respiratory symptoms in a general population cohort related to HRQoL, measured by the Short-Form (SF)-12 questionnaire. Materials and Methods Study Population and Sample In 1985, 3,786 randomly selected subjects aged 15 to 70 years in the Norwegian city of Bergen and 11 surrounding municipalities were mailed a questionnaire regarding respiratory symptoms. After up to two reminder letters, 3,370 subjects (89%) responded to the questionnaire.15 A follow-up was conducted in 1996/1997. Then, in addition to being sent a questionnaire on respiratory symptoms, the subjects were invited to a clinical examination. Between 1985 and
1996, 189 persons were deceased, leaving 3,181 eligible for the second phase. A total of 2,819 subjects (89%) returned the questionnaire on respiratory symptoms, and 2,405 subjects (76%) attended the clinical examination.16 While attending the clinical examination, the subjects were invited to fill in the SF-12 QoL questionnaire. For calculation of the SF-12 scores, the questionnaire had to be fully completed, which was the case for 2,306 subjects. Informed consent was obtained from each participant prior to each survey, and both surveys were approved by the Regional Committee of Medical Research Ethics. Questionnaires The wording of the questions on respiratory symptoms is given in the Appendix. The questions regarding respiratory symptoms has been validated against lung function and bronchial reactivity,17,18 and compared with the British Medical Research Council questionnaire on chronic bronchitis.19 The SF-12 questionnaire is a 12-item, general, QoL questionnaire previously validated20 and widely used in both general and diseased populations.7,21 The SF-12 measures QoL in the physical component scale (PCS) and the mental component scale (MCS).22 The scales range from 0 to 70, and a higher score indicates a better HRQoL. Changes in smoking habits were defined by whether they were neversmokers, ex-smokers, or current smokers at baseline and at follow-up, and recoded into five categories: neversmoker to neversmoker, nonsmoker to current smoker, current smoker to current smoker, current smoker to ex-smoker, and ex-smoker to ex-smoker. The nonsmoker-to-current smoker category consisted of both neversmokers to current smokers, and ex-smokers to current smokers. Pack-years smoked was estimated by multiplying the duration of smoking in years by the number of cigarettes smoked per day and dividing the product by 20. Educational level was determined at follow-up and divided into three categories: up to 9 years of
Table 1—Description of the Study Population by Gender Variables
Women (n ⫽ 1,193)
Men (n ⫽ 1,113)
Mean age (SD), yr Education,* No. (%) Primary Secondary University Smoking history,† No. (%) Neversmoker to neversmoker Ex-smoker to ex-smoker Nonsmoker to current smoker Current smoker to current smoker Current smoker to ex-smoker Pack-years,‡ No. (%) 0 1–9 10–19 20⫹ BMI,§ No. (%) Underweight Normal weight Overweight Obese
50.6 (15.3)
48.8 (14.4)
233 (19.5) 682 (57.2) 258 (21.6)
183 (16.4) 641 (57.6) 276 (24.8)
513 (45.3) 140 (12.3) 59 (5.2) 309 (27.3) 112 (9.9)
344 (32.0) 216 (20.1) 55 (5.1) 331 (30.8) 129 (12.0)
513 (45.5) 308 (27.3) 181 (16.0) 126 (11.2)
344 (32.7) 236 (22.4) 224 (21.3) 248 (23.6)
13 (1.1) 640 (54.2) 361 (30.5) 168 (14.2)
10 (0.9) 459 (41.4) 521 (46.9) 120 (10.8)
*Thirty-three subjects did not report an educational level. †Ninety-eight subjects had inconsistent answers regarding smoking. ‡One hundred ninety-eight subjects did not report pack-years. §Fourteen subjects were not weighed. www.chestjournal.org
CHEST / 131 / 6 / JUNE, 2007
1891
schooling (primary), a degree requiring 12 years of schooling (secondary), and a higher degree of education (university).23 Height and weight were measured during the clinical examination. Body mass index (BMI) was estimated by dividing the person’s weight in kilograms by the square of his/her height in meters, and categorized into four groups according to the World Health Organization24: (1) underweight (BMI ⬍ 18.5 kg/m2); (2) normal weight (BMI, 18.5 to 24.9 kg/m2); (3) overweight (BMI, 25 to 29.9 kg/m2); and (4) obese (BMI ⬎ 30 kg/m2).24 Statistical Analysis All statistical analysis was conducted using statistical software (Stata version 8.0; StataCorp; College Station, TX). Categorical variables were described by frequency distributions and tables, and continuous variables were described by means and SDs in different subgroups of the sample (Tables 1, 2). Univariate relationships between SF-12 scores and the six respiratory symptoms (morning cough, chronic cough, phlegm cough, wheeze, dyspnea attacks, and dyspnea grade 2) were examined by the Kruskal-Wallis test. The multivariate relationship between the SF-12 scores and the explanatory variables were examined in robust linear regression models. In the models, gender, educational level, smoking habits, pack-years, BMI, all symptoms, and all the first-order interaction terms were included. Age was included as a continuous variable. Insignificant interactions terms
Table 2—Description of the Symptoms Population by Gender* Respiratory Symptoms Morning cough Never Incidence Remission Persistent Chronic cough Never Incidence Remission Persistent Phlegm cough Never Incidence Remission Persistent Wheeze Never Incidence Remission Persistent Dyspnea attacks Never Incidence Remission Persistent Dyspnea grade 2 Never Incidence Remission Persistent
Women (n ⫽ 1,193)
Men (n ⫽ 1,113)
801 (67.1) 171 (14.3) 87 (7.3) 134 (11.2)
719 (64.6) 134 (12.0) 116 (10.4) 144 (12.9)
988 (82.8) 110 (9.2) 54 (4.5) 41 (3.5)
906 (81.4) 86 (7.7) 70 (6.3) 51 (4.6)
832 (69.8) 155 (13.0) 103 (8.6) 103 (8.6)
679 (61.0) 155 (13.9) 138 (12.4) 141 (12.7)
817 (68.5) 156 (13.1) 98 (8.2) 122 (10.2)
748 (67.2) 123 (11.1) 114 (10.2) 128 (11.5)
910 (76.3) 130 (10.9) 78 (6.5) 75 (6.3)
902 (81.0) 92 (8.3) 67 (6.0) 52 (4.7)
892 (74.8) 162 (13.6) 57 (4.7) 82 (6.9)
937 (84.2) 86 (7.7) 42 (3.8) 48 (4.3)
2† 0.014
0.07
⬍ 0.001
0.14
0.032
⬍ 0.001
*Data are presented as No. (%). †For gender differences in distributions of symptom change. 1892
were omitted in the final model. To partially adjust for the skewed distributions of the SF-12 summary scores, robust variance estimation, which uses the Huber/White/sandwich estimator of variance in place of traditional variance estimation, was used. Due to the large number of significance tests performed, a nominal significance level of 0.01 was used.
Results Characteristics of the study sample are presented in Table 1. A larger percentage of men had a university educational level. More women than men were neversmokers, and men tended to have higher smoking loads measured by pack-years (Table 1). The distribution of the changes in respiratory symptoms between genders is shown in Table 2. The distribution varied between genders for the symptoms phlegm cough and dyspnea grade 2. The occurrence of not having a symptom at either time point was higher for the cough symptoms among women, and for the dyspnea symptoms among men (Table 2). Both MCS and PCS had similar and skewed distribution patterns for all the symptoms. Figure 1 illustrates the distributions of PCS for dyspnea grade 2 and chronic cough. The univariate relationships between changes in respiratory symptoms and the PCS and MCS are presented in Table 3. All six symptoms affected both the PCS and the MCS in a nonrandom manner (p ⬍ 0.01). For all symptoms, there was strikingly similar pattern; in which subjects never having a symptom have the highest component scores (both PCS and MCS), subjects with remission have the second highest component scores, subjects with incidence have the second lowest component scores, and subjects with persistent symptoms have the lowest component scores. Whereas the pattern was similar for both the PCS and MCS, the difference between the highest and lowest component scores was larger for the PCS (Table 3). The magnitude of the effect on PCS and MCS also differed between symptoms, with chronic cough, dyspnea attacks, and dyspnea grade 2 having the largest negative influence. The relationships between respiratory symptoms and QoL persisted in the multivariate analyses after adjustment for gender, age, educational level, smoking, BMI, and all other respiratory symptoms (Table 4). The symptoms with the largest effect on QoL were dyspnea attacks and dyspnea grade 2 in the multivariate analyses. However, the magnitude of the effect was reduced after adjustment for the confounders. After adjustments, chronic cough was not associated with reduced QoL. This was more pronounced for the PCS, so that the adjusted negative effect of incidence or persistence of attacks of dyspnea and dyspnea grade 2 was almost as large on the MCS as on the PCS (Table 4). Significant interactions were found between wheeze and age, wheeze and smoking, and dyspnea grade 2 and Original Research
.1
.15
Distribution of physical scores among subjects without chronic cough
0
.05
Density
.1 0
.05
Density
.15
Distribution of physical scores among subjects without dyspnea grade two
0
20
40 Physical score
60
80
0
20
40
60
Physical score
Distribution of physical scores among subjects with incidence of chronic cough
.06 Density
.02 0
10
60
.1
Distribution of physical scores among subjects with remission of chronic cough
0
0
.02
.02
.04
.06
Density
.06
.08
.08
Distribution of physical scores among subjects with remission of dyspnea grade two
.04
Density
.04
.03 .02 0
.01
Density
.04
.08
Distribution of physical scores among subjects with incidence of dyspnea grade two
10
20
30 40 Physical score
50
60
20
40 Physical score
50
60
Density
.02
.03
.04
Distribution of physical scores among subjects with persistence of chronic cough
0
.01
.02 0
.01
Density
.03
.04
Distribution of physical scores among subjects with persistence of dyspnea grade two
30
10
20
30 40 Physical score
50
60
10
20
30
40 Physical score
50
60
Figure 1. Distributions of PCS for dyspnea grade 2 and chronic cough.
age on the PCS. For the MCS, significant interactions were found between smoking and phlegm cough, age and dyspnea attacks, age and dyspnea grade 2, and pack-years smoked, and dyspnea grade 2. Significant interactions were included in the final multivariate models. Among the confounders most interactions were found with age, and those are presented in Table 5. In younger subjects, persistence of dyspnea grade 2 affected the MCS more than the PCS, compared to older subjects. Persistence of wheeze affected the PCS in younger subjects more than older subjects, whereas the opposite was true for dyspnea attacks and the MCS (Table 5). www.chestjournal.org
Discussion Among subjects with incidence or persistence of any of the six examined respiratory symptoms, the PCS and MCS were lower than among subjects without symptoms. In general, the PCS was more reduced than the MCS in symptomatic subjects; however, this trend was reduced after adjustment for the confounder’s gender, age, educational level, and smoking status. Dyspnea grade 2 had a larger negative impact on the PCS in older subjects, whereas it affected the MCS more in younger subjects. First, there are some methodologic issues to disCHEST / 131 / 6 / JUNE, 2007
1893
Table 3—Physical and Mental SF-12 Scores by Symptoms PCS Symptoms
Subjects, No.
Mean (SD)
1,520 305 203 278
50.5 (8.8) 46.8 (10.3) 48.9 (8.6) 44.3 (11.4)
1,894 196 124 92
50.0 (9.0) 44.8 (11.4) 48.5 (8.3) 41.1 (11.8)
1,511 310 241 244
50.1 (8.9) 47.7 (10.2) 49.8 (8.9) 44.3 (11.4)
1,354 490 112 350
50.2 (9.0) 47.6 (10.4) 48.5 (9.5) 44.5 (10.7)
1,812 222 145 127
50.4 (8.5) 44.1 (11.8) 47.7 (10.1) 41.5 (11.9)
1,829 248 99 130
51.1 (7.7) 41.5 (11.7) 48.4 (9.3) 36.2 (11.7)
Morning cough Never Incidence Remission Persistent Chronic cough Never Incidence Remission Persistent Phlegm cough Never Incidence Remission Persistent Wheeze Never Incidence Remission Persistent Dyspnea attaks Never Incidence Remission Persistent Dyspnea 2 Never Incidence Remission Persistent
MCS p Value*
Mean (SD)
⬍ 0.01
P Value* ⬍ 0.01
51.6 (8.5) 49.0 (9.4) 50.7 (0.9) 48.0 (10.2) ⬍ 0.01
⬍ 0.01 51.3 (8.5) 46.8 (10.9) 50.9 (9.6) 46.6 (10.9)
⬍ 0.01
⬍ 0.01 51.5 (8.5) 49.0 (9.7) 51.0 (9.2) 48.2 (10.3)
⬍ 0.01
⬍ 0.01 51.4 (8.7) 48.9 (9.8) 50.4 (8.5) 48.1 (9.0)
⬍ 0.01
⬍ 0.01 51.5 (8.7) 47.5 (9.9) 49.6 (8.3) 46.5 (9.8)
⬍ 0.01
⬍ 0.01 51.5 (8.6) 47.6 (10.0) 49.4 (9.5) 46.6 (10.4)
*Based on Kruskal Wallis tests.
cuss. The strengths of this study are that it has a longitudinal design, a high response rate, and a randomly sampled community cohort with a wide age span. However, although the study is longitudinal, and true incidence and remission can be examined as predictors of the outcome at follow-up, we only have one time point for HRQoL. Thus, whether QoL has decreased due to the incidence of a symptom, or been stationary low and thereby preceded the incident symptom, cannot be resolved. The concept of HRQoL was developed in the 1970s and early 1980s, but the extensive use of QoL measurements began in the 1990s.25 Hence, the lack of QoL data in the first phase of this study is due to the state of the art of HRQoL in 1985. Second, there are several instruments available for measuring HRQoL. SF-12 has the advantage of being easy to complete in a short amount of time. The SF-12 questionnaire has shown a high degree of correspondence to SF-36 in general population surveys in nine European countries.26 SF-12 has also been validated against SF-36 and EQ-5D in patient populations with similar results.27,28 However, be1894
cause SF-12 only has two summary scales, it does not have an overall HRQoL score, nor does it measure specific domains of HRQoL such as social functioning or other domains found in SF-36. Third, the PCS and MCS scores were not normally distributed. This has prevented some earlier studies of providing multivariate modeling through standard techniques like linear regression. However, with computational advances, robust linear regression is now readily available, which enables multivariate adjustment. The distribution of PCS differed from fairly normal in those with persistent dyspnea and chronic cough to skewed in those without these symptoms and in those experiencing incidence or remission (Fig 1). However, normal distribution in some subgroups still requires statistical models that can take the overall skewedness of the distribution onto account. Fourth, socioeconomic status was assessed in terms of educational level. This is the most commonly used measurement of socioeconomic status in epidemiologic studies,29 and educational level has been shown to be an independent predictor of respiratory symptoms in a previous study23 of this cohort. Original Research
Table 4 —Multivariate Robust Linear Regression Analysis of Mean PCS and Mean MCS With Respect to Respiratory Symptoms: Adjusted for Age, Smoking Habits, Pack-Years, Educational Level, and BMI; Significant Interactions Taken Into Account*
Variables Morning cough Never Incidence Remission Persistent Chronic cough Never Incidence Remission Persistent Phlegm cough Never Incidence Remission Persistent Wheeze Never Incidence Remission Persistent Dyspnea attacks Never Incidence Remission Persistent Dyspnea 2 Never Incidence Remission Persistent
PCS Coefficient (SE); R2 ⫽ 0.318; R2 ⫽ 0.143†
MCS Coefficient (SE); R2 ⫽ 0.117; R2 ⫽ 0.034†
⫺ 0.87 (0.6) ⫺ 0.12 (0.7) ⫺ 1.20 (0.8)
⫺ 0.83 (0.6) 0.16 (0.8) 0.62 (0.9)
⫺ 1.32 (0.8) 0.90 (0.8) ⫺ 2.36 (1.2)
⫺ 1.73 (0.9) 1.19 (1.0) ⫺ 0.83 (1.4)
⫺ 0.16 (0.6) 0.33 (0.6) ⫺ 1.97 (0.7)‡
0.86 (1.0) 0.31 (1.1) 0.40 (1.4)
⫺ 0.21 (0.6) ⫺ 0.47 (0.6) ⫺ 1.2 (0.8)
⫺ 0.80 (0.7) 0.65 (0.7) ⫺ 0.53 (0.8)
⫺ 2.72 (0.7)‡ ⫺ 1.24 (0.8) ⫺ 3.52 (1.0)‡
⫺ 2.51 (0.8)‡ ⫺ 1.90 (0.8)‡ ⫺ 2.14 (0.9)‡
⫺ 6.39 (0.8)‡ ⫺ 0.52 (1.0) ⫺ 8.73 (1.2)‡
⫺ 2.52 (0.9)‡ ⫺ 3.42 (1.4)‡ ⫺ 7.59 (1.6)‡
*Age had significant interactions with wheeze and dyspnea grade 2 for PCS and with dyspnea attacks and dyspnea grade 2 for MCS. The coefficients are given for age 50 years. †R2 value from robust linear regressions with the covariates, without the symptoms in the model. ‡p ⬍ 0.01.
Fifth, the exact wording of two of the questions on respiratory symptoms was changed between baseline and follow-up (Appendix). For the question on wheeze, the time frame was changed from everwheeze to wheeze during the last 12 months. This is likely to lead to a lower estimate of wheeze at follow-up compared to baseline, hence lowering the incidence and persistence rates. If this were to affect the SF-12 scores, we believe it likely to underestimate the effect of wheeze on the PCS and MCS. To our knowledge, this is the first study to examine the relationship between changes in respiratory symptoms and HRQoL in a general population. Hence, there are not many studies with which to compare. In a Welsh study13 of an elderly population www.chestjournal.org
sample of 500 subjects, breathless subjects had reduced SF-36 scores of MCS and PCS, compared with nonbreathless subjects. In a study30 of 426 young adults in Melbourne, Australia, having wheeze was associated with lower SF-36 PCS and MCS. In a larger previous community study14 from Manchester, UK, HRQoL varied with the prevalence of four respiratory symptoms and obstructive airways disease. In this study,14 having any of the four symptoms was associated with a reduced QoL score for all five domains examined (mobility, self care, usual activities, pain/discomfort, and anxiety/depression), compared to not having the symptom. Due to the skewed distribution of the EQ-5D scores, multivariate adjustments were not made in this study.14 From studies on patient samples, symptoms have been found to predict a lower HRQoL. From a lung and allergy clinic in Massachusetts, patients with chronic cough had significantly reduced HRQoL scores while symptomatic, compared to after treatment and remission of symptoms.9 In a sample of COPD patients from Italy, an increase in dyspnea on exertion corresponded with a decrease in HRQoL, measured by St. George Respiratory Questionnaire.12 And, among subjects with asthma in a South Australian population sample, asthmatics with dyspnea and night awakenings had significantly lower SF-36 PCS and MCS than asthmatics without symptoms.11 In a retrospective analysis10 of 27 randomized clinical trials, both lung function impairment and having respiratory symptoms were associated with a decrease in HRQoL. However, comparing results from patient samples to population studies may not be valid. For instance, the impact on a subject’s state of mind of having chronic cough may not be the same for an asthmatic who knows why he/she coughs, compared with a subject for whom the cough has an unknown cause. Indeed, in a population sample such as ours, some subjects will know why they have a certain symptom, whereas others will not. We conducted the same multivariate analyses that are presented in Table 4 after exclusion of subjects with self-reported chronic lung disease. Almost no differences were seen in the results (data not shown). This could be due to a larger number of subjects without known chronic lung disease in this general population sample. However, it indicates that having respiratory symptoms is a predictor of reduced HRQoL, regardless of whether one knows the cause or not. The results from the current study show that some symptoms are more potent predictors of a reduced PCS and MCS than others, namely chronic cough, dyspnea attacks, and dyspnea grade 2. Presumably, these symptoms have a larger impact on activities in daily living than a symptom like morning cough, or could be indicators of more serious disease. However, CHEST / 131 / 6 / JUNE, 2007
1895
Table 5—Interaction Details for the Physical and Mental Scores PCS, Coefficient (SE) Variables Wheeze Never Incidence Remission Persistent Dyspnea attacks Never Incidence Remission Persistent Dyspnea 2† Never Incidence Remission Persistent
Subjects, No.
35 yr
65 yr
1,921 329 263 306
0 ⫺ 0.25 (0.7) 0.65 (0.7) ⫺ 3.2 (0.9)*
0 ⫺ 0.18 (1.0) ⫺ 1.6 (1.1) 0.86 (1.1)
2,220 256 194 149 2,235 302 126 156
0 ⫺ 4.79 (1.0)* 0.19 (1.2) ⫺ 5.64 (1.9)*
0 ⫺ 8.00 (1.0)* ⫺ 1.16 (1.9) ⫺ 11.6 (1.1)*
MCS, Coefficient (SE) 35 yr
65 yr
0 ⫺ 2.39 (1.0) ⫺ 3.44 (1.2)* ⫺ 0.33 (1.3)
0 ⫺ 2.69 (1.1)* ⫺ 0.35 (1.1) ⫺ 3.9 (1.2)
0 ⫺ 3.93 (1.2)* ⫺ 5.04 (2.0)* ⫺ 10.70 (2.2)*
0 ⫺ 1.23 (1.0) ⫺ 1.6 (1.4) ⫺ 4.46 (1.5)*
*p ⬍ 0.01. †The contrasts for interaction between dyspnea 2 and age (MCS) are given for 0 pack-years because there were significant interactions between dyspnea grade 2 and pack-years.
especially dyspnea is a complex symptom because it can be part of several different disorders. The connection between dyspnea and psychological problems like depression, panic disorders, and anxiety is complex. Dyspnea is a core feature in panic attacks and anxiety disorders,30,31 and conversely people with dyspnea often have panic attacks. The prevalence of panic disorders is higher in subjects with pulmonary disease than in the general population.32,33 HRQoL has been found to vary with gender, age, and educational level.34 In addition to adjust for these confounders in the robust linear regression models, we conducted extensive analyses of possible interactions. With gender, no interactions were found. Previous studies35–37 on patient samples have hinted at possible gender differences in how respiratory symptoms relate to HRQoL. None of these studies examined changes in respiratory symptoms. Too few studies have been undertaken to resolve whether there are gender differences in how respiratory symptoms relate to HRQoL; however, in this general population study none were found. With age, there were several significant interactions with symptoms affecting HRQoL. Younger subjects with persistent wheeze reported a lower PCS than older subjects with persistent wheeze. Remission of dyspnea attacks was negatively associated with MCS in younger subjects but not in older subjects. And, whereas persistent dyspnea grade 2 had a larger negative impact on PCS in the older subjects compared to the younger subjects, the reverse was true for the MCS. However, the small number of subjects in these groups (Table 5) implies that substantial caution must be taken in interpreting these findings. 1896
The larger picture from this study is to be found in the pattern of how changes in respiratory symptoms influence the PCS and MCS. Compared with subjects without symptoms at either time point, subjects with remission of a symptom have about the same HRQoL scores, whereas subjects with incidence or persistence of symptoms have lower HRQoL scores. Subjects with persistent symptoms have lower HRQoL scores than subjects with incident symptoms. Thus, a heavier symptom load has a larger influence on HRQoL. And, with remission of a symptom will subjects have a HRQoL score comparable to subjects without symptoms. A considerable number of subjects have respiratory symptoms, and for many of them this has an impact on their QoL. For some, the symptoms are due to a known chronic disease; for others they are not. Regardless, QoL is affected by respiratory symptoms that underscore the need to diagnose and treat the symptoms properly. Appendix The wording of the questions on respiratory symptoms was identical at baseline and follow-up except for the questions regarding chronic cough and wheezing: 1985 Do you usually cough or clear your throat in the morning? (yes, no) Do you usually have phlegm when coughing? (yes, no) Do you have a daily cough for ⱖ 3 months altogether during a year? (yes, no) Are you breathless when you climb two flights of stairs at an ordinary pace? (yes, no) [dyspnea grade 2] Original Research
Do you sometimes experience attacks of breathlessness? (yes, no) Do you ever have wheezing in your chest? (yes, no) 17 1996/1997 Do you have a cough for ⱖ 3 months altogether during a year? (yes, no) Have you had wheezing in your chest in the last 12 months? (yes, no)
18
ACKNOWLEDGMENT: Borghild Hovland and the late Ms. Bjørg Meidel made significant contributions to the collecting and punching of the data.
19 20
References 1 Kaplan RM, Ries AL, Reilly J, et al. Measurement of health-related quality of life in the National Emphysema Treatment Trial. Chest 2004; 126:781–789 2 Graf J, Koch M, Dujardin R, et al. Health-related quality of life before, 1 month after, and 9 months after intensive care in medical cardiovascular and pulmonary patients. Crit Care Med 2003; 31:2163–2169 3 van den Boom G, Rutten-van Molken MP, Tirimanna PR, et al. Association between health-related quality of life and consultation for respiratory symptoms: results from the DIMCA programme. Eur Respir J 1998; 11:67–72 4 Osman IM, Godden DJ, Friend JA, et al. Quality of life and hospital re-admission in patients with chronic obstructive pulmonary disease. Thorax 1997; 52:67–71 5 Seemungal TA, Donaldson GC, Paul EA, et al. Effect of exacerbation on quality of life in patients with chronic obstructive pulmonary disease. Am J Respir Crit Care Med 1998; 157:1418 –1422 6 Merikallio VJ, Mustalahti K, Remes ST, et al. Comparison of quality of life between asthmatic and healthy school children. Pediatr Allergy Immunol 2005; 16:332–340 7 Osman LM, Calder C, Robertson R, et al. Symptoms, quality of life, and health service contact among young adults with mild asthma. Am J Respir Crit Care Med 2000; 161:498 –503 8 Schlenk EA, Erlen JA, Dunbar-Jacob J, et al. Health-related quality of life in chronic disorders: a comparison across studies using the MOS SF-36. Qual Life Res 1998; 7:57– 65 9 French CL, Irwin RS, Curley FJ, et al. Impact of chronic cough on quality of life. Arch Intern Med 1998; 158:1657–1661 10 Carranza Rosenzweig JR, Edwards L, Lincourt W, et al. The relationship between health-related quality of life, lung function and daily symptoms in patients with persistent asthma. Respir Med 2004; 98:1157–1165 11 Goldney RD, Ruffin R, Fisher LJ, et al. Asthma symptoms associated with depression and lower quality of life: a population survey. Med J Aust 2003; 178:437– 441 12 Peruzza S, Sergi G, Vianello A, et al. Chronic obstructive pulmonary disease (COPD) in elderly subjects: impact on functional status and quality of life. Respir Med 2003; 97:612– 617 13 Ho SF, O’Mahony MS, Steward JA, et al. Dyspnoea and quality of life in older people at home. Age Ageing 2001; 30:155–159 14 Hazell M, Frank T, Frank P. Health related quality of life in individuals with asthma related symptoms. Respir Med 2003; 97:1211–1218 15 Bakke P, Gulsvik A, Lilleng P, et al. Postal survey on airborne occupational exposure and respiratory disorders in Norway: causes and consequences of non-response. J Epidemiol Commun Health 1990; 44:316 –320 16 Eagan TM, Eide GE, Gulsvik A, et al. Nonresponse in a www.chestjournal.org
21 22 23 24 25 26
27 28 29 30 31 32 33
34 35 36 37
community cohort study: predictors and consequences for exposure-disease associations. J Clin Epidemiol 2002; 55:775–781 Gulsvik A, Bakke P, Humerfelt S, et al. Measurements of respiratory symptoms and sample size to detect a given difference between treatment groups in obstructive lung disease. Eur Respir Rev 1991; 436 – 443 Welle I, Eide GE, Bakke PS, et al. The single-breath transfer factor for carbon monoxide and respiratory symptoms in a Norwegian community sample. Eur Respir J 1999; 14:1320 – 1325 Brogger JC, Bakke PS, Gulsvik A. Comparison of respiratory symptoms questionnaires. Int J Tuberc Lung Dis 2000; 4:83–90 Ware J Jr, Kosinski M, Keller SD. A 12-Item Short-Form Health Survey: construction of scales and preliminary tests of reliability and validity. Med Care 1996; 34:220 –233 Sturm R, Wells KB. Does obesity contribute as much to morbidity as poverty or smoking? Public Health 2001; 115: 229 –235 Ware JE. How to score the SF-12 physical & mental summary scales. 3rd ed. Lincoln, RI: Quality Metric, 1998 Eagan TM, Gulsvik A, Eide GE, et al. The effect of educational level on the incidence of asthma and respiratory symptoms. Respir Med 2004; 98:730 –736 Body mass index. Geneva, Switzerland: World Health Organization, 2006. Available at: http://www.who.int/bmi/ index.jsp?introPage⫽intro_3.html/. Accessed May 2, 2007 Fayers P, Machin D. Quality of life assessment, analysis and interpretation. 1st ed. New York, NY: John Wiley & Sons Ltd, 2000 Gandek B, Ware JE, Aaronson NK, et al. Cross-validation of item selection and scoring for the SF-12 Health Survey in nine countries: results from the IQOLA Project; International Quality of Life Assessment. J Clin Epidemiol 1998; 51:1171–1178 Macran S, Weatherly H, Kind P. Measuring population health: a comparison of three generic health status measures. Med Care 2003; 41:218 –231 Riddle DL, Lee KT, Stratford PW. Use of SF-36 and SF-12 health status measures: a quantitative comparison for groups versus individual patients. Med Care 2001; 39:867– 878 Liberatos P, Link BG, Kelsey JL. The measurement of social class in epidemiology. Epidemiol Rev 1988; 10:87–121 Matheson M, Raven J, Woods RK, et al. Wheeze not current asthma affects quality of life in young adults with asthma. Thorax 2002; 57:165–167 Anderson DJ, Noyes R Jr, Crowe RR. A comparison of panic disorder and generalized anxiety disorder. Am J Psychiatry 1984; 141:572–575 Verburg K, Griez E, Meijer J, et al. Respiratory disorders as a possible predisposing factor for panic disorder. J Affect Disord 1995; 33:129 –134 Spinhoven P, Ros M, Westgeest A, et al. The prevalence of respiratory disorders in panic disorder, major depressive disorder and V-code patients. Behav Res Ther 1994; 32:647– 649 Loge JH, Kaasa S. Short form 36 (SF-36) health survey: normative data from the general Norwegian population. Scand J Soc Med 1998; 26:250 –258 Belloch A, Perpina M, Martinez-Moragon E, et al. Gender differences in health-related quality of life among patients with asthma. J Asthma 2003; 40:945–953 French CT, Fletcher KE, Irwin RS. Gender differences in health-related quality of life in patients complaining of chronic cough. Chest 2004; 125:482– 488 Wijnhoven HA, Kriegsman DM, Snoek FJ, et al. Gender differences in health-related quality of life among asthma patients. J Asthma 2003; 40:189 –199 CHEST / 131 / 6 / JUNE, 2007
1897
Original Research HEALTH-RELATED QUALITY OF LIFE
Changes in Respiratory Symptoms and Health-Related Quality of Life* Marianne Voll-Aanerud, MD; Tomas M. L. Eagan, MD, PhD; Tore Wentzel-Larsen, MSc; Amund Gulsvik, MD, PhD; and Per S. Bakke, MD, PhD
Background: For a number of chronic diseases, health-related quality of life (HRQoL) has become an important outcome measure. Little data are available on how incidence, remission, or persistence of respiratory symptoms affect HRQoL. Methods: The Hordaland County Cohort Study was conducted between 1985 and 1997, and comprised 3,786 subjects, randomly selected, and aged 15 to 70 years in 1985. Respiratory symptoms were assessed both in 1985 and 1996/1997, and HRQoL was measured by the Short-Form 12 questionnaire in 1996/1997. Robust linear regression analysis was used to examine the relationship between changes in six respiratory symptoms and the physical component score (PCS) and mental component score (MCS). Results: Among subjects with incidence or persistence of any of the six examined respiratory symptoms, PCS and MCS were significantly lower than among subjects without symptoms. The PCS was more reduced than the MCS in symptomatic subjects; however, this trend was reduced after adjustment for the confounder’s gender, age, educational level, body mass index, and smoking status. Dyspnea attacks and dyspnea grade 2 had the largest negative impact on both PCS and MCS. Conclusions: This is the first longitudinal population study to show the negative impact of incidence and persistence of respiratory symptoms on HRQoL. (CHEST 2007; 131:1890 –1897) Key words: health-related quality of life; respiratory symptoms; Short-Form-12 Abbreviations: BMI ⫽ body mass index; EQ-5D ⫽ EuroQol five dimension; HRQoL ⫽ health-related quality of life; MCS ⫽ mental component score; PCS ⫽ physical component score; QoL ⫽ quality of life; SF ⫽ Short-Form
a number of chronic diseases, health-related F orquality of life (HRQoL) measured by standard questionnaires has become an important measure of the well-being of those affected. HRQoL is used to evaluate the magnitude of the impact of disease in observational studies, and increasingly as an outcome *From the Department of Thoracic Medicine (Drs. Voll-Aanerud and Eagan), Centre for Clinical Research (Mr. Wentzel-Larsen), and Institute of Medicine (Drs. Gulsvik and Bakke), Haukeland University Hospital, Bergen, Norway. None of the authors have any conflicts of interest related to the article or the research described. Manuscript received October 26, 2006; revision accepted February 17, 2006. Reproduction of this article is prohibited without written permission from the American College of Chest Physicians (www.chestjournal. org/misc/reprints.shtml). Correspondence to: Marianne Voll-Aanerud, MD, Department of Thoracic Medicine, Haukeland University Hospital, N-5021 Bergen, Norway; e-mail: [email protected] DOI: 10.1378/chest.06-2629 1890
measure in clinical trials.1,2 Further, HRQoL can be important as a predictor of the use of health-care resources.3–5 A disease often involves several different symptoms that may or may not have an impact on the patient well-being. Some symptoms may affect HRQoL more than others. Knowing the effect of individual symptoms on HRQoL may expand on the knowledge gained by studies of how disease affects HRQoL. Although several studies6 – 8 have found that patients with asthma and COPD have lower HRQoL scores than healthy subjects, very few studies have examined the role of respiratory symptoms. Chronic cough,9 dyspnea,9 –12 wheezing,10 and morning or night awakenings11 were associated with a worsened quality of life (QoL) in studies of patient populations with asthma. Only two population studies13,14 have Original Research
examined the impact of respiratory symptoms on HRQoL, both of which were cross-sectional in design. In South Wales, a random sample of 500 subjects aged ⱖ 70 years was examined, and the presence of dyspnea was found to significantly impair Short-Form (SF)-36 questionnaire scores.13 And, in a large patient register sample from England, the presence of wheeze, chest tightness, shortness of breath, and cough was associated with lower EuroQol five-dimension (EQ-5D) questionnaire scores.14 Respiratory symptoms are often transient, with high remission and recurrence rates. Whether HRQoL varies with the incidence or remission of respiratory symptoms has not been previously examined. The aim of the current study was to examine how the incidence, remission, and persistence of six respiratory symptoms in a general population cohort related to HRQoL, measured by the Short-Form (SF)-12 questionnaire. Materials and Methods Study Population and Sample In 1985, 3,786 randomly selected subjects aged 15 to 70 years in the Norwegian city of Bergen and 11 surrounding municipalities were mailed a questionnaire regarding respiratory symptoms. After up to two reminder letters, 3,370 subjects (89%) responded to the questionnaire.15 A follow-up was conducted in 1996/1997. Then, in addition to being sent a questionnaire on respiratory symptoms, the subjects were invited to a clinical examination. Between 1985 and
1996, 189 persons were deceased, leaving 3,181 eligible for the second phase. A total of 2,819 subjects (89%) returned the questionnaire on respiratory symptoms, and 2,405 subjects (76%) attended the clinical examination.16 While attending the clinical examination, the subjects were invited to fill in the SF-12 QoL questionnaire. For calculation of the SF-12 scores, the questionnaire had to be fully completed, which was the case for 2,306 subjects. Informed consent was obtained from each participant prior to each survey, and both surveys were approved by the Regional Committee of Medical Research Ethics. Questionnaires The wording of the questions on respiratory symptoms is given in the Appendix. The questions regarding respiratory symptoms has been validated against lung function and bronchial reactivity,17,18 and compared with the British Medical Research Council questionnaire on chronic bronchitis.19 The SF-12 questionnaire is a 12-item, general, QoL questionnaire previously validated20 and widely used in both general and diseased populations.7,21 The SF-12 measures QoL in the physical component scale (PCS) and the mental component scale (MCS).22 The scales range from 0 to 70, and a higher score indicates a better HRQoL. Changes in smoking habits were defined by whether they were neversmokers, ex-smokers, or current smokers at baseline and at follow-up, and recoded into five categories: neversmoker to neversmoker, nonsmoker to current smoker, current smoker to current smoker, current smoker to ex-smoker, and ex-smoker to ex-smoker. The nonsmoker-to-current smoker category consisted of both neversmokers to current smokers, and ex-smokers to current smokers. Pack-years smoked was estimated by multiplying the duration of smoking in years by the number of cigarettes smoked per day and dividing the product by 20. Educational level was determined at follow-up and divided into three categories: up to 9 years of
Table 1—Description of the Study Population by Gender Variables
Women (n ⫽ 1,193)
Men (n ⫽ 1,113)
Mean age (SD), yr Education,* No. (%) Primary Secondary University Smoking history,† No. (%) Neversmoker to neversmoker Ex-smoker to ex-smoker Nonsmoker to current smoker Current smoker to current smoker Current smoker to ex-smoker Pack-years,‡ No. (%) 0 1–9 10–19 20⫹ BMI,§ No. (%) Underweight Normal weight Overweight Obese
50.6 (15.3)
48.8 (14.4)
233 (19.5) 682 (57.2) 258 (21.6)
183 (16.4) 641 (57.6) 276 (24.8)
513 (45.3) 140 (12.3) 59 (5.2) 309 (27.3) 112 (9.9)
344 (32.0) 216 (20.1) 55 (5.1) 331 (30.8) 129 (12.0)
513 (45.5) 308 (27.3) 181 (16.0) 126 (11.2)
344 (32.7) 236 (22.4) 224 (21.3) 248 (23.6)
13 (1.1) 640 (54.2) 361 (30.5) 168 (14.2)
10 (0.9) 459 (41.4) 521 (46.9) 120 (10.8)
*Thirty-three subjects did not report an educational level. †Ninety-eight subjects had inconsistent answers regarding smoking. ‡One hundred ninety-eight subjects did not report pack-years. §Fourteen subjects were not weighed. www.chestjournal.org
CHEST / 131 / 6 / JUNE, 2007
1891
schooling (primary), a degree requiring 12 years of schooling (secondary), and a higher degree of education (university).23 Height and weight were measured during the clinical examination. Body mass index (BMI) was estimated by dividing the person’s weight in kilograms by the square of his/her height in meters, and categorized into four groups according to the World Health Organization24: (1) underweight (BMI ⬍ 18.5 kg/m2); (2) normal weight (BMI, 18.5 to 24.9 kg/m2); (3) overweight (BMI, 25 to 29.9 kg/m2); and (4) obese (BMI ⬎ 30 kg/m2).24 Statistical Analysis All statistical analysis was conducted using statistical software (Stata version 8.0; StataCorp; College Station, TX). Categorical variables were described by frequency distributions and tables, and continuous variables were described by means and SDs in different subgroups of the sample (Tables 1, 2). Univariate relationships between SF-12 scores and the six respiratory symptoms (morning cough, chronic cough, phlegm cough, wheeze, dyspnea attacks, and dyspnea grade 2) were examined by the Kruskal-Wallis test. The multivariate relationship between the SF-12 scores and the explanatory variables were examined in robust linear regression models. In the models, gender, educational level, smoking habits, pack-years, BMI, all symptoms, and all the first-order interaction terms were included. Age was included as a continuous variable. Insignificant interactions terms
Table 2—Description of the Symptoms Population by Gender* Respiratory Symptoms Morning cough Never Incidence Remission Persistent Chronic cough Never Incidence Remission Persistent Phlegm cough Never Incidence Remission Persistent Wheeze Never Incidence Remission Persistent Dyspnea attacks Never Incidence Remission Persistent Dyspnea grade 2 Never Incidence Remission Persistent
Women (n ⫽ 1,193)
Men (n ⫽ 1,113)
801 (67.1) 171 (14.3) 87 (7.3) 134 (11.2)
719 (64.6) 134 (12.0) 116 (10.4) 144 (12.9)
988 (82.8) 110 (9.2) 54 (4.5) 41 (3.5)
906 (81.4) 86 (7.7) 70 (6.3) 51 (4.6)
832 (69.8) 155 (13.0) 103 (8.6) 103 (8.6)
679 (61.0) 155 (13.9) 138 (12.4) 141 (12.7)
817 (68.5) 156 (13.1) 98 (8.2) 122 (10.2)
748 (67.2) 123 (11.1) 114 (10.2) 128 (11.5)
910 (76.3) 130 (10.9) 78 (6.5) 75 (6.3)
902 (81.0) 92 (8.3) 67 (6.0) 52 (4.7)
892 (74.8) 162 (13.6) 57 (4.7) 82 (6.9)
937 (84.2) 86 (7.7) 42 (3.8) 48 (4.3)
2† 0.014
0.07
⬍ 0.001
0.14
0.032
⬍ 0.001
*Data are presented as No. (%). †For gender differences in distributions of symptom change. 1892
were omitted in the final model. To partially adjust for the skewed distributions of the SF-12 summary scores, robust variance estimation, which uses the Huber/White/sandwich estimator of variance in place of traditional variance estimation, was used. Due to the large number of significance tests performed, a nominal significance level of 0.01 was used.
Results Characteristics of the study sample are presented in Table 1. A larger percentage of men had a university educational level. More women than men were neversmokers, and men tended to have higher smoking loads measured by pack-years (Table 1). The distribution of the changes in respiratory symptoms between genders is shown in Table 2. The distribution varied between genders for the symptoms phlegm cough and dyspnea grade 2. The occurrence of not having a symptom at either time point was higher for the cough symptoms among women, and for the dyspnea symptoms among men (Table 2). Both MCS and PCS had similar and skewed distribution patterns for all the symptoms. Figure 1 illustrates the distributions of PCS for dyspnea grade 2 and chronic cough. The univariate relationships between changes in respiratory symptoms and the PCS and MCS are presented in Table 3. All six symptoms affected both the PCS and the MCS in a nonrandom manner (p ⬍ 0.01). For all symptoms, there was strikingly similar pattern; in which subjects never having a symptom have the highest component scores (both PCS and MCS), subjects with remission have the second highest component scores, subjects with incidence have the second lowest component scores, and subjects with persistent symptoms have the lowest component scores. Whereas the pattern was similar for both the PCS and MCS, the difference between the highest and lowest component scores was larger for the PCS (Table 3). The magnitude of the effect on PCS and MCS also differed between symptoms, with chronic cough, dyspnea attacks, and dyspnea grade 2 having the largest negative influence. The relationships between respiratory symptoms and QoL persisted in the multivariate analyses after adjustment for gender, age, educational level, smoking, BMI, and all other respiratory symptoms (Table 4). The symptoms with the largest effect on QoL were dyspnea attacks and dyspnea grade 2 in the multivariate analyses. However, the magnitude of the effect was reduced after adjustment for the confounders. After adjustments, chronic cough was not associated with reduced QoL. This was more pronounced for the PCS, so that the adjusted negative effect of incidence or persistence of attacks of dyspnea and dyspnea grade 2 was almost as large on the MCS as on the PCS (Table 4). Significant interactions were found between wheeze and age, wheeze and smoking, and dyspnea grade 2 and Original Research
.1
.15
Distribution of physical scores among subjects without chronic cough
0
.05
Density
.1 0
.05
Density
.15
Distribution of physical scores among subjects without dyspnea grade two
0
20
40 Physical score
60
80
0
20
40
60
Physical score
Distribution of physical scores among subjects with incidence of chronic cough
.06 Density
.02 0
10
60
.1
Distribution of physical scores among subjects with remission of chronic cough
0
0
.02
.02
.04
.06
Density
.06
.08
.08
Distribution of physical scores among subjects with remission of dyspnea grade two
.04
Density
.04
.03 .02 0
.01
Density
.04
.08
Distribution of physical scores among subjects with incidence of dyspnea grade two
10
20
30 40 Physical score
50
60
20
40 Physical score
50
60
Density
.02
.03
.04
Distribution of physical scores among subjects with persistence of chronic cough
0
.01
.02 0
.01
Density
.03
.04
Distribution of physical scores among subjects with persistence of dyspnea grade two
30
10
20
30 40 Physical score
50
60
10
20
30
40 Physical score
50
60
Figure 1. Distributions of PCS for dyspnea grade 2 and chronic cough.
age on the PCS. For the MCS, significant interactions were found between smoking and phlegm cough, age and dyspnea attacks, age and dyspnea grade 2, and pack-years smoked, and dyspnea grade 2. Significant interactions were included in the final multivariate models. Among the confounders most interactions were found with age, and those are presented in Table 5. In younger subjects, persistence of dyspnea grade 2 affected the MCS more than the PCS, compared to older subjects. Persistence of wheeze affected the PCS in younger subjects more than older subjects, whereas the opposite was true for dyspnea attacks and the MCS (Table 5). www.chestjournal.org
Discussion Among subjects with incidence or persistence of any of the six examined respiratory symptoms, the PCS and MCS were lower than among subjects without symptoms. In general, the PCS was more reduced than the MCS in symptomatic subjects; however, this trend was reduced after adjustment for the confounder’s gender, age, educational level, and smoking status. Dyspnea grade 2 had a larger negative impact on the PCS in older subjects, whereas it affected the MCS more in younger subjects. First, there are some methodologic issues to disCHEST / 131 / 6 / JUNE, 2007
1893
Table 3—Physical and Mental SF-12 Scores by Symptoms PCS Symptoms
Subjects, No.
Mean (SD)
1,520 305 203 278
50.5 (8.8) 46.8 (10.3) 48.9 (8.6) 44.3 (11.4)
1,894 196 124 92
50.0 (9.0) 44.8 (11.4) 48.5 (8.3) 41.1 (11.8)
1,511 310 241 244
50.1 (8.9) 47.7 (10.2) 49.8 (8.9) 44.3 (11.4)
1,354 490 112 350
50.2 (9.0) 47.6 (10.4) 48.5 (9.5) 44.5 (10.7)
1,812 222 145 127
50.4 (8.5) 44.1 (11.8) 47.7 (10.1) 41.5 (11.9)
1,829 248 99 130
51.1 (7.7) 41.5 (11.7) 48.4 (9.3) 36.2 (11.7)
Morning cough Never Incidence Remission Persistent Chronic cough Never Incidence Remission Persistent Phlegm cough Never Incidence Remission Persistent Wheeze Never Incidence Remission Persistent Dyspnea attaks Never Incidence Remission Persistent Dyspnea 2 Never Incidence Remission Persistent
MCS p Value*
Mean (SD)
⬍ 0.01
P Value* ⬍ 0.01
51.6 (8.5) 49.0 (9.4) 50.7 (0.9) 48.0 (10.2) ⬍ 0.01
⬍ 0.01 51.3 (8.5) 46.8 (10.9) 50.9 (9.6) 46.6 (10.9)
⬍ 0.01
⬍ 0.01 51.5 (8.5) 49.0 (9.7) 51.0 (9.2) 48.2 (10.3)
⬍ 0.01
⬍ 0.01 51.4 (8.7) 48.9 (9.8) 50.4 (8.5) 48.1 (9.0)
⬍ 0.01
⬍ 0.01 51.5 (8.7) 47.5 (9.9) 49.6 (8.3) 46.5 (9.8)
⬍ 0.01
⬍ 0.01 51.5 (8.6) 47.6 (10.0) 49.4 (9.5) 46.6 (10.4)
*Based on Kruskal Wallis tests.
cuss. The strengths of this study are that it has a longitudinal design, a high response rate, and a randomly sampled community cohort with a wide age span. However, although the study is longitudinal, and true incidence and remission can be examined as predictors of the outcome at follow-up, we only have one time point for HRQoL. Thus, whether QoL has decreased due to the incidence of a symptom, or been stationary low and thereby preceded the incident symptom, cannot be resolved. The concept of HRQoL was developed in the 1970s and early 1980s, but the extensive use of QoL measurements began in the 1990s.25 Hence, the lack of QoL data in the first phase of this study is due to the state of the art of HRQoL in 1985. Second, there are several instruments available for measuring HRQoL. SF-12 has the advantage of being easy to complete in a short amount of time. The SF-12 questionnaire has shown a high degree of correspondence to SF-36 in general population surveys in nine European countries.26 SF-12 has also been validated against SF-36 and EQ-5D in patient populations with similar results.27,28 However, be1894
cause SF-12 only has two summary scales, it does not have an overall HRQoL score, nor does it measure specific domains of HRQoL such as social functioning or other domains found in SF-36. Third, the PCS and MCS scores were not normally distributed. This has prevented some earlier studies of providing multivariate modeling through standard techniques like linear regression. However, with computational advances, robust linear regression is now readily available, which enables multivariate adjustment. The distribution of PCS differed from fairly normal in those with persistent dyspnea and chronic cough to skewed in those without these symptoms and in those experiencing incidence or remission (Fig 1). However, normal distribution in some subgroups still requires statistical models that can take the overall skewedness of the distribution onto account. Fourth, socioeconomic status was assessed in terms of educational level. This is the most commonly used measurement of socioeconomic status in epidemiologic studies,29 and educational level has been shown to be an independent predictor of respiratory symptoms in a previous study23 of this cohort. Original Research
Table 4 —Multivariate Robust Linear Regression Analysis of Mean PCS and Mean MCS With Respect to Respiratory Symptoms: Adjusted for Age, Smoking Habits, Pack-Years, Educational Level, and BMI; Significant Interactions Taken Into Account*
Variables Morning cough Never Incidence Remission Persistent Chronic cough Never Incidence Remission Persistent Phlegm cough Never Incidence Remission Persistent Wheeze Never Incidence Remission Persistent Dyspnea attacks Never Incidence Remission Persistent Dyspnea 2 Never Incidence Remission Persistent
PCS Coefficient (SE); R2 ⫽ 0.318; R2 ⫽ 0.143†
MCS Coefficient (SE); R2 ⫽ 0.117; R2 ⫽ 0.034†
⫺ 0.87 (0.6) ⫺ 0.12 (0.7) ⫺ 1.20 (0.8)
⫺ 0.83 (0.6) 0.16 (0.8) 0.62 (0.9)
⫺ 1.32 (0.8) 0.90 (0.8) ⫺ 2.36 (1.2)
⫺ 1.73 (0.9) 1.19 (1.0) ⫺ 0.83 (1.4)
⫺ 0.16 (0.6) 0.33 (0.6) ⫺ 1.97 (0.7)‡
0.86 (1.0) 0.31 (1.1) 0.40 (1.4)
⫺ 0.21 (0.6) ⫺ 0.47 (0.6) ⫺ 1.2 (0.8)
⫺ 0.80 (0.7) 0.65 (0.7) ⫺ 0.53 (0.8)
⫺ 2.72 (0.7)‡ ⫺ 1.24 (0.8) ⫺ 3.52 (1.0)‡
⫺ 2.51 (0.8)‡ ⫺ 1.90 (0.8)‡ ⫺ 2.14 (0.9)‡
⫺ 6.39 (0.8)‡ ⫺ 0.52 (1.0) ⫺ 8.73 (1.2)‡
⫺ 2.52 (0.9)‡ ⫺ 3.42 (1.4)‡ ⫺ 7.59 (1.6)‡
*Age had significant interactions with wheeze and dyspnea grade 2 for PCS and with dyspnea attacks and dyspnea grade 2 for MCS. The coefficients are given for age 50 years. †R2 value from robust linear regressions with the covariates, without the symptoms in the model. ‡p ⬍ 0.01.
Fifth, the exact wording of two of the questions on respiratory symptoms was changed between baseline and follow-up (Appendix). For the question on wheeze, the time frame was changed from everwheeze to wheeze during the last 12 months. This is likely to lead to a lower estimate of wheeze at follow-up compared to baseline, hence lowering the incidence and persistence rates. If this were to affect the SF-12 scores, we believe it likely to underestimate the effect of wheeze on the PCS and MCS. To our knowledge, this is the first study to examine the relationship between changes in respiratory symptoms and HRQoL in a general population. Hence, there are not many studies with which to compare. In a Welsh study13 of an elderly population www.chestjournal.org
sample of 500 subjects, breathless subjects had reduced SF-36 scores of MCS and PCS, compared with nonbreathless subjects. In a study30 of 426 young adults in Melbourne, Australia, having wheeze was associated with lower SF-36 PCS and MCS. In a larger previous community study14 from Manchester, UK, HRQoL varied with the prevalence of four respiratory symptoms and obstructive airways disease. In this study,14 having any of the four symptoms was associated with a reduced QoL score for all five domains examined (mobility, self care, usual activities, pain/discomfort, and anxiety/depression), compared to not having the symptom. Due to the skewed distribution of the EQ-5D scores, multivariate adjustments were not made in this study.14 From studies on patient samples, symptoms have been found to predict a lower HRQoL. From a lung and allergy clinic in Massachusetts, patients with chronic cough had significantly reduced HRQoL scores while symptomatic, compared to after treatment and remission of symptoms.9 In a sample of COPD patients from Italy, an increase in dyspnea on exertion corresponded with a decrease in HRQoL, measured by St. George Respiratory Questionnaire.12 And, among subjects with asthma in a South Australian population sample, asthmatics with dyspnea and night awakenings had significantly lower SF-36 PCS and MCS than asthmatics without symptoms.11 In a retrospective analysis10 of 27 randomized clinical trials, both lung function impairment and having respiratory symptoms were associated with a decrease in HRQoL. However, comparing results from patient samples to population studies may not be valid. For instance, the impact on a subject’s state of mind of having chronic cough may not be the same for an asthmatic who knows why he/she coughs, compared with a subject for whom the cough has an unknown cause. Indeed, in a population sample such as ours, some subjects will know why they have a certain symptom, whereas others will not. We conducted the same multivariate analyses that are presented in Table 4 after exclusion of subjects with self-reported chronic lung disease. Almost no differences were seen in the results (data not shown). This could be due to a larger number of subjects without known chronic lung disease in this general population sample. However, it indicates that having respiratory symptoms is a predictor of reduced HRQoL, regardless of whether one knows the cause or not. The results from the current study show that some symptoms are more potent predictors of a reduced PCS and MCS than others, namely chronic cough, dyspnea attacks, and dyspnea grade 2. Presumably, these symptoms have a larger impact on activities in daily living than a symptom like morning cough, or could be indicators of more serious disease. However, CHEST / 131 / 6 / JUNE, 2007
1895
Table 5—Interaction Details for the Physical and Mental Scores PCS, Coefficient (SE) Variables Wheeze Never Incidence Remission Persistent Dyspnea attacks Never Incidence Remission Persistent Dyspnea 2† Never Incidence Remission Persistent
Subjects, No.
35 yr
65 yr
1,921 329 263 306
0 ⫺ 0.25 (0.7) 0.65 (0.7) ⫺ 3.2 (0.9)*
0 ⫺ 0.18 (1.0) ⫺ 1.6 (1.1) 0.86 (1.1)
2,220 256 194 149 2,235 302 126 156
0 ⫺ 4.79 (1.0)* 0.19 (1.2) ⫺ 5.64 (1.9)*
0 ⫺ 8.00 (1.0)* ⫺ 1.16 (1.9) ⫺ 11.6 (1.1)*
MCS, Coefficient (SE) 35 yr
65 yr
0 ⫺ 2.39 (1.0) ⫺ 3.44 (1.2)* ⫺ 0.33 (1.3)
0 ⫺ 2.69 (1.1)* ⫺ 0.35 (1.1) ⫺ 3.9 (1.2)
0 ⫺ 3.93 (1.2)* ⫺ 5.04 (2.0)* ⫺ 10.70 (2.2)*
0 ⫺ 1.23 (1.0) ⫺ 1.6 (1.4) ⫺ 4.46 (1.5)*
*p ⬍ 0.01. †The contrasts for interaction between dyspnea 2 and age (MCS) are given for 0 pack-years because there were significant interactions between dyspnea grade 2 and pack-years.
especially dyspnea is a complex symptom because it can be part of several different disorders. The connection between dyspnea and psychological problems like depression, panic disorders, and anxiety is complex. Dyspnea is a core feature in panic attacks and anxiety disorders,30,31 and conversely people with dyspnea often have panic attacks. The prevalence of panic disorders is higher in subjects with pulmonary disease than in the general population.32,33 HRQoL has been found to vary with gender, age, and educational level.34 In addition to adjust for these confounders in the robust linear regression models, we conducted extensive analyses of possible interactions. With gender, no interactions were found. Previous studies35–37 on patient samples have hinted at possible gender differences in how respiratory symptoms relate to HRQoL. None of these studies examined changes in respiratory symptoms. Too few studies have been undertaken to resolve whether there are gender differences in how respiratory symptoms relate to HRQoL; however, in this general population study none were found. With age, there were several significant interactions with symptoms affecting HRQoL. Younger subjects with persistent wheeze reported a lower PCS than older subjects with persistent wheeze. Remission of dyspnea attacks was negatively associated with MCS in younger subjects but not in older subjects. And, whereas persistent dyspnea grade 2 had a larger negative impact on PCS in the older subjects compared to the younger subjects, the reverse was true for the MCS. However, the small number of subjects in these groups (Table 5) implies that substantial caution must be taken in interpreting these findings. 1896
The larger picture from this study is to be found in the pattern of how changes in respiratory symptoms influence the PCS and MCS. Compared with subjects without symptoms at either time point, subjects with remission of a symptom have about the same HRQoL scores, whereas subjects with incidence or persistence of symptoms have lower HRQoL scores. Subjects with persistent symptoms have lower HRQoL scores than subjects with incident symptoms. Thus, a heavier symptom load has a larger influence on HRQoL. And, with remission of a symptom will subjects have a HRQoL score comparable to subjects without symptoms. A considerable number of subjects have respiratory symptoms, and for many of them this has an impact on their QoL. For some, the symptoms are due to a known chronic disease; for others they are not. Regardless, QoL is affected by respiratory symptoms that underscore the need to diagnose and treat the symptoms properly. Appendix The wording of the questions on respiratory symptoms was identical at baseline and follow-up except for the questions regarding chronic cough and wheezing: 1985 Do you usually cough or clear your throat in the morning? (yes, no) Do you usually have phlegm when coughing? (yes, no) Do you have a daily cough for ⱖ 3 months altogether during a year? (yes, no) Are you breathless when you climb two flights of stairs at an ordinary pace? (yes, no) [dyspnea grade 2] Original Research
Do you sometimes experience attacks of breathlessness? (yes, no) Do you ever have wheezing in your chest? (yes, no) 17 1996/1997 Do you have a cough for ⱖ 3 months altogether during a year? (yes, no) Have you had wheezing in your chest in the last 12 months? (yes, no)
18
ACKNOWLEDGMENT: Borghild Hovland and the late Ms. Bjørg Meidel made significant contributions to the collecting and punching of the data.
19 20
References 1 Kaplan RM, Ries AL, Reilly J, et al. Measurement of health-related quality of life in the National Emphysema Treatment Trial. Chest 2004; 126:781–789 2 Graf J, Koch M, Dujardin R, et al. Health-related quality of life before, 1 month after, and 9 months after intensive care in medical cardiovascular and pulmonary patients. Crit Care Med 2003; 31:2163–2169 3 van den Boom G, Rutten-van Molken MP, Tirimanna PR, et al. Association between health-related quality of life and consultation for respiratory symptoms: results from the DIMCA programme. Eur Respir J 1998; 11:67–72 4 Osman IM, Godden DJ, Friend JA, et al. Quality of life and hospital re-admission in patients with chronic obstructive pulmonary disease. Thorax 1997; 52:67–71 5 Seemungal TA, Donaldson GC, Paul EA, et al. Effect of exacerbation on quality of life in patients with chronic obstructive pulmonary disease. Am J Respir Crit Care Med 1998; 157:1418 –1422 6 Merikallio VJ, Mustalahti K, Remes ST, et al. Comparison of quality of life between asthmatic and healthy school children. Pediatr Allergy Immunol 2005; 16:332–340 7 Osman LM, Calder C, Robertson R, et al. Symptoms, quality of life, and health service contact among young adults with mild asthma. Am J Respir Crit Care Med 2000; 161:498 –503 8 Schlenk EA, Erlen JA, Dunbar-Jacob J, et al. Health-related quality of life in chronic disorders: a comparison across studies using the MOS SF-36. Qual Life Res 1998; 7:57– 65 9 French CL, Irwin RS, Curley FJ, et al. Impact of chronic cough on quality of life. Arch Intern Med 1998; 158:1657–1661 10 Carranza Rosenzweig JR, Edwards L, Lincourt W, et al. The relationship between health-related quality of life, lung function and daily symptoms in patients with persistent asthma. Respir Med 2004; 98:1157–1165 11 Goldney RD, Ruffin R, Fisher LJ, et al. Asthma symptoms associated with depression and lower quality of life: a population survey. Med J Aust 2003; 178:437– 441 12 Peruzza S, Sergi G, Vianello A, et al. Chronic obstructive pulmonary disease (COPD) in elderly subjects: impact on functional status and quality of life. Respir Med 2003; 97:612– 617 13 Ho SF, O’Mahony MS, Steward JA, et al. Dyspnoea and quality of life in older people at home. Age Ageing 2001; 30:155–159 14 Hazell M, Frank T, Frank P. Health related quality of life in individuals with asthma related symptoms. Respir Med 2003; 97:1211–1218 15 Bakke P, Gulsvik A, Lilleng P, et al. Postal survey on airborne occupational exposure and respiratory disorders in Norway: causes and consequences of non-response. J Epidemiol Commun Health 1990; 44:316 –320 16 Eagan TM, Eide GE, Gulsvik A, et al. Nonresponse in a www.chestjournal.org
21 22 23 24 25 26
27 28 29 30 31 32 33
34 35 36 37
community cohort study: predictors and consequences for exposure-disease associations. J Clin Epidemiol 2002; 55:775–781 Gulsvik A, Bakke P, Humerfelt S, et al. Measurements of respiratory symptoms and sample size to detect a given difference between treatment groups in obstructive lung disease. Eur Respir Rev 1991; 436 – 443 Welle I, Eide GE, Bakke PS, et al. The single-breath transfer factor for carbon monoxide and respiratory symptoms in a Norwegian community sample. Eur Respir J 1999; 14:1320 – 1325 Brogger JC, Bakke PS, Gulsvik A. Comparison of respiratory symptoms questionnaires. Int J Tuberc Lung Dis 2000; 4:83–90 Ware J Jr, Kosinski M, Keller SD. A 12-Item Short-Form Health Survey: construction of scales and preliminary tests of reliability and validity. Med Care 1996; 34:220 –233 Sturm R, Wells KB. Does obesity contribute as much to morbidity as poverty or smoking? Public Health 2001; 115: 229 –235 Ware JE. How to score the SF-12 physical & mental summary scales. 3rd ed. Lincoln, RI: Quality Metric, 1998 Eagan TM, Gulsvik A, Eide GE, et al. The effect of educational level on the incidence of asthma and respiratory symptoms. Respir Med 2004; 98:730 –736 Body mass index. Geneva, Switzerland: World Health Organization, 2006. Available at: http://www.who.int/bmi/ index.jsp?introPage⫽intro_3.html/. Accessed May 2, 2007 Fayers P, Machin D. Quality of life assessment, analysis and interpretation. 1st ed. New York, NY: John Wiley & Sons Ltd, 2000 Gandek B, Ware JE, Aaronson NK, et al. Cross-validation of item selection and scoring for the SF-12 Health Survey in nine countries: results from the IQOLA Project; International Quality of Life Assessment. J Clin Epidemiol 1998; 51:1171–1178 Macran S, Weatherly H, Kind P. Measuring population health: a comparison of three generic health status measures. Med Care 2003; 41:218 –231 Riddle DL, Lee KT, Stratford PW. Use of SF-36 and SF-12 health status measures: a quantitative comparison for groups versus individual patients. Med Care 2001; 39:867– 878 Liberatos P, Link BG, Kelsey JL. The measurement of social class in epidemiology. Epidemiol Rev 1988; 10:87–121 Matheson M, Raven J, Woods RK, et al. Wheeze not current asthma affects quality of life in young adults with asthma. Thorax 2002; 57:165–167 Anderson DJ, Noyes R Jr, Crowe RR. A comparison of panic disorder and generalized anxiety disorder. Am J Psychiatry 1984; 141:572–575 Verburg K, Griez E, Meijer J, et al. Respiratory disorders as a possible predisposing factor for panic disorder. J Affect Disord 1995; 33:129 –134 Spinhoven P, Ros M, Westgeest A, et al. The prevalence of respiratory disorders in panic disorder, major depressive disorder and V-code patients. Behav Res Ther 1994; 32:647– 649 Loge JH, Kaasa S. Short form 36 (SF-36) health survey: normative data from the general Norwegian population. Scand J Soc Med 1998; 26:250 –258 Belloch A, Perpina M, Martinez-Moragon E, et al. Gender differences in health-related quality of life among patients with asthma. J Asthma 2003; 40:945–953 French CT, Fletcher KE, Irwin RS. Gender differences in health-related quality of life in patients complaining of chronic cough. Chest 2004; 125:482– 488 Wijnhoven HA, Kriegsman DM, Snoek FJ, et al. Gender differences in health-related quality of life among asthma patients. J Asthma 2003; 40:189 –199 CHEST / 131 / 6 / JUNE, 2007
1897