Cost-Effective Spinal Surgery: Reality or Oxymoron?

Fig. 13.1

This is a typical graph seen in cost-effectiveness analysis literature comparing incremental costs versus incremental QALYs. The x-axis represents the health benefit in terms of QALYs. The y-axis represents the cost. After a CEA is conducted, if a technology is found to be beneficial from the perspective of QALYs and cost less than its comparator, it is said to be dominant (southeast quadrant). Likewise, a technology found to be less beneficial and more costly than its comparator is said to be dominated (northwest quadrant). Such results require only common sense to interpret. Most often, however, CEAs will report that a technology is more beneficial and more costly (northeast quadrant) or less beneficial and less costly (southwest quadrant). The interpretation of such results depends on the perspective of the interpreter. A technology that is marginally better in terms of utility at great cost may not be “cost-effective” if the more important aspect is cost control. On the other hand, a technology that is marginally worse than its comparator but at much less cost may be judged to be cost-effective if cost is the most important aspect of the decision process

The “northwest” corner of the graph is considered to be bad. This will also give a negative ICER and corresponds to a more expensive worse outcome. This situation does occur from time to time and illustrates the need for high-quality research to investigate “standard” practices. For example, the use of steroids in patients with severe head injury was studied and was found to both worsen outcome and accrue increased costs. The “southwest” corner reflects a less expensive but worse outcome. Again, whether this is a good or bad result can be subjected to interpretation.

At face value, CEA is a relatively simple analysis. There are, after all, only four input categories: (1) cost of the medical technology of interest, (2) cost of the comparator medical technology, (3) health utility associated with the new technology, and (4) health utility associated with the comparator medical technology.

13.3 Establishing Health Utilities or What Is a QALY?

QALYs are a unique outcome measure because they represent a composite measure of quality and quantity of life. It was first used to evaluate the effectiveness of hypertension medications [4]. QALYs are derived from health utilities multiplied by time (typically years). This is important to understand because not all QALYs are created equal (see Fig. 13.2). For example, one full year of life in perfect health represents one QALY (Fig. 13.2). Similarly, one full year of life in a health state valued at 0.5 (half as good as perfect health) represents 0.5 QALY (Fig. 13.2). Figure 13.2 provides a demonstration of two very different scenarios in which there are equal QALY values.

Fig. 13.2

Time, measured in years, is on the x–axis. Health utility is measured on the y–axis with a value of 1 representing perfect health and 0 representing death. Note that there is space for utilities valued below 0 reserved for health states valued as worse than death. In graphical format, the number of QALYs is equal to the total shaded area. In the case of patient A, one half-year of life was spent in a state of health valued at 0.5 followed by sudden death, resulting in a total of 0.5 QALY. Patient B, on the other hand, experiences a gradual decrement in health state utility until death at 1 year. This also translates into 0.5 QALY. While QALYs represent a way to reduce benefits across patients and interventions into a comparable number, it is important to understand that two very different life courses can represent the exact same number of QALYs

Health utilities are assigned using preference-based health state questionnaires in which populations respond to questionnaires developed to inquire about a number of functional domains. Some commonly used health state questionnaires are the EuroQol-5D, HUI2 and HUI3, etc. The EQ-5D, for example, has 243 possible health states, each with a unique utility value. Utilities for each health state are obtained by asking a general population to rate a sample set of those 243 health states (mathematical modeling was used for the remainder). Instruments such as the HUI2 and HUI3 have even more possible states of health (requiring even more mathematical modeling).

There are a number of important questions to consider that affect the generalizability of CEA results. First, the assignment of health state utilities is biased based on the group providing the ratings. In the case of the EQ-5D, the utilities were derived based on the preferences of the general British population. Since its initial inception, there have been a number of other utility value sets derived for American, Danish, German, and many other populations. However, if a study uses a health state questionnaire that has not been validated in a particular study population, then the health utilities and thus QALYs derived from these questionnaires should be viewed critically. Second, many important considerations (economic, social, etc.) are omitted due to complexity, insufficient data, or limitations of the perspective used for the study [5,6]. If the economic benefits of return to work or the ability to maintain a higher level of employment, for example, is not considered, then the reported QALY may be understated. Lastly, it is also important to recognize that much of the available data comes from carefully designed clinical trials designed to answer one or maybe two hypotheses. Very little pragmatic data is included in these trials. Much like the narrowly defined randomized control trial, the generalizability of a CEA study that uses this design is often not generalizable to many clinical situations. Like other aspects of CEAs discussed above, rather than invalidating CEA results, knowledge of such limitations helps inform each reader’s final interpretation.

13.4 Establishing Cost

At first glance, cost seems like the simplest aspect of a CEA. Currency is much easier to understand than utility. Yet, significant heterogeneity exists from hospital to hospital and clinic to clinic clouding the “true cost” of a good. Additionally, confusion over the origins and accuracy of hospital charges makes determining the true cost even more difficult [7]. Finally, there are often costs that are not included in studies. These can include the cost of missing work, not only to the patient but to the company; the cost of the time family members and friends spend providing “free” or informal care to patients; the cost to society of diverting resources that could have been used elsewhere to health care instead. Most frequently these are the costs of informal care, but may also include many other aspects that go into the “true cost” of health care.

13.5 Establishing Comparison Technologies

A CEA involves comparison between two existing technologies. The important detail, however, is that one is not free to choose any existing technology as a baseline. Instead, the comparator (i.e., the technology serving as the baseline) to which the new technology is being compared must be the next best treatment that remains cost-effective. For example, in evaluating new methods to promote fusion in spinal surgery, it is not acceptable to compare each new intervention to medical therapy alone; rather, the comparison must be between the new technology and the standard surgical fusion procedure (see example below). This has two implications. First, in designing a CEA, it is important to choose a baseline technology that is not already known to be dominated (i.e., more effective at less cost, Fig. 13.3a). Second, inappropriate choice of comparators can lead to calculation of an extremely misleading value known as the average cost-effectiveness ratio (ACER) rather than the ICER (Fig. 13.3c).

Fig. 13.3

(a) A standard cost-effectiveness curve with health effect, often reported as QALYs, on the x-axis and cost on the y-axis. Four different technologies are represented here representing four different treatments for the same disease. Technology D is both more costly and less effective than technology A. This is the definition of dominance, and therefore, technology D would be referred to as being dominated by technology A (sometimes this concept is reported as a negative ICER). Therefore, there is no sense in conducting a CEA for technology D versus any of the other technologies because there is already an alternative that is better in terms of cost and health effect. (b) Lines connecting technology A to technology B and technology B to technology C have been drawn. These lines are a graphical representation of what is measured in a CEA, i.e., the difference in cost compared to the difference in health effect. The slopes of these lines, referred to as Slope AB and Slope BC, respectively, represent the ICERs of technology A compared to technology B and technology B compared to technology C. Remember that the slope of a line on an xy–axis is the amount of change in vertical units (y–axis units) for an increase of one horizontal unit (x–axis units). In this case, Slope AB and Slope BC represent the cost of A relative to B and B relative to C for each increase of one QALY. This is otherwise known as the ICER. (c) In this case, technology A is compared directly to technology C. The line connecting technology A to technology C has a slope designated “Slope AC.” Note that even though technology B derives some benefit for patients relative to technology C, a comparison such as this ascribes all of the health benefits to technology A. This is despite the fact that technology A adds very little benefit on top of technology B at a significantly higher cost. (d) The effect of inappropriately choosing a comparator that is not the next-best non-dominated alternative. If a CEA analysis were performed comparing technology A to technology C, ignoring technology B, the result would be reported as an ICER equal to the slope of the line connecting technology A to technology C, denoted here as slope AC. This is in contrast to Slope AB referring to the slope of the line connecting technology A to technology B. A more horizontal line represents a lesser slope, and therefore, Slope AC is much less than Slope AB. If only slope AC was reported, technology A would appear to be much more cost-effective than it really is. Therefore, understanding the alternatives available and choosing the appropriate alternative technology for comparison are crucial in conducting CEAs and are critical points that must be understood and examined when interpreting a CEA

Only gold members can continue reading. Log In or Register to continue