In healthcare, it is well recognized that quality of life is important, and as a conceptual matter, health embodies quality as well as quantity of life.1-3 It is therefore generally accepted that quality of life merits consideration when payers and policy makers weigh the clinical and economic value of medical treatments. Such assessments typically focus on health states2; however, the concept of quality of life, and the impact of healthcare on quality of life, plausibly extend beyond conventional health states. There is growing evidence that patients care about nonclinical aspects of care. For example, many hospital patients elect to be treated relatively far from home, in part because distant hospitals offer better food, more attentive staff, or more pleasant surroundings.4,5 For specific treatments, it is likewise possible that patients value not only clinical efficacy and side effects, but the overall experience, potentially including the burden of administering the treatment, for instance.
 
The treatment of cystic fibrosis (CF) is a test case for this proposition. CF is the most common lethal inherited disease affecting individuals of Caucasian descent6 and is characterized as a multisystem disease that primarily results in chronic and progressive pulmonary dysfunction, nutritional abnormalities, and heightened inflammation secondary to chronic respiratory infections. Life expectancy with CF, while improving, remains substantially shortened, with a median predicted survival of 41 years.7 CF affects not only life expectancy, but also functional capabilities such as exercise capacity, and significant lung disease can lead to multiple chronic physical and psychosocial symptoms.8 As such, quality-of-life assessments of individuals with CF are increasingly useful adjuncts to standard clinical care pathways.9
 
Improvements in CF health outcomes have arisen in part because of the early introduction of chronic respiratory therapies to treat airway abnormalities and chronic infections. Many of these treatments require a significant time to complete, leading to high treatment burden for these individuals. Time may be especially valuable to individuals with shortened life expectancy, yet common treatments have been time-consuming and otherwise burdensome. Studies have estimated that adults with CF spend approximately 2 hours per day on routine treatments,10 and the treatment burden has been increasing over time11; moreover, this burden is also associated with a decline in quality of life.10 
 
For individuals with CF and chronic infection with Pseudomonas aeruginosa, alternate-month nebulized administration of tobramycin is a guideline-recommended standard of care.12 Chronic therapy with nebulized tobramycin has been associated with improvements in lung function and decreased mortality,13-16 but despite this clinical efficacy, adherence to inhaled tobramycin in the population is low.17 Multiple barriers to adherence have been studied in the CF population, and time for treatment administration is often cited. With a standard nebulizer model, delivery of this medication, together with setup and cleaning, takes 50 to 70 minutes daily. In addition, the equipment weighs in excess of 5 pounds and must be connected to a power source.
 
Recently, a new mode of administration for inhaled tobramycin has been introduced: an inhaler that delivers tobramycin as an inhaled dry powder.18 The efficacy and safety of this formulation have been demonstrated in clinical trials19-22; this treatment requires only about 10-15 minutes daily, and the inhaler is compact and lightweight.
 
This study assesses patient preferences for treatment modalities, focusing specifically on treatment administration and burden. We developed and conducted a survey with adult CF patients undergoing chronic respiratory treatment. The survey applied the established and validated time trade-off (TTO) methodology to treatments that differed in treatment burden. Survey results were translated into the benchmark quality-adjusted-life-year (QALY) metric for the distinct treatments.
 
METHODS
Survey Overview
We surveyed CF patients receiving care in the adult CF clinic at Boston Children’s Hospital between August and October 2013. Study participants were >18 years old, had previously been diagnosed with P aeruginosa lung infection, and had used nebulized antipseudomonal therapy; these criteria were verified with medical records. We reviewed the medical records of all study participants to document the most recent lung function (as measured by percent-predicted forced expiratory volume in 1 second [FEV1]).
 
Study participants were administered the Cystic Fibrosis Questionnaire-Revised (CFQ-R), a validated and commonly used health-related quality-of-life measure that assesses the impact of CF on 9 domains, including overall health perceptions and perceived treatment burden. The reliability coefficients on the domains (measured by Cronbach’s α) range from 0.18 to 0.94, with a majority exceeding the threshold of 0.70.9 The CFQ-R also elicits patient characteristics such as age, gender, and marital/partner status. We obtained data on whether a participant had any children via a review of clinic administrative records. Our survey also included a module that elicited family income.
 
Assessment of Treatments
To assess patient preferences for treatment modalities that plausibly differ in treatment burden, we used the TTO approach, which presents a respondent with 2 situations and asks him or her how many years of life expectancy he or she would be willing to give up to have his or her preferred situation. This approach has been widely used to compare health states, often for the purpose of measuring quality of life in the specification of QALYs.23 More recently, the method has been used to assess non-clinical aspects of treatment.24,25
 
We developed a TTO module for the survey. This module compared 2 treatments for respiratory lung infections caused by P aeruginosa. Therapy A corresponded to an unbranded nebulized therapy and therapy B corresponded to an unbranded dry powder inhaler. The treatments were contrasted in their burden of administration, vis-à-vis equipment used, setup and cleaning, medicine storage, and total treatment time per day; side effects were also compared. These treatment parameters were calibrated to real-world treatments based on existing clinical trial data and a literature review.18 The module specified that the treatments would be equally effective in treating infection.
 
Study participants were asked which treatment would be preferred if the out-of-pocket costs were the same. Next, each participant was asked whether he or she would choose to live a certain number of years with his or her preferred treatment, or a longer time (40 years) with the alternative treatment—in either case, these years represent remaining life expectancy. The “ping pong” method was used to specify life-years with the preferred treatment.26 If a participant chose (or alternatively, did not choose) the preferred treatment at a given life expectancy level, his or her life-years with the preferred treatment were decreased (or alternatively, increased) until he or she expressed indifference about the treatments and their life expectancies. Forty years of possible life expectancy was chosen to limit anxiety about mortality, and because this value has become more realistic for this population.7
 
To determine QALY values associated with each treatment, study participants were presented with a second TTO between a shorter life span without CF and 40 years with the disease while using their preferred treatment. In addition, the survey qualitatively assessed the 2 treatments by asking each study participant to agree or disagree with statements such as, “Therapy B is an improvement over therapy A.” Additional details are presented in the eAppendix (available at www.ajmc.com).
 
Data Analysis
We quantified the number of years of life (if any) that patients with CF would give up to have their preferred treatments instead of the alternative. We also quantified willingness to give up life-years to be in perfect health, instead of having CF and using the preferred treatment. We calculated pairwise correlations between the numbers of life-years willingly given up and patient characteristics, including age, gender, having a partner or child, family income, percent-predicted FEV1, current inhaled tobramycin use, and CFQ-R scores on 2 domains: health perceptions and treatment burden (each scale is scored 0-100). Because income was elicited in ranges, we created a continuous variable using midpoints for closed intervals (eg, $20,000 for $15,000-$24,999) and $200,000 for the highest value ($150,000 or more). We also performed multivariate linear regressions of the number of years that respondents would willingly give up for their preferred treatment on patient characteristics such as age. Income was excluded in sensitivity analysis because of occasional nonresponse. Statistical significance was assessed based on a 10% level.
 
RESULTS
We approached 100 adults with CF to participate in the study. Of these 100 patients, 75 were approached after the TTO module had been finalized. One patient declined to participate, and 1 did not complete the survey. Descriptive statistics for the 73 patients who completed the finalized TTO module are presented in Table 1. (There were no significant differences at a 10% level in measured characteristics between this sample of patients and those patients not in the TTO sample.) The average age in the TTO sample was 30.2 years (SD ± 11 years), and a majority of participants were female (58%). More than a third had a partner or child, and more than half reported an annual family income in excess of $50,000. In terms of patient health, the average FEV1 was 64.9% predicted. On the CFQ-R questionnaire, the average score on the health perceptions domain was 60.9, while the treatment burden score averaged 51.4.
 
Of the 73 patients, 65 (89%) would prefer therapy B (the dry powder inhaler) over therapy A (nebulized therapy) if their out-of-pocket costs were the same. The 8 remaining patients were indifferent about the treatments. Figure 1 shows the number of years with the dry powder inhaler that would be equivalent to 40 years with a nebulized therapy. Eleven percent of patients would have been equally happy with 40 years with a dry powder formulation or 40 years with a nebulized therapy, whereas another 36% viewed 39.5 years with a dry powder formulation as equivalent to 40 years with a nebulized therapy. Fifty-three percent would be as happy having the dry powder inhaler for 38.5 years or less as with nebulized therapy for 40 years. That is, these respondents would be willing to give up at least 1.5 years of life expectancy to have the dry powder formulation. On average, respondents viewed 34.7 years with a dry powder formulation as equivalent to 40 years with a nebulized therapy and would have been willing to give up 5.3 years of life expectancy to have the dry powder inhaler instead of nebulized therapy.
 
Table 2 shows how the number of years of life that patients would be willing to give up for the dry powder inhaler is related to their clinical characteristics and reported quality of life. No association was observed with age, gender, or living situation. The correlation coefficient between the number of years given up and the CFQ-R score for treatment burden is –0.221 (P = .060); because a higher score corresponds to a lower burden, respondents who reported a higher burden from CF treatment were willing to give up more life-years for the dry powder formulation than patients with a lower burden. This relationship remains significant (P = .052) in the multivariate regression that includes income: a 1-SD increase in treatment burden is associated with a 0.104-SD increase in the number of years a respondent would give up for the dry powder inhaler. In the larger analysis sample that excludes income, the association is somewhat smaller in magnitude and statistically insignificant. In 1 of the 2 specifications, FEV1 is negatively related (at a 10% significance level) to years given up. Specifically, a 1-SD decrease in lung function is associated with a 0.074-SD increase in years willingly given up for the dry powder formulation. In additional analysis (available from the authors upon request), individuals with an FEV1 above the sample median of 62 would be willing to give up an average of 3.9 years out of 40 years of life expectancy to have the dry powder formulation rather than nebulized therapy (the confidence around 3.9 years does not include zero); individuals with an FEV1 below the median would give up 6.7 years (P = .11 for the difference).
 
Figure 2 shows the number of years with perfect health (no CF) that survey respondents viewed as equivalent to 40 years living with CF while using the dry powder inhaler. Twenty-five percent of patients would have been equally happy with 39.5 years of perfect health, or 40 years with CF while using the dry powder inhaler; all respondents would have given up some life expectancy (1 half year or more) to not have CF. At the extreme, 5% of patients would give up 25 or more years of life to not have CF. On average, respondents viewed 31.9 years with perfect health as equivalent to 40 years with CF while using the dry powder inhaler, and would have been willing to give up 8.1 years to not have CF.
 
Table 3 shows how the number of years of life that patients would be willing to give up for perfect health is related to their clinical characteristics and quality of life. Age, gender, and living situation were not associated with this outcome. The correlation coefficient between the number of years willingly given up and the CFQ-R score for health perceptions is –0.278 (P = .025). That is, patients who reported lower overall health perceptions were willing to give up more life-years to not have CF than were patients who were healthier.
 
Applying the chained TTO approach, the average QALY value for having CF and using nebulized therapy (therapy A) was 0.704. The average QALY for having CF while using a dry powder inhaler (therapy B) was 0.797. These values exclude the 8 patients who were indifferent about the treatments and were not administered the second TTO. If we use multivariate regression to predict the second TTO value for these patients, the average QALY values are very similar (0.693 and 0.797, respectively).
 
In the qualitative assessment of CF treatments, 78% of patients in the TTO sample agreed with the statement, “Therapy B is an improvement over therapy A”; 52% strongly agreed with this statement. (Responses of “Don’t know/refuse” are treated as disagreement for this purpose.) Sixty-two percent agreed with the statement, “If I could use therapy B, my life would be better.” In terms of treatment attributes, 95% of respondents rated time spent on treatment as important; 90% rated ease of setup and cleaning as important; and 88% rated the portability of equipment and ability to administer treatment anywhere as important. 
 
DISCUSSION
In this study, we assessed how treatment burden affects quality of life in patients with CF, from the patient point of view. To do so, we surveyed adults with CF about their attitudes toward treatments that differed in burden of administration. We applied the established and validated TTO methodology and translated the results into the benchmark QALY metric for the distinct treatments.
 
The vast majority of survey participants would prefer a dry powder inhaler to nebulized therapy if their out-of-pocket costs were the same. In our survey, the key distinction between the dry powder inhaler and the nebulized therapy was that the inhaler took substantially less time out of the patient’s day and was otherwise less burdensome to administer. In terms of the TTO, the average adult with CF would willingly give up 5.3 years of life, out of a possible 40 years, to have the less burdensome treatment. In terms of QALYs, having CF and using the less burdensome treatment had an average value of 0.797, compared with an average value of 0.704 for the treatment that is more burdensome to administer. Thus, the less burdensome treatment mitigated one-third of the quality-of-life impact of having CF.
 
We also found that assessments of the treatments varied with patient characteristics in reasonable ways. For example, there was some evidence that patients who reported that their treatment burden was relatively high were willing to give up more life-years to have the less burdensome treatment than were patients with a low treatment burden. It is also noteworthy that nearly all patients deemed important the time spent on treatment, as well as ease of setup and cleaning, and the ability to administer treatment anywhere.
 
Altogether, our findings about patient preferences toward CF treatments are consistent with trial evidence of improved patient satisfaction with administration of tobramycin by dry powder inhaler instead of as a nebulized solution.21 It is noteworthy that comparable quality-of-life impacts have been found for a preferred mode of treatment (inhalation rather than injection) among patients with diabetes.25 Our findings raise the possibility that use of a dry powder might lead to improvement in adherence to chronic therapy, a critical issue in CF care.27-29
 
Limitations
Nevertheless, our study does have a number of limitations. We analyzed a sample of 73 individuals, using a 10% significance level; these individuals did not have direct experience with a dry powder inhaler at the time of the survey. Moreover, the generalizability of our findings to the broader population of US adults with CF is a natural concern. It is somewhat reassuring that the measured characteristics of our study sample are comparable to the average health characteristics of the adult CF population as reported by the US CF Foundation.7 From a methodological point of view, there is ongoing debate about the strengths and weaknesses of alternative approaches to eliciting patient utilities/valuations, and indeed debate about the implications of varied implementations of the TTO approach.23 Although QALY estimates with CF are not widely available, the agreement between our QALY value for CF with nebulized therapy and the QALY value for CF from a prior TTO assessment (0.70) lends some validity to our results.30 A comparable value (0.75) has also been used in cost-effectiveness analysis of neonatal screening for CF.31 This study has not addressed the value of reduced treatment burden in economic terms. It is also not clear whether our key finding—that the burden of administering the CF treatments studied matters a good deal to CF patients—applies broadly to other therapies and diseases.

Implications
The implications of this research are significant, as the importance of the patient perspective is increasingly being recognized in US healthcare. In its report entitled Crossing the Quality Chasm, the Institute of Medicine offered 6 criteria by which to judge the performance of the healthcare system.32 Among them, healthcare should be “patient centered,” that is, “respectful of and responsive to individual patient preferences, needs and values.” Consistent with this view, CMS has sponsored the development of survey instruments for patient experience of care, and publicly reports on the performance of providers and even pays for “value” partly on this basis.33 More recently, the Affordable Care Act established a Patient Centered Outcomes Research Institute, which funds research that “is inclusive of an individual’s preferences, autonomy and needs.”34
 
CONCLUSIONS

This study suggests that the burden of administering treatments is of concern to adults with CF, and that non-clinical aspects of treatments may be important more generally. Clinicians should assess patient preferences for treatment delivery systems, certainly when recommending chronic respiratory therapies for patients with CF, and potentially in other settings. Payers should be cognizant of the potential importance to patients of the burden that can be imposed in administering treatments, because coverage decisions may have competitive consequences in insurance markets. Finally, health technology assessment thought leaders and practitioners should seriously consider the possibility that non-clinical aspects of treatment play a role in these assessments.