Statistics from Altmetric.com
In a fine recent editorial, Drs Melissa and Gary Brown raised issues at the nexus of health policy and clinical science.1 As utility assessment is relatively new to the visual sciences, understanding both the assumptions behind this work and the consequences of relaxing those assumptions is essential for the conduct of high quality research and appropriate interpretation of the results.
The use of community elicited utilities (that is, including people without the disease in the elicitation study) in economic evaluation should be given more than minimal consideration. Economic evaluations are intended to inform health policy makers by assessing the value society places on the cure or prevention of disease. Community based utilities typically reflect larger estimates of utility loss than those elicited from patients and result in a more favourable analysis of the cost effectiveness of preventive interventions than those relying on patient elicited utilities.2 At the same time, estimating community elicited utilities requires the development of easily understood scenarios to assist community members in understanding life with the disease,3 after leading investigators prefer to rely on patient elicited utilities. Rather than dismiss the community elicited approach, economic evaluation in ophthalmology would be greatly facilitated by development of a catalogue of community elicited utilities related to old disease developed through the standard gamble or time trade-off methods or responses to health status questionnaires that include algorithms to estimate health utilities.
While the Browns caution against the use of functionally based health related quality of life instruments (for example, the NEI-VFQ) in economic evaluation, we would like to offer an alternative explanation for this concern. Most disease specific instruments are based in psychometric theory and designed to measure change in the patient’s self reported health status in investigator defined domains.4 Domain scores do not reflect the importance the respondent assigns to the activities, but scoring algorithms developed by the instrument designer. The result is a metric that is often meaningful to clinicians but does not reflect the value the patient or society places on the health state. This limits generalisability across disease groups, as well as investigators’ ability to comment on the most efficient method to screen for, or treat, an ophthalmic condition affecting multiple areas of physical, mental, or emotional function.
Finally, the standard gamble elicitation method should not be dismissed off handedly. More frequent use of the time trade-off reflects the method’s intuitive appeal rather than theoretical superiority. As opposed to the time trade-off in which the anchor event (typically, death, blindness, etc) occurs in the future, in the standard gamble the event is immediate. This provides an estimate of the person’s risk preference unconfounded by time. The time trade-off consistently results in higher estimates of utility loss than the standard gamble,5,6 potentially resulting in an overestimation of the cost-effectiveness of treatment or prevention.
We hope that our comments will help future work to be pragmatic and theoretically sound. This is necessary if we are to properly characterise the appropriateness of our methods as well as the value of our findings.
We thank Drs Kymes and Frick for their excellent letter regarding utility analysis as a health related quality of life instrument. We agree that the use of primarily function based quality of life instruments such as the NEI-VFQ-25 may result in missing many important variables in the quality of life arena, as well as limit applicability across all diseases.1 In contrast, preference based quality of life instruments, such as utility analysis, are applicable across all diseases and encompass all variables that comprise quality of life, as well as the weighting of those variables. Of great additional importance is the fact that preference based instruments can be used in healthcare economic analyses, especially utility analysis, while most function based instruments have not been successfully used.1,2
Concerning the use of time trade-off and standard gamble utility analysis, we have found that the time trade-off methodology is easier for patients to comprehend and also is more sensitive to milder health states since there is risk aversion to the consequence of immediate death associated with the standard gamble variant.1,2 Froberg and Kane3 have also shown that the time trade-off method of utility has greater test-retest reliability, intra-rater reliability and inter-rater reliability than standard gamble methodology. In our experience, time trade-off utilities generally demonstrate better construct validity1 and a wider range between pre-intervention and post-intervention values than standard gamble utilities, thus resulting in more favourable cost utility analysis, rather then less favourable analyses.
With regard to quality of life respondents, we remain firm in our adherence to the fact that a basic pillar of value based medicine is the use of utility values obtained from respondents with a health state in question.1,2 We have found that utility value diminution in patients who actually have age related macular degeneration ranges from 103% to 750% greater than the decrement estimated by treating ophthalmologists for the same condition.4,5 This has been noted as well for non-ophthalmological health states.6
We respectfully disagree that community utility values generally overestimate the degree to which a disease decreases quality of life. In contrast, we and others4–9 have noted that community and provider participants asked to evaluate the quality of life associated with a health state using utility value analysis generally underestimate the decrement in quality of life compared to patients with that health state. In essence, patients who have lived with a health state are those best able to ascertain the quality of life associated with that health state. And it is usually worse than others imagine.
In conclusion, we thank Kymes and Frick for their interest and comments and look forward to additional awareness in the arena of value based medicine. As increasing numbers of those who allocate healthcare resources become aware that value based medicine allows for higher quality care (by incorporating quality of life parameters that evidence based primary clinical trials often ignore) and the most efficient use of resources, it will have a considerably greater role in the delivery of cost effective, quality healthcare. When that takes place, all will benefit.