Last updated: Sun, Jul 28, 2024
What is your pain intensity level on a scale from 0 to 10, where 0 is no pain and 10 is the worst pain you can imagine?
You've been asked this question if you are a pain patient. It is used clinically and in pain reserach. You answer it by a mysterious process that to me seems much like a kid's “pick a number from one to ten” game. Several concepts are hidden in this apparently straightforwad question. These are: 1) pain has a quality called “intensity”; 2) any two pains (or two time periods?) can be compared based on this quality, and 3) you should be able to evaluate the intensity to a precision of one part in eleven. This seems reasonable. After all, if you don't know what your pain intensity level is, who does? (Also, if we can't measure pain levels, how can we say that drug X is Y% better than drug Z?)
This scale is properly called a numerical rating scale, since it asks you to provide a numeric response. It is commonly called instead the visual analog scale or the VAS. The visual analog scale is actually a line or bar on a piece of paper. One end is labeled “no pain,” the other “worst pain you can imagine,” and you are asked to mark the line to correspond to your pain intensity level. Someone must measure the distance between the zero mark and your mark to determine your pain score. The formal differences are subtle. With the VAS you can choose any value, not just whole numbers. With the VAS you estimate your pain visually, while with the numerical rating scale you estimate based on your “number sense,” which adds a basically unknowable source of variability to this measurement.
The question has become a standard in pain treatment. It must be the most-commonly asked question in a clinical setting, implying that it is regarded as important information by care providers. It originated in a research context. The Handbook of Pain Assessment has this to say about it:
...there is much evidence supporting the validity of VASs of pain intensity. Such scales demonstrate positive relations to other self-report measures of pain intensity. They are sensitive to treatment effects and are distinct from measures of other subjective components of pain. The scores from VASs appear to have the qualities of ratio scale data for groups of people. Thus, differences in pain intensity (for groups, but not necessarily for individuals), as measured by VAS scales, represent actual differences in magnitude.1
I am going to go through this statement in some detail because we don't have a better way to measure pain intensity (more later on what pain intensity may be). Many of the research results that determine our medical treatment as pain patients are absolutely dependent on the quality of this metric. In many cases our medical treatment is highly dependent on our doctor's beliefs about the intensity of our pain. You'll see later in the book that this turns around and swallows its own tail once we visit the doctor. So, here goes.
Scientists have developed some fairly sophisticated ideas about when a measurement technique is appropriate and when it is not. Validity is the extent to which a measurement accurately represents what it is intended to represent. The central problem with VAS pain intensity is that there is no way to measure the validity of this measurement, since we have no better measure of pain intensity level than the VAS. We can't hook people up to a Pain-o-Meter and compare the reading to their VAS number. Scientists have broken the overall idea of validity
down into several types of validity. By looking at these types of validity, we can see validity somewhat more clearly.
Face validity refers to whether the measurement seems right and fitting to people. My experience is that most people think that the VAS question makes sense. (I'm the only one who doesn't.) So, “Check” on the face validity of the VAS.
Construct validity refers to the relation between the measurement process and the theory that it is based on. The measurement process should be consistent with the theory of what pain intensity is. My research hasn't revealed any such theory to me, beyond the concepts implied by the VAS itself. In fact, there is much head-scratching about it, much perplexity about why different people respond differently to stimulation that seems similar, why people report pain when the observer thinks perhaps they oughtn't, and so on. Similarly, while there is much discussion about 'dimensions' of pain, a scale such as the VAS recognizes only one. The structure of the VAS implies that any two pains can be compared, and that either one is greater than the other or that they have the same “size.” Perhaps a “Question mark” of construct validity.
Content validity refers to whether the measurement covers everything that we mean when we say “pain intensity level.” In my mind this is another poser. The VAS measurement implies that pain intensity level is a single number. It is reasonable that there should be a size or a greatness to pain, since it seems self-evident that some pains are greater than others. Certainly I can testify that some pains are preferable to others. On the other hand, it is completely unclear and unspecified as to what aspects of pain, which we know is complex, this measurement covers. Perhaps it measures the degree to which an individual dislikes the pain. Perhaps it means different things depending on the context it is used in. Perhaps it means different things to different people. There is no way to know for sure. I suggest another “Question mark” for content validity.
So, what kind of validity do the authors of The Pain Assessment Handbook think that the VAS has? For one thing, they say, scales such as the VAS “demonstrate positive relations” to other self-report measures of intensity. In other words, if you ask the question in a different way, you get similar answers. They also say that such scales are “sensitive to treatment effects.” In other words, when you expect the VAS to be different (because you amped up the pain stimulus or applied some treatment intended to change the pain intensity), it is in fact different. This is dangerous because it defines accuracy in terms of what was expected, and it invokes the tension between paradigm and skepticism that I discussed in Confirmation Bias. However, when applied across large number of studies, the VAS has behaved reasonably.
The authors say that scales like the VAS are distinct from measures of other subjective components of pain.
This is a good thing if true, but raises a further set of concerns: All of the questions that can be raised about "pain intensity level" can be raised about these other subjective components of pain. It is possible, and it has been done, to develop another measurement scale for another "dimension," and it is possible and has been done to verify statistically that it measures something different than the VAS does. But the questions about exactly what is being measured remain. This general topic is discussed in the following section on The McGill Pain Questionnaire, which I think helps to clarify, again, what is known and what isn't.
Next, The scores from VASs appear to have the qualities of ratio scale data for groups of people.
Ratio scale data means data that it makes sense to form ratios from. It makes sense, for example, to say that car A is going 50% faster than car B because there is such a thing as no speed (sitting still), and because the difference between a speed of, say, 12 mph and 14 mph is the same as the difference between, say, a speed of 17 and 19 mph. Simply assigning numerical values to things is not enough to make a ratio scale. It makes no sense to say that my telephone number is half as much as yours.
It is significant that the authors specify that the VAS produces a ratio scale for groups of people.
By implication they are saying that there is no reason to assume that individual VAS scores, or groups of scores from an individual, have these same qualities. This is (roughly speaking) because the scores of groups of people average out irregularities in the scoring of individuals. If Subject 3679 always rates his pain a 6 when it's “really” a 3 (whatever this might mean), this won't make much difference in an average of hundreds of subjects. If Subject 3679 does this, Subject 3681 may well make the opposite error and cancel 3679 out. On the other hand, if 3679 and 3681 each reports a 6 to his doctor, the doctors may make poor medical decisions if they rely on the comparability of the scores.
Most pain clinics I've been to have their own version of the scale that retains the zero-to-ten range of scores and adds some description of intermediate scores. This is called anchoring the scale. I don't know whether these exist because the patients are uncomfortable without them or because their doctors are. I do know that each seems to be a little different. It makes me very skeptical that people can communicate using this vocabulary.
The VAS meets a need, the need to estimate the intensity, size, or seriousness of pain. It is well-suited to research, where it is always used as group averages. Its validity is limited by the nature of pain itself and/or our current thinking about it. Its use has been expanded into the clinic, where the limited claims it has to validity are much weaker. The McGill Pain Questionnaire, discussed in the next section, provides an illuminating contrast.
Within this section...
VAS and Similar Intensity Scales
Or skip to...
Pain Thresholds (This page is incomplete.)