What outcomes should be used to measure if a medicine is a “good” medicine? Improved survival? Improved ability to function? Few side effects? Ability to get back to work? Living to attend your daughter’s wedding? All of the above?
Oftentimes, health economists measure value of a treatment for the average person. Preferences for efficacy, safety, dosing and other factors are averaged across a whole population. However, could we actually measure whether a treatment helped more people attain their personalized goals?
That is specifically the aim of Goal Attainment Scaling (GAS). As described in Zaza et al. (1999), the GAS methodology relies on six primary steps.
1. Selection of Goal Areas. A patient’s problem areas are assessed and a set of priority goal areas (usually 3–5) is identified. Identified goals should be those that will be the focus of care planning or intervention—a patient may have many problems that may not be expected to change, or are not relevant to specific treatment programs. As a general rule, a guide should have at least three goals.32
If new goals are identified during the course of treatment, these can be added.
2.Weighting. As an optional step, the goal areas may be assigned subjective weights to reflect their estimated relative importance. Weighting of the goal areas may be helpful clinically, but if not, this step can usually be omitted without major effects on the GAS scores.
3.Follow-up Time Selection. The anticipated time frame for goal attainment is determined. This is usually the same for all goals, but may vary for each goal.
4.Statement of Expected Outcomes. The most probable result that would be expected if treatments and interventions are effectively implemented is recorded as the “0” category in the center of the follow-up guide. Attainment levels should be descriptive, observable, and objective—an independent rater should be able to reliably assess the patient’s attainment level for each goal. “Change” words or relative terms such as “better” or “improved” should be avoided in favor of descriptions of observable states. Although there may be a temptation to use GAS to measure activity, workload, or processes of care, GAS is best used to scale outcomes, not the specific procedures or processes that will be used to achieve the outcome. Therefore, “process goals,” such as “arrange CT scan,” are to be avoided.
5.Completion of the Other Scale Levels. The remaining scale levels are filled in for each goal area. These indicate plausible outcomes that would be much less than expected (−2), somewhat less than expected (−1), somewhat better than expected (+1), and much better than expected (+2). As with the expected outcome identified in Step 4, these are to be objective and observable indicators of the patient’s status on each goal area. Goals should be clearly scorable at only one level. As with any scale, goal levels should not overlap. If it is difficult to imagine a better outcome than the expected (“0”) level, the “+1” and “+2” levels might be scaled as achieving the same expected level, but in less time, or with fewer resources. Standardized rating scales may be helpful in providing descriptors for the scale levels.
6.Follow-up. At the end of the follow-up period(s), the patient is assessed in terms of the goal attainment levels and her or his status is marked on the follow-up guide. Follow-up status is usually marked with an asterisk, and the initial assessment level with a checkmark.
Using this approach, the final goal attainment score for an individual can be calculated as follows:
Here, the term wi is the weight assigned to the ith goal, and xi is the numerical value of the attainment level of the ith goal. The value of xi ranges between -2 to +2. If all goals are weighted equally, then wi=1. In general, a score of 50 or better mean that the patient is achieving their goals on average.
While GAS can be used to measure the value across a variety of interventions, the Zaza et al. (1999) article provides examples from individuals with pain including cancer pain, geriatric pain, pediatric pain, and work-related nonmalignant pain.
A systematic literature review of GAS by Gaasterland et al. (2016) identified 58 articles who used GAS, 38 of them were drug studies using GAS was used as an outcome measure, and 20 studies used GAS in other (i.e., non-drug) settings. While GAS is an attractive approach for measuring the impact of a drug given its individualized nature, additional research is needed to determine how best to implement GAS. For instance, Tennant (2007) notes that using ordinal rather than linear scales may be preferred for identifying minimum clinically important differences (MCID) in outcomes.