Time and Health Status in Health Economics

…consumers demand health, defined broadly to include llness-free days in a given year and life expectancy…

Grossman, 2016

Introduction

Much empirical health economics research depends centrally on the measurement of individuals’ health. Less obvious perhaps is the level of effort dedicated in this research to appropriate conceptualization of health status and to the measurement of these concepts. Ultimately the question arises: Are the data on health status on which much of our empirical work depends worthy of the elegant models and econometric methods used to understand them? This editorial was prompted by some current work in this area (Burns and Mullahy, 2016), as well as by revisiting important decades-old work by Manning, Newhouse, and Ware (1982; henceforth “MNW”), who devoted considerable thoughtful attention to the importance of and strategies for careful measurement of health status in the context of empirical investigations in health economics. True to the full title of MNW’s seminal chapter, important health status measurement considerations arise far “beyond excellent, good, fair, and poor.”

It seems obvious that health economists ought to care about the measurement of health in their empirical analyses. At least two reasons are readily apparent. The first is intrinsic interest for scientific, policy, regulatory, etc. purposes in which the focus is on understanding health as an outcome arising from treatments or other factors or, conversely, as a predictor of other economic phenomena of interest (e.g. labor market outcomes). The second reason is an instrumental one: For scientific, policy, regulatory, etc. purposes, health outcomes play a central role in evaluation (cost-effectiveness analysis, cost-benefit analysis) criteria, e.g. incremental cost-effectiveness ratios, net health benefit criteria, etc. In these or other imaginable contexts, thoughtful conceptualization and measurement of what is meant by “health” should lead to better understanding of and/or better decisionmaking regarding the questions at hand. In the clinical sciences concern is increasing about the use of so-called surrogate measures (e.g. biomarkers) in evaluations of interventions for the reason that they are unlikely to capture the kinds of health endpoints that matter to stakeholders like how they feel, how they function, and whether they survive. Analogous concerns should be raised in health economics about how well our health status measures describe the features of health that matter to decisionmakers.

This editorial suggests the importance of health status measurement in one particular area of interest in applied health economics research: the temporal dimensions of health status and its measurement. Health necessarily plays out in time, but the manner in which it does so often goes unnoticed or underappreciated measuring health status for purposes of analysis. After providing a brief conceptual context, I describe several measures of health status in common usage in health economics and note whether they are built on firm conceptual foundations that respect health’s temporal dimensions.

Health Measurement in Time

Many discussions of health in health economics begin with Grossman’s, 1972, human capital model of the demand for health. One feature of Grossman’s framework is its careful distinction between health capital and health status. Health status is a flow that in Grossman’s framework is at time “t” proportional to accumulated health capital stocks, ${\mathrm{h}}_{\mathrm{t}}={\mathrm{\varphi }}_{\mathrm{t}}{\mathrm{H}}_{\mathrm{t}}$ (analogous to how income flows, e.g., derive from wealth stocks). As a flow, health status is necessarily defined by some time frame or time dimension. In Grossman’s framework, this notion translates into considering poor health or sickness as a particular type of time use within an individual’s overall time budget $\mathrm{T}{\mathrm{H}}_{\mathrm{t}}+\mathrm{T}{\mathrm{L}}_{\mathrm{t}}+\mathrm{T}{\mathrm{W}}_{\mathrm{t}}+{\mathrm{T}}_{\mathrm{t}}=\mathrm{\Omega }$, wherein TW is labor supply, T is time dedicated to productive activities apart from human capital investment, and TL is unhealthy time defined by $\mathrm{T}{\mathrm{L}}_{\mathrm{t}}=\mathrm{\Omega }-{\mathrm{h}}_{\mathrm{t}}$. Grossman’s characterization of health status is fundamentally important when it comes to understanding the roles of health in some domains of economic analysis.

Kindred to Grossman’s framework, and featuring prominently in the technology assessment literature, is a class of health-related quality-of-life measure in which lifespan, life years, or life expectancy provides the essential time dimension. Measures like quality-adjusted life years (QALYs) and healthy-year equivalents (HYEs) are designed to characterize how health status is manifested over time. While technology assessment and human capital applications are often directed towards different audiences, they both rely on sound characterizations and measurements of health and, in general, the manner in which health evolves over some relevant time dimension(s).

Time and Health Status Measures in Empirical Health Economics

Health economists frequently encounter data on concepts like work-loss days, restricted-activity days, bed-disability days, and school-loss days, often reported over a tightly defined two-week recall or lookback period. In other contexts, health economists will often consider data on chronic illness obtained from survey questions like: “Has a healthcare provider ever told you have ______?” Notwithstanding considerations of recall bias and access bias, such chronic-illness data are also based on well defined time windows – birth to current age, in principle – even though they are uninformative about the timepoints of onsets and possible remissions of episodes of such illnesses.

While it may or may not be important to draw a distinction between health capital and healthy time (i.e. “health status”) in a particular application, some other measures of health elicited via familiar survey questions will often be difficult to interpret because of their lack of correspondence to either health capital or healthy time. For instance, compare a survey question: “In general, would you say your health is excellent, very good, good, fair, or poor?” with an alternative: “Over the past week, would you say your health is excellent, very good, good, fair, or poor?” The former version – whose variants are in common use in many surveys across the globe – provides no time-frame anchor within which respondents can interpret unambiguously what is meant by “in general”. As such, it is challenging to interpret a response to such a question as a measure of healthy time or a health flow. Indeed, such a survey item may – depending on respondents’ framing – capture respondents’ senses of their health capital as opposed to their “instantaneous” healthiness.

In reference to the structure of the RAND Health Insurance Study from which their work evolved, MNW offer compelling arguments for why many health-status-related survey items will have firmer conceptual and empirical foundations when cast within particular timeframes:

To enhance measurement reliability, the HIS fielded unambiguous questionnaire responses, such as “My health is now excellent” as opposed to “Health is good”; the latter response could refer to either the value placed on health or to the goodness of one’s health, and its time frame is ambiguous. Multiple-response, as opposed to dichotomous, choices further improved reliability. For example, respondents were asked: “During the past month, how much of the time have you felt depressed?” Six choices, ranging from “All of the time” to “None of the time,” were offered, as opposed to asking “During the past month, have you been depressed?” with responses of “Yes” and “No.” (MNW, 1982, p. 148)

Healthy-Time Health Status Measures

Beyond time-denominated health status measures, other metrics are defined with time units themselves as the outcomes. In keeping with Grossman’s time-budget framework, the notion that healthiness is a flow corresponding to some form of time use or time allocation within a given time budget or time frame is both conceptually as well as intuitively appealing. For instance, two measures of health status that have been used as summary measures of health-related quality of life – used in, e.g., the U.S. Behavioral Risk Factors Surveillance System and Medicare Health Outcomes Survey – measure physically and mentally healthy time in the month preceding the survey (e.g. “Now, thinking about your physical health, which includes physical illness and injury, for how many days during the past 30 was your physical health not good?”).

Also noteworthy are innovations in what might be termed utilization-based time allocation measures of health outcomes. Such measures include Days Alive out of Hospital (ESCAPE, 2005), Healthy Days at Home (MedPAC, 2015), Contact Days (Dartmouth Institute, 2016), and Days Residing in the Community (Mathematica Policy Research, 2016). For all such measures, the ostensible goal is to utilize existing data on healthcare utilization and mortality to characterize outcomes that are likely to be considered important by patients and other stakeholders, and ultimately to understand the factors – treatments, quality initiatives, provider incentives, etc. – that produce the observed time-allocation outcomes. The fundamental premise is that a given day within the accounting period has positive value if the individual is alive and not in contact with the healthcare system on that day (a premise that clearly merits scrutiny in some contexts). Burns and Mullahy, 2016, provide additional discussion.

Conclusions

High-impact empirical research in health economics depends on taking seriously the measurement of health. Yet at a general level health measurement issues are far from resolved. Among other important areas of application, measurement of health has become increasingly important in an era of patient-centered clinical care and research, patient-reported outcomes, and value-based health care delivery (Lynn et al., 2015; Porter et al., 2015; U.S. FDA, 2009).

Understanding the timeframes over which health status is measured is conceptually logical and empirically important. In econometric analyses that strive to understand health either as an outcome or as a determinant of outcomes, the time contexts of such relationships – and the measures that describe them – should be central considerations. Issues like survey recall bias are real, but such considerations should not per se compromise the use of conceptually sound measures in empirical analysis, and bias-variance tradeoffs may be necessary to address in such cases. Ultimately, the policy or treatment question at hand ought to dictate how to measure health and the timeframes within which it should be characterized.