the preventable burden of productivity loss due to suboptimal asthma control

CHEST Original ResearchASTHMA

journal.publications.chestnet.org CHEST / 145 / 4 / APRIL 2014 787

Productivity loss is the opportunity cost due to foregone labor; it can either be in the form of

absenteeism (the withdrawal of labor) or presentee-ism (ineffi ciency of labor due to the impairment). 1 Several studies have shown that the indirect costs of asthma surpass the direct costs. 2-16 A landmark early study reported that among the total population of US adults aged ! 18 years, asthma-related indirect costs amounted to $1.268 billion, compared with the direct costs of $1.197 billion (in 1990 US dollars). 2 Similarly, an early Canadian study reported the indi-rect costs of asthma to amount to 50% of the total costs of the disease. 16 Studies in North America also

reported high economic burden due to asthma-related productivity loss. 17,18 A recent European study has esti-mated the indirect costs to constitute 62.5% of total costs of asthma. 3

Despite the central role of clinical control in the contemporary management of asthma, only a few stud-ies have reported on the relationship between pro-ductivity loss and asthma control. 10,19 In addition, a simple reporting of productivity loss across levels of control, though informative on its own, is of limited value to policy makers as it does not infer any causal association. This is because of the impact of con-founding factors affecting both asthma control and

Background: Productivity loss is an overlooked aspect of the burden of chronic health conditions. While modern guidelines emphasize achieving clinical control in asthma management, few studies have reported on the relationship between asthma control and productivity loss. We calculated the productivity loss that can be avoided by achieving and maintaining clinical control in employed adults with asthma. Methods: We prospectively recruited a population-based random sample of adults with asthma in British Columbia, Canada. We measured productivity loss due to absenteeism and presenteeism using validated instruments, and ascertained asthma control according to the GINA (Global Ini-tiative for Asthma) classifi cation. We estimated the average gain in productivity for each indi-vidual if the individual’s asthma was controlled in the past week, by fi tting two-part regression models associating asthma control and productivity loss, controlling for potential confounding variables. Results: The fi nal sample included 300 employed adults (mean age, 47.9 years [SD 12.0]; 67.3% women). Of these, 49 (16.3%) reported absenteeism, and 137 (45.7%) reported presentee-ism. Productivity loss due to presenteeism, but not absenteeism, was associated with asthma control. A person with uncontrolled asthma would avoid $184.80 (Canadian dollars [CAD]) in productivity loss by achieving clinical control during a week, CAD$167.50 (90.6%) of which would be due to presenteeism. The corresponding value was CAD$34.20 for partially controlled asthma and was not statistically signifi cant. Conclusions: Our results indicate that substantial gain in productivity can be obtained by achieving asthma control. Presenteeism is more responsive than absenteeism to asthma control, and, thus, is a more important source of preventable burden. CHEST 2014; 145(4):787–793

Abbreviations: CAD 5 Canadian dollars; VOLP 5 Valuation of Lost Productivity; WPAI 5 Work Productivity and Activity Impairment

The Preventable Burden of Productivity Loss Due to Suboptimal Asthma Control A Population -Based Study

Mohsen Sadatsafavi , MD, PhD ; Roxanne Rousseau , BSc ; Wenjia Chen , MSc ; Wei Zhang , PhD ; Larry Lynd , PhD ; J. Mark FitzGerald , MD ; and the Economic Burden of Asthma Study Team *

Downloaded From: http://journal.publications.chestnet.org/ on 10/07/2016

http://journal.publications.chestnet.org

788 Original Research

National Occupation Classifi cation code for the year 2010 from Statistics Canada. 22 All costs were, therefore, in 2010 Canadian dollars (CAD).

The effect of a patient’s asthma on their work was assessed using the Work Productivity and Activity Impairment (WPAI) 23 and the Valuation of Lost Productivity (VOLP) 24 questionnaires. The WPAI questionnaire captures the work time lost due to absenteeism and presenteeism, with a recall period of 1 week. 23 Specifi cally, for absenteeism, it asks the participant about how many hours they missed from work in the past 7 days because of their health, asking the respondent to consider hours missed on sick days and the time they went in late or left early. For pre-senteeism, the questionnaire asks about the extent to which indi-vidual’s health affected their productivity while the person was working, asking the respondent to think about days they were limited in the amount or kind of work they could do, days they accomplished less than they would like, or days they could not do their work as carefully as usual. The VOLP questionnaire captures aspects of the current work environment, such as the contribu-tion of the individual to team productivity, availability of a replace-ment, and time sensitivity of the job. 24 Such information is used to produce multipliers, separately for absenteeism and presen-teeism, that convert the hourly wage to marginal value of produc-tivity for an hour of work, considering the impact of individual’s productivity loss in the work environment. 25

Assessing Asthma Control

We applied the Global Initiative for Asthma (GINA) defi nition of asthma control. 26 In this defi nition, asthma is categorized as controlled, partially controlled, or uncontrolled, based on measures of perceived impairment as well as the ratio of FEV 1 , obtained through spirometry, to its predicted value. We used the Third National Health and Nutrition Examination Survey reference stan-dards for calculating the predicted FEV 1 . 27

Statistical Analysis

All analyses were performed using SAS, version 9.3 (SAS Insti-tute Inc). Two-tailed P values at .05 were considered statistically signifi cant. We performed descriptive analysis of baseline variables according to the level of asthma control, using analysis of variance for continuous variables and the x 2 test for categorical variables to compare their distribution across asthma control levels.

Productivity loss at the individual level in this study is the prod-uct of three terms: the weekly hours of work lost, the individual’s hourly wage, and the VOLP multiplier converting wage to produc-tivity rate. Unadjusted analysis included reporting the percent-ages of individuals with lost work due to presenteeism/absenteeism, and weekly hours of lost work across levels of asthma control.

Adjusted Analysis: Estimating the Preventable Productivity Loss: We fi tted regression models with hours of productivity loss (separately for absenteeism and presenteeism) as the dependent variable, and asthma control and potential confounding variables as independent variables. Asthma control entered the models as two dummy variables representing partially controlled and uncon-trolled asthma (with the reference being controlled asthma). The potential confounding variables, selected from a larger set of candidates based on exploratory analysis of crude associations, consisted of sex, age at baseline visit, socioeconomic status, educa-tion, the type of residence (urban or rural), and the possession of drug insurance (no coverage, partial coverage, and full coverage for medications).

Given the zero-infl ated and right-skewed nature of the pro-ductivity loss data, we used two-part regression models with logis-tic and Gamma (generalized linear model with logarithmic link

productivity, and the fact that even patients with con-trolled asthma might still experience some produc-tivity loss, which means a fraction of productivity loss is unavoidable even when achieving total asthma con-trol. Our research question in the present work was: How much productivity loss, from the societal perspec-tive and considering both absenteeism and presen-teeism, can be averted by achieving and maintaining clinical control in individuals with partially controlled or uncontrolled asthma?

Materials and Methods

We used the prospectively collected cross-sectional data of the Economic Burden of Asthma (EBA) study, a longitudinal study with the ultimate aim of measuring the burden of asthma in Canada (University of British Columbia Human Ethics No. H10-01542). The study’s catchment areas were composed of two census sub-di visions in British Columbia, Canada: the Vancouver subdivision and Central Okanagan subdivision, with a 2011 population of 603,502 and 179,830, respectively. 20 The rationale for the selection of the two areas was to increase the generalizability of fi ndings by including both urban and rural populations. Individuals aged 1 to 85 years with a self-reported diagnosis of asthma by a physician were identifi ed using random-digit dialing (including both land-line and cellular numbers). Eligibility criteria also included having had at least one encounter with the health-care system (eg, visit to a doctor, ED, or hospital, or receiving an inhaler medication) in the past 5 years. Eligible individuals were invited to the study centers, and for those who gave written informed consent, a series of questionnaires was administered. The data collected in the baseline visit included demographics, socioeconomic status, asthma-related symptoms, employment status (and if employed, job title and description), quality of life, mental health status, and pro-ductivity loss. In addition, a lung function test was administered by a trained technician. The sample for the present study included adults who were employed at the time of recruitment.

We matched the stated job titles and descriptions to the National Occupation Classifi cation codes. 21 This was performed indepen-dently by two investigators, and disagreements were resolved by a third investigator. The hourly wage for each individual was esti-mated from the sex- and age-specifi c values for the corresponding

Manuscript received July 15, 2013; revision accepted November 15, 2013. Affi liations: From the Institute for Heart and Lung Health (Drs Sadatsafavi and FitzGerald and Ms Rousseau), Department of Medicine; Centre for Clinical Epidemiology and Evaluation (Drs Sadatsafavi and FitzGerald); Collaboration for Outcomes Research and Evaluation, Faculty of Pharmaceutical Sciences (Ms Chen and Dr Lynd); and the Centre for Health Evaluation and Outcome Sciences (Drs Zhang and Lynd), the University of British Columbia, Vancouver, BC, Canada. *A complete list of the study team members can be found in e-Appendix 1. Funding/Support : This study is funded through the Collabo-rative Innovative Research Fund, an investigator-initiated, peer-reviewed competition sponsored by GlaxoSmithKline Canada. Correspondence to: Mohsen Sadatsafavi, MD, PhD, Centre for Clinical Epidemiology and Evaluation, 7th Floor, 828 W 10th Ave, Research Pavilion, Vancouver, BC, V5Z 1M9, Canada; e-mail: [email protected] © 2014 American College of Chest Physicians. Reproduction of this article is prohibited without written permission from the American College of Chest Physicians. See online for more details. DOI: 10.1378/chest.13-1619


mailto:[email protected]


number of hours of work lost per week was 5.1 for controlled, 6.2 for partially controlled, and 10.1 for uncontrolled asthma ( P 5 .03). The unadjusted time loss was signifi cantly different across control levels for presenteeism ( P , .01) but not for absenteeism ( P 5 .40).

Adjusted Analysis

The results of the regression analysis associating asthma control and productivity loss, controlling for the effect of potential confounders, are presented in Table 3 . Compared with controlled asthma, neither partially controlled nor uncontrolled asthma was asso-ciated with absenteeism, whereas partially controlled asthma (OR 5 2.03; 95% CI, 1.02-4.05) and uncon-trolled asthma (OR 5 3.41; 95% CI, 1.72-6.76) were associated with presenteeism.

The cost of productivity loss due to presenteeism, but not absenteeism, was associated with asthma con-trol ( Table 3 ). When absenteeism- and presenteeism-related productivity loss estimates were combined, the preventable burden of uncontrolled asthma was esti-mated to be 4.10 (95% CI, 0.83-7.37) work hours per week, equivalent to CAD$184.80 (95% CI, 42.2-327.4). The estimated preventable burden of 0.22 h ( P 5 .68) or CAD$34.20 ( P 5 .60) for partially controlled asthma was not statistically signifi cant. Coeffi cients for cova-riates obtained from the regression model are pre-sented in the e-Tables 1-3.

function and Gamma distribution) components. 28 The logistic component calculated an OR for the association between asthma control and presence of productivity loss, and the Gamma compo-nent generated a rate ratio (RR) for the association of asthma control with the hours of lost work in the subset of individuals who reported productivity loss. The two measures were com-bined to generate an overall incremental effect of uncontrolled or partially controlled asthma on hours of lost work, with controlled asthma as the reference category. This incremental value was then converted to monetary value of lost productivity using the indi-vidual wage and VOLP multipliers. CIs and P values were esti-mated using parametric bootstrapping by repeating the entire process 100 times. 28

Unlike the unadjusted analysis, this analysis took into account the impact of potential confounders, and by taking individuals with controlled asthma as the reference group, it removed the impact of any non-asthma-related productivity loss as well as the residual asthma-related productivity loss that even patients with controlled asthma can experience. As such, the coeffi cients of productivity loss associated with partially controlled and uncon-trolled asthma in this framework can have a causal interpretation, answering our research question.

Missing Values: A small percentage of covariates were missing. To optimally analyze the available data yet properly account for the uncertainty around the missing values, we performed multi-ple imputation of such missing data by imputing the missing data within each of the 100 bootstraps . 29

Sensitivity Analyses: Sensitivity analyses included using a regres-sion model with the dependent variable being the monetary value of productivity loss (instead of the hours of lost work), repeating the analysis after removing observations with outlying values of VOLP multiplier, and alternative handling of equivocal responses to asthma-symptoms questionnaires.

Results

Figure 1 illustrates the fl owchart of number of indi-viduals according to their asthma control and pro-ductivity loss status. The final sample consisted of 300 individuals recruited between January 2011 and July 2012.

In 59 (19.7%), 119 (39.7%), and 122 (40.7%) indi-viduals, asthma was classified respectively as con-trolled, partially controlled, and uncontrolled. Baseline characteristics of the study subjects are presented in Table 1 . 30,31

Unadjusted Analysis

Table 2 presents productivity loss overall and across asthma control levels. In total, 146 individuals (48.7%) reported productivity loss due to asthma. Presenteeism was more common than absenteeism, with 137 report-ing presenteeism (45.7%) and 49 reporting absen-teeism (16.3%).

On average, an employed adult with asthma lost 7.6 h per week for health-related productivity loss, equal to 20.9% of the time they worked; 5.5 h (72.4%) of the lost productivity was due to presenteeism. The

Figure 1. Flow chart of the sample selection.




Sensitivity Analyses: Results of the sensitivity anal-yses are provided in e-Table 4. In none of the analyses did asthma control become associated with absen-teeism. On the other hand, asthma control remained signifi cantly associated with presenteeism in all the analyses.

Discussion

In this prospective cross-sectional study, we studied the association of productivity loss with asthma con-trol. In our study, presenteeism was not only the larger component of productivity loss, it was more respon-sive to asthma control than absenteeism. Because the contemporary belief is that control is achievable in the majority of individuals with asthma, the adjusted

weekly productivity loss of CAD$184.80 among indi-viduals with uncontrolled asthma is a preventable source of burden. To put this value in context, patients with current asthma were estimated to cost the health-care system of British Columbia, on average, CAD$461 (in 2010 CAD) per year in billed medical goods and services, 32 equal to 2.5 weeks of preventable produc-tivity loss in patients with uncontrolled asthma. This simple comparison suggests that by solely relying on billed services and ignoring indirect costs, policy mak-ers severely underestimate the burden of the disease and risk misallocating resources.

We believe there are plausible mechanisms explain-ing the observed association between asthma control and presenteeism and lack of such association with absenteeism. Individuals with asthma might use all their allowed sick leave and so be forced to be present

Table 1— Demographic Characteristics of the Study Sample, According to the Level of Control

Characteristics All (N 5 300 ) Controlled (n 5 59)Partially Controlled

(n 5 119) Uncontrolled (n 5 122) P ValueSubjects With

Missing Data, No.

Age, mean (SD), y 47.88 (11.98) 46.00 (12.49) 47.99 (11.90) 48.68 (11.81) .37 0Sex .11 0 Male 32.7 30.5 39.5 27.0 Female 67.3 69.5 60.5 73.0 Household income .61 11 Low 38.0 42.4 34.5 39.3 High ( . CAD$60K/y) a 62.0 57.6 65.5 60.7 Education b .08 0 Low 19.7 25.4 13.4 23.0 High 80.3 74.6 86.6 77.0 Place of birth .67 0 Canada 72.0 71.2 74.8 69.7 Foreign born 28.0 28.8 25.2 30.3 Residence type c .85 0 Urban 92.0 93.2 92.4 91.0 Rural 8.0 6.8 7.6 9.0 Insurance for medications .93 11 None 13.3 14.0 11.9 14.3 Partial 58.6 60.3 55.9 60.5 Complete 0.226 22.0 25.4 20.2

Data given as % unless otherwise indicated. CAD 5 Canadian dollars. a Cutoff of CAD$60K was chosen because it is close to the median household income in Canada. 30 b High education was defi ned as having obtained a 4-y college or university degree or higher. c Rural areas were defi ned as postal codes with , 400 people/km 2 . 31

Table 2— Productivity Loss by Type (Absenteeism or Presenteeism) and by Level of Asthma Control

Productivity Loss Outcomes All Controlled Partially Controlled Uncontrolled P Value

Participants with productivity loss 300 59 119 122 … Any type 146 (48.7) 22 (37.3) 53 (44.5) 71 (58.2) .02 a Absenteeism 49 (16.3) 10 (16.9) 12 (10.1) 27 (22.1) .04 a Presenteeism 137 (45.7) 17 (28.8) 51 (42.9) 69 (56.6) , .01 a Total hours worked, mean (SE) 36.1 (17.8) 37.0 (15.2) 35.0 (16.7) 36.8 (19.9) .68Hours lost, mean (SE) [%] Any type 7.6 (0.8) [20.9] 5.1 (1.5) [13.7] 6.2 (1.2) [17.6] 10.1 (1.5) [27.5] .03 a Absenteeism 2.1 (0.4) [5.8] 2.4 (1.0) [6.6] 1.4 (0.6) [4.0] 2.6 (0.7) [7.0] .40 Presenteeism 5.5 (0.6) [15.2] 2.6 (0.8) [7.1] 4.8 (0.9) [13.6] 7.6 (1.1) [20.5] , .01 a

Data given as No. (%) unless otherwise indicated. a Signifi cant at the .05 level.



at work despite asthma impairment. This will cause a ceiling effect for absenteeism, weakening its asso-ciation with asthma control. Unfortunately, we did not have data on whether individuals had used up their sick leave to test this hypothesis. In addition, absen-teeism is most likely due to an episode of exacerba-tion, while presenteeism is likely due to the presence of daily asthma impairment. The current methods of assessing asthma control are heavily based on mea-sures of impairment, and, as such, achieving clinical control according to such instruments will most likely result in minimal impairment, making presenteeism appear to be more responsive to asthma control than absenteeism.

To our knowledge, our study is the fi rst in asthma that has used productivity rate instead of wage rate to estimate the cost of lost work time, and has measured the preventable loss of productivity. Some of the pre-vious studies that measured productivity loss have used nonvalidated instruments. 9,16 For example, one relied on descriptive self-report of days absent from work or days of work with limited activity due to asthma. 16 Other studies have used patients’ disability claims to estimate indirect costs. 8 Further, many studies have not measured presenteeism, 6,8 which, as our results indicate, is the most important source of indirect costs. Some studies have only reported unadjusted levels of productivity loss across levels of asthma con-

trol or severity, but such results do not represent the preventable burden due to the confounding effect of other variables, as well as inevitability of residual pro-ductivity loss in patients with well-controlled asthma. Finally, estimates of productivity loss are sensitive to the instrument used. The WPAI questionnaire used to estimate the hours of lost work in this study tends to give a higher estimate of presenteeism compared with other measurement methods. 33 On the other hand, the proportion of our sample who reported pro-ductivity loss and the magnitude of the loss is in-line with other reports in other respiratory conditions, 34 as well as the productivity loss in the general North American population. 35,36

The limitations of our study should be acknowl-edged. None of the adults in the EBA study reported being unemployed because of asthma; thus, the poten-tial contribution of this aspect of the burden of asthma was not captured in our sample. Cost data often have large variations. As such, even with a sample size of 300 and prospectively collected data, our study might have been underpowered to capture rare events such as exacerbations (and, thus, absenteeism). Individuals’ asthma-related impairment might vary even within a given control level, and this residual variation might act as a confounding factor in relation with productivity loss. Finally, we have estimated the burden of produc-tivity loss associated with suboptimal asthma control,

Table 3— Results of the Regression Analysis of Productivity Loss on Control Level

Variable Absenteeism Presenteeism Overall Productivity Loss a

Partially controlled (vs controlled) Adjusted OR for reporting

productivity loss b 0.53 (0.21-1.32) [.17] 2.03 (1.02-4.05) [.04] c 1.48 (0.77-2.86) [.24]

Adjusted ratio of productivity loss among those who reported productivity loss d

0.59 (0.26-1.35) [.21] 1.24 (0.76-2.01) [.39] 0.98 (0.60-1.61) [.95]

Adjusted incremental effect on hours of productivity loss per week

2 1.31 ( 2 2.94-0.33) [.12] 1.93 ( 2 0.05-3.91) [.06] 0.62 (-2.37-3.61) [.68]

Adjusted incremental effect on productivity loss (CAD$2010) per week

2 $52.9 ( 2 $120.0-$14.3) [.12] $87.1 ( 2 $0.4-$174.6) [.05] $34.2 ( 2 $92.0-$160.5) [.60]

Uncontrolled (vs controlled) Adjusted OR for reporting

productivity loss b 1.30 (0.57-2.95) [.53] 3.41 (1.72-6.76) [ , .01] c 2.41 (1.25-4.64) [ , .01] c

Adjusted ratio of productivity loss among those who reported productivity loss d

0.82 (0.41-1.62) [.57] 1.49 (0.94-2.37) [.09] 1.31 (0.82-2.09) [.26]

Adjusted incremental effect on hours of productivity loss per week

0.42 ( 2 1.55-2.40) [.67] 3.68 (1.67-5.69) [ , .01] c 4.10 (0.83-7.37) [ , .01] c

Adjusted incremental effect on productivity loss (CAD$2010) per week

$17.4 ( 2 $62.9-$97.7) [.67] $167.4 ($76.6-$258.1) [ , .01] c $184.8 ($42.2-$327.4) [.01] c

Data in parentheses are 95% CIs ; those in brackets are P values. Monetary values are given in 2010 CAD. See Table 1 legend for expansion of abbreviation. a Adjusted incremental effects for overall productivity loss were estimated by summing the individual-level predicted values for absenteeism and presenteeism. The OR of reporting productivity loss and the ratio of productivity loss among those who reported productivity loss were estimated from fi tting a two-part model with the overall productivity loss as the dependent variable. b From the fi rst part of the two-part model: a logistic regression with the occurrence of productivity loss as the dependent variable. c Signifi cant at the .05 level. d From the second part of the two-part model: a generalized linear model with hours of work lost as the dependent variable, with Gamma distribu-tion and logarithmic link function.




and our results can be used to evaluate the benefi ts of policies aimed at improving asthma management. However, lack of a reference nonasthma group has precluded us from assessing the burden of productivity loss due to asthma itself; as such, our results cannot be used to inform preventive policies (eg, controlling environmental risk factors) aimed at reducing the incidence of asthma.

Productivity loss is an underappreciated source of economic loss. In other chronic diseases, consider-ation of productivity loss in policy analyses has resulted in profound changes in the choice of therapies. 37,38 Even when studies consider productivity loss, the emphasis has been based on absenteeism. 39 As our results suggest, presenteeism can be more responsive to change in the level of control than absenteeism, thus being a more important source of preventable burden. Future cost-effectiveness and burden-of-disease studies should put more emphasis on produc-tivity loss, and future studies evaluating productivity loss should use more rigorous and standardized methodology.

Acknowledgments Author contributions: Drs Sadatdafavi and FitzGerald had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Dr Sadatsafavi: contributed to the study concept, design, and pro-tocol and the statistical analysis; wrote the fi rst draft of the manu-script; approved the manuscript; and served as principal author. Ms Rousseau: contributed to study design and approved the manuscript. Ms Chen: contributed to the statistical analysis and approved the manuscript. Dr Zhang: contributed to approval of the manuscript. Dr Lynd: contributed to the study protocol and approved the manuscript. Dr FitzGerald: contributed to the study concept, design, and pro-tocol and approved the manuscript. Financial/nonfi nancial disclosures : The authors have reported to CHEST the following confl icts of interest: Dr Sadatsafavi receives salary support from the National Sanitarium Association. Dr Lynd has received other money from the Canadian Institutes of Health Research, BC Pharmacy Association, and Canadian Foundation for Pharmacy; unrestricted research grants from Genzyme Canada Inc and AbbVie Inc; and has provided methodology consultation to Pfi zer Canada Inc and Sanofi Aventis Canada. Dr FitzGerald has served on advisory boards for GlaxoSmithKline plc, AstraZeneca plc, Novartis AG, Pfi zer Inc, Boehringer-Ingelheim GmbH, Altana AG, Merck & Co Inc, and Topigen Pharmaceutiques Inc ; has been a mem-ber of speakers’ bureaus for GlaxoSmithKline plc, AstraZeneca plc, Boehringer-Ingelheim GmbH, Pfi zer Inc, and Merck & Co Inc; and has received research funding paid directly to the Univer-sity of British Columbia from the Canadian Institutes of Health Research, AstraZeneca plc, GlaxoSmithKline plc, Boehringer-Ingelheim GmbH, Merck & Co Inc, Pfi zer Canada, Bayer Schering Pharma, Genentech Inc, and Topigen Pharmaceutiques Inc. The remaining authors have reported to CHEST that no potential confl icts of interest exist with any companies/organizations whose products or services may be discussed in this article. Role of sponsors : The sponsor had no role in the design of the study, the collection and analysis of the data, or the preparation of the manuscript. Additional information: The e-Appendix and e-Tables can be found in the “Supplemental Materials” area of the online article.

References 1 . Ernst R . Indirect costs and cost-effectiveness analysis . Value

Health . 2006 ; 9 ( 4 ): 253 - 261 . 2 . Weiss KB , Gergen PJ , Hodgson TA . An economic evaluation

of asthma in the United States . N Engl J Med . 1992 ; 326 ( 13 ): 862 - 866 .

3 . Accordini S , Corsico AG , Braggion M , et al . The cost of per-sistent asthma in Europe: an international population-based study in adults . Int Arch Allergy Immunol . 2013 ; 160 ( 1 ): 93 - 101 .

4 . Awadh Behbehani N , Grunfeld A , FitzGerald JM . Health care costs associated with acute asthma: a prospective economic analysis . Can Respir J . 1999 ; 6 ( 6 ): 521 - 525 .

5 . Birnbaum HG , Berger WE , Greenberg PE , et al . Direct and indirect costs of asthma to an employer . J Allergy Clin Immunol . 2002 ; 109 ( 2 ): 264 - 270 .

6 . Chew FT , Goh DY , Lee BW . The economic cost of asthma in Singapore . Aust N Z J Med . 1999 ; 29 ( 2 ): 228 - 233 .

7 . Cisternas MG , Blanc PD , Yen IH , et al . A comprehensive study of the direct and indirect costs of adult asthma . J Allergy Clin Immunol . 2003 ; 111 ( 6 ): 1212 - 1218 .

8 . Colice G , Wu EQ , Birnbaum H , Daher M , Marynchenko MB , Varghese S . Healthcare and workloss costs associated with patients with persistent asthma in a privately insured popula-tion . J Occup Environ Med . 2006 ; 48 ( 8 ): 794 - 802 .

9 . Godard P , Chanez P , Siraudin L , Nicoloyannis N , Duru G . Costs of asthma are correlated with severity: a 1-yr prospective study . Eur Respir J . 2002 ; 19 ( 1 ): 61 - 67 .

10 . Herjavecz I , Nagy GB , Gyurkovits K , et al . Cost, morbidity, and control of asthma in Hungary: the Hunair Study . J Asthma . 2003 ; 40 ( 6 ): 673 - 681 .

11 . Jansson S-A , Rönmark E , Forsberg B , Löfgren C , Lindberg A , Lundbäck B . The economic consequences of asthma among adults in Sweden . Respir Med . 2007 ; 101 ( 11 ): 2263 - 2270 .

12 . Serra-Batlles J , Plaza V , Morejón E , Comella A , Brugués J . Costs of asthma according to the degree of severity . Eur Respir J . 1998 ; 12 ( 6 ): 1322 - 1326 .

13 . Smith DH , Malone DC , Lawson KA , Okamoto LJ , Battista C , Saunders WB . A national estimate of the economic costs of asthma . Am J Respir Crit Care Med . 1997 ; 156 ( 3 pt 1 ): 787 - 793 .

14 . Stock S , Redaelli M , Luengen M , Wendland G , Civello D , Lauterbach KW . Asthma: prevalence and cost of illness . Eur Respir J . 2005 ; 25 ( 1 ): 47 - 53 .

15 . Szucs TD , Anderhub H , Rutishauser M . The economic bur-den of asthma: direct and indirect costs in Switzerland . Eur Respir J . 1999 ; 13 ( 2 ): 281 - 286 .

16 . Ungar WJ , Coyte PC , Chapman KR , MacKeigan L ; Pharmacy Medication Monitoring Program Advisory Board . The patient level cost of asthma in adults in south central Ontario . Can Respir J . 1998 ; 5 ( 6 ): 463 - 471 .

17 . Sullivan PW , Ghushchyan VH , Slejko JF , Belozeroff V , Globe DR , Lin S-L . The burden of adult asthma in the United States: evidence from the Medical Expenditure Panel Survey . J Allergy Clin Immunol . 2011 ; 127 ( 2 ): 363 - 369.e1-3 .

18 . Shenolikar R , Song X , Anderson JA , Chu BC , Cantrell CR . Costs of asthma among US working adults . Am J Manag Care . 2011 ; 17 ( 6 ): 409 - 416 .

19 . Doz M , Chouaid C , Com-Ruelle L , et al . The association between asthma control, health care costs, and quality of life in France and Spain . BMC Pulm Med . 2013 ; 13 : 15 .

20 . BCStats. Open data from the 2011 census, British Columbia. BCStats website. http://www.bcstats.gov.bc.ca/StatisticsBySubject/Census/OpenData.aspx . Accessed January 14, 2013 .

21 . Human Resources and Skills Development Canada. National Occupational Classifi cation 2006. Human Resources and Skills Development Canada website. http://www30.hrsdc.gc.ca/NOC/english/NOC/2006/Welcome.aspx . Accessed January 15, 2013.


http://www.bcstats.gov.bc.ca/StatisticsBySubject/Census/OpenData.aspx

http://www30.hrsdc.gc.ca/NOC/english/NOC/2006/Welcome.aspx


sum-som/l01/cst01/famil108a-eng.htm . Accessed February 10, 2013.

31 . Statistics Canada. Population, urban and rural, by province and territory (British Columbia). 2009. Government of Canada website. http://www.statcan.gc.ca/tables-tableaux/sum-som/l01/cst01/demo62k-eng.htm . Accessed January 28, 2013.

32 . Bedouch P , Marra CA , FitzGerald JM , Lynd LD , Sadatsafavi M . Trends in asthma-related direct medical costs from 2002 to 2007 in British Columbia, Canada: a population based-cohort study . PLoS One . 2012 ; 7 ( 12 ): e50949 .

33 . Zhang W , Gignac MAM , Beaton D , Tang K , Anis AH ; Canadian Arthritis Network Work Productivity Group . Productivity loss due to presenteeism among patients with arthritis: estimates from 4 instruments . J Rheumatol . 2010 ; 37 ( 9 ): 1805 - 1814 .

34 . DiBonaventura MD , Paulose-Ram R , Su J , et al . The impact of COPD on quality of life, productivity loss, and resource use among the elderly United States workforce . COPD . 2012 ; 9 ( 1 ): 46 - 57 .

35 . Mitchell RJ , Bates P . Measuring health-related productivity loss . Popul Health Manag . 2011 ; 14 ( 2 ): 93 - 98 .

36 . Wang J , Schmitz N , Smailes E , Sareen J , Patten S . Workplace characteristics, depression, and health-related presenteeism in a general population sample . J Occup Environ Med . 2010 ; 52 ( 8 ): 836 - 842 .

37 . van den Hout WB , Goekoop-Ruiterman YPM , Allaart CF , et al . Cost-utility analysis of treatment strategies in patients with recent-onset rheumatoid arthritis . Arthritis Rheum . 2009 ; 61 ( 3 ): 291 - 299 .

38 . Grover SA , Ho V , Lavoie F , Coupal L , Zowall H , Pilote L . The importance of indirect costs in primary cardiovascular disease prevention: can we save lives and money with statins? Arch Intern Med . 2003 ; 163 ( 3 ): 333 - 339 .

39 . Johns G . Presenteeism in the workplace: a review and research agenda . J Organiz Behav . 2010 ; 31 ( 4 ): 519 - 542 .

22 . Statistics Canada. Labour force survey estimates (LFS), wages of employees by type of work, National Occupational Classi fi -ca tion for Statistics (NOC-S), sex and age group. Government of Canada website. http://www5.statcan.gc.ca/cansim/a26?lang 5 eng&retrLang 5 eng&id 5 2820070&paSer 5 &pattern 5 &stByVal 5 1&p1 5 1&p2 5 38&tabMode 5 dataTable&csid . Accessed January 14, 2013.

23 . Andreasson W , Svensson K , Berggren F . The validity of the work productivity and activity impairment questionnaire for patients with asthma (WPAIasthma): results from a web-based study . Value Health . 2003 ; 6 ( 6 ): 780 .

24 . Zhang W , Bansback N , Boonen A , Severens JL , Anis AH . Development of a composite questionnaire, the valuation of lost productivity, to value productivity losses: application in rheumatoid arthritis . Value Health . 2012 ; 15 ( 1 ): 46 - 54 .

25 . Pauly MV , Nicholson S , Polsky D , Berger ML , Sharda C . Valuing reductions in on-the-job illness: ‘ presenteeism’ from mana-gerial and economic perspectives . Health Econ . 2008 ; 17 ( 4 ): 469 - 485 .

26 . Global Initiative for Asthma (GINA) . Global strategy for asthma management and prevention. GINA website . http://www.ginasthma.org/local/uploads/fi les/GINA_Report_2010_1.pdf . Accessed December 4, 2013.

27 . Hankinson JL , Odencrantz JR , Fedan KB . Spirometric refer-ence values from a sample of the general U.S. population . Am J Respir Crit Care Med . 1999 ; 159 ( 1 ): 179 - 187 .

28 . Mihaylova B , Briggs A , O’Hagan A , Thompson SG . Review of statistical methods for analysing healthcare resources and costs . Health Econ . 2011 ; 20 ( 8 ): 897 - 916 .

29 . Schafer JL . Multiple imputation: a primer . Stat Methods Med Res . 1999 ; 8 ( 1 ): 3 - 15 .

30 . Statistics Canada. Median total income, by family type, by province and territory (All census families). 2012. Government of Canada website. http://www.statcan.gc.ca/tables-tableaux/



http://www.statcan.gc.ca/tables-tableaux/sum-som/l01/cst01/demo62k-eng.htm

http://www5.statcan.gc.ca/cansim/a26?lang=eng&retrLang=eng&id=2820070&paSer=&pattern=&stByVal=1&p1=1&p2=38&tabMode=dataTable&csid

http://www.ginasthma.org/local/uploads/files/GINA_Report_2010_1.pdf

http://www.statcan.gc.ca/tables-tableaux/sum-som/l01/cst01/famil108a-eng.htm

http://www.statcan.gc.ca/tables-tableaux/sum-som/l01/cst01/famil108a-eng.htm

http://www.statcan.gc.ca/tables-tableaux/sum-som/l01/cst01/demo62k-eng.htm

!

Online supplements are not copyedited prior to posting.

©2014 American College of Chest Physicians. Reproduction of this article is prohibited without written permission from the American College of Chest Physicians. See online for more details. DOI: 10.1378/chest.13-1619

CHEST 2014; 145(4):787-793

!!!

The Preventable Burden of Productivity Loss Due to Suboptimal Asthma Control A Population -Based Study Mohsen Sadatsafavi, MD, PhD; Roxanne Rousseau, BSc; Wenjia Chen, MSc; Wei Zhang, PhD; Larry Lynd, PhD; J. Mark FitzGerald, MD; and the Economic Burden of Asthma Study Team*

e-Appendix 1. The Economic Burden of Asthma Study Team: Satvir Dhoot; Lisa Dinh; Jennie Chan; Jesmin

Dhillon; Gurleen Gill; Jessika Iwanski; Zaakir Jiwa; Intan Agoes; Richie Li; Jordan Deppiesse; Samantha Gray;

Elena Terekhova; Nicole Brunton; Dayna Taylor; Madeline Ludwig; Laura FitzGerald; Douglass Rolfe; Wan Tan

Hogg

!!!!!!!!!


!



CHEST 2014; 145(4):787-793

!e-Table 1: Regression coefficients for the main analysis (absenteeism) Variable Coefficient

(logistic part) Coefficient (generalized linear model part)

Incremental value (hours)

Incremental value ($)

Partially controlled (versus controlled)

0.53 (0.21 , 1.32) P=0.17

0.59 (0.26 , 1.35) P=0.21

-1.31 (-2.94 , 0.33) P=0.12

-$52.9 (-$120.0 , $14.3) P=0.12

Uncontrolled (versus controlled)

1.30 (0.57 , 2.95) P=0.53

0.82 (0.41 , 1.62) P=0.57

0.42 (-1.55 , 2.40) P=0.67

$17.4 (-$62.9 , $97.7) P=0.67

Age at baseline 1.53 (0.67 , 3.46)

P=0.31 0.51 (0.23 , 1.12)

P=0.09 0.61 (-1.00 , 2.21)

P=0.46 $25.9 (-$40.9 , $92.7) P=0.45

Sex (female versus male)

1.23 (0.61 , 2.49) P=0.57

0.36 (0.19 , 0.71) P=<0.01*

-0.05 (-1.49 , 1.38) P=0.94

-$7.1 (-$61.6 , $47.3) P=0.80

Income (high versus low)

0.61 (0.30 , 1.21) P=0.16

1.00 (0.57 , 1.74) P=0.99

-1.16 (-2.95 , 0.62) P=0.20

-$39.0 (-$111.7 , $33.8) P=0.29

Education (high versus low)

1.34 (0.57 , 3.18) P=0.50

0.71 (0.33 , 1.55) P=0.39

0.66 (-1.05 , 2.37) P=0.22

$29.9 (-$38.2 , $98.0) P=0.39

Foreign born (versus Canadian born)

0.83 (0.41 , 1.70) P=0.62

0.48 (0.24 , 0.96) P=0.04*

-0.52 (-1.84 , 0.80) P=0.44

-$24.9 (-$77.0 , $27.3) P=0.35

Residence type (rural versus urban)

0.69 (0.19 , 2.49) P=0.57

0.89 (0.24 , 3.24) P=0.85

-0.71 (-3.00 , 1.58) P=0.54

-$30.5 (-$123.5 , $62.5) P=0.52

Drug insurance coverage (partial versus none)

1.86 (0.72 , 4.83) P=0.20

0.96 (0.42 , 2.20) P=0.92

1.23 (-1.01 , 3.46) P=0.28

$51.6 (-$39.2 , $142.3) P=0.27

Drug insurance coverage (complete versus none)

1.27 (0.41 , 3.93) P=0.68

1.43 (0.53 , 3.85) P=0.48

0.80 (-2.10 , 3.70) P=0.59

$35.1 (-$91.5 , $161.8) P=0.59

*Significant at 0.05 level

!!!!!!!!


!



CHEST 2014; 145(4):787-793

!e-Table 2: Regression coefficients for the main analysis (presenteeism) Variable Coefficient





2.03 (1.02 , 4.05) P=0.04*

1.24 (0.76 , 2.01) P=0.39

1.93 (-0.05 , 3.91) P=0.06

$87.1 (-$0.4 , $174.6) P=0.05


3.41 (1.72 , 6.76) P<0.01*

1.49 (0.94 , 2.37) P=0.09

3.68 (1.67 , 5.69) P<0.01*

$167.4 ($76.6 , $258.1) P<0.01*

Age at baseline 0.99 (0.56 , 1.77)

P=0.98 0.94 (0.65 , 1.35)

P=0.72 -0.07 (-1.82 , 1.68)

P=0.94 $4.7 (-$74.5 , $83.8) P=0.91


1.27 (0.76 , 2.13) P=0.35

1.02 (0.73 , 1.42) P=0.93

0.59 (-0.96 , 2.14) P=0.45

$14.1 (-$58.7 , $87.0) P=0.70


0.76 (0.43 , 1.35) P=0.35

0.93 (0.66 , 1.30) P=0.67

-0.95 (-2.48 , 0.58) P=0.22

-$16.4 (-$88.6 , $55.8) P=0.66


0.75 (0.40 , 1.44) P=0.39

0.94 (0.64 , 1.37) P=0.74

-0.62 (-2.31 , 1.07) P=0.47

-$14.4 (-$91.4 , $62.6) P=0.71


0.80 (0.47 , 1.36) P=0.41

0.97 (0.69 , 1.37) P=0.86

-0.69 (-2.23 , 0.86) P=0.38

-$41.9 (-$116.0 , $32.2) P=0.27


0.40 (0.15 , 1.02) P=0.06

0.56 (0.28 , 1.11) P=0.10

-2.59 (-4.91 , -0.28) P=0.03*

-$120.8 (-$224.7 , -$16.9) P=0.02*


1.09 (0.57 , 2.09) P=0.80

0.65 (0.43 , 0.98) P=0.04*

-0.15 (-1.99 , 1.69) P=0.87

$7.1 (-$77.7 , $91.9) P=0.87


1.15 (0.53 , 2.49) P=0.73

0.76 (0.48 , 1.23) P=0.27

0.17 (-1.94 , 2.28) P=0.87

$15.2 (-$85.0 , $115.5) P=0.77


!!!!!!!


!



CHEST 2014; 145(4):787-793

!e-Table 3: Regression coefficients for the main analysis (overall productivity loss) Variable Coefficient





1.48 (0.77 , 2.86) P=0.24

0.98 (0.60 , 1.61) P=0.95

0.62 (-2.37 , 3.61) P=0.68

$34.2 (-$92.0 , $160.5) P=0.60


2.41 (1.25 , 4.64) P<0.01*

1.31 (0.82 , 2.09) P=0.26

4.10 (0.83 , 7.37) P=0.01*

$184.8 ($42.2 , $327.4) P=0.01*

Age at baseline 1.13 (0.64 , 2.01)

P=0.67 0.86 (0.58 , 1.29)

P=0.48 0.54 (-2.30 , 3.37)

P=0.71 $30.6 (-$93.9 , $155.1) P=0.63


1.20 (0.72 , 1.98) P=0.49

0.90 (0.63 , 1.29) P=0.58

0.54 (-1.93 , 3.01) P=0.67

$7.0 (-$101.0 , $115.0) P=0.90


0.76 (0.43 , 1.34) P=0.35

0.87 (0.60 , 1.26) P=0.46

-2.11 (-4.92 , 0.70) P=0.14

-$55.4 (-$178.6 , $67.8) P=0.38


0.62 (0.33 , 1.18) P=0.15

0.92 (0.61 , 1.39) P=0.70

0.04 (-2.59 , 2.68) P=0.97

$15.5 (-$98.7 , $129.7) P=0.79


0.94 (0.56 , 1.59) P=0.81

0.84 (0.58 , 1.22) P=0.36

-1.21 (-3.53 , 1.12) P=0.31

-$66.8 (-$172.0 , $38.4) P=0.21


0.45 (0.18 , 1.13) P=0.09

0.69 (0.34 , 1.43) P=0.32

-3.31 (-7.18 , 0.57) P=0.09

-$151.3 (-$318.7 , $16.1) P=0.08


1.20 (0.63 , 2.29) P=0.58

0.81 (0.51 , 1.28) P=0.36

1.08 (-2.50 , 4.65) P=0.55

$58.7 (-$94.1 , $211.4) P=0.45


1.10 (0.51 , 2.37) P=0.80

0.93 (0.55 , 1.59) P=0.80

0.97 (-3.36 , 5.31) P=0.66

$50.4 (-$144.9 , $245.6) P=0.61


!!!!!!!


!



CHEST 2014; 145(4):787-793

e-Table 4: Results of alternative model specification (with main model also reported for comparison) and sensitivity analyses Scenario absenteeism presenteeism Overall productivity loss

1) Main model As described in the text Partially controlled vs.

controlled -$52.9 (-$120.0 , $14.3)

P=0.12 $87.1 (-$0.4 , $174.6)

P=0.05 $34.2 (-$92.0 , $160.5)

P=0.60 Uncontrolled vs. controlled $17.4 (-$62.9 , $97.7)

P=0.67 $167.4 ($76.6 , $258.1)

P<0.01* $184.8 ($42.2 , $327.4)

P=0.01* 2) Alternative model Two-part model regression directly performed on the cost of productivity loss

Partially controlled vs. controlled

-$49.4 (-$106.7 , $7.9) P=0.09

$72.3 ($11.7 , $133.0) P=0.02*

$22.9 (-$49.6 , $95.4) P=0.54

Un controlled vs. controlled $19.1 (-$61.7 , $99.9) P=0.64

$165.7 ($76.0 , $255.5) P<0.01*

$184.8 ($43.8 , $325.9) P=0.01*

3) Asthma status responses uncertain as missing value

"Don't know" responses to the presence of asthma symptoms (used to assess asthma control) were taken as the lack of symptom in the main analysis. Here they are treated as missing value and are imputed using multiple imputation methods.


-$49.6 (-$106.5 , $7.4) P=0.09

$71.4 ($11.6 , $131.3) P=0.02*

$21.9 (-$50.6 , $94.4) P=0.55

Uncontrolled vs. controlled $19.5 (-$59.5 , $98.6) P=0.63

$164.8 ($76.1 , $253.6) P<0.01*

$184.4 ($45.1 , $323.6) P<0.01*

4) Extreme VOLP multipliers removed

In this sensitivity analysis, individuals with VOLP multipliers more than or equal to 5 were excluded from the analysis. There were a total 13 individuals with such values.


-$45.3 (-$97.7 , $7.1) P=0.09

$63.8 ($8.3 , $119.3) P=0.02*

$18.5 (-$50.0 , $86.9) P=0.60

Uncontrolled vs. controlled $17.4 (-$57.9 , $92.7) P=0.65

$145.5 ($70.5 , $220.6) P<0.01*

$163.0 ($40.1 , $285.9) P<0.01*

* significant at 0.05 level Inference on models 2,3, and 4 is based on parametric bootstrap with 20 replications. S


the preventable burden of productivity loss due to suboptimal asthma control

Documents