- Email: [email protected]
Clinically Important Changes—What's Important and Whose Change Is It Anyway?
Vol. 25 No. 5 May 2003
Journal of Pain and Symptom Management
395
Editorial
Clinically Important Changes—What’s Important and Whose Change Is It Anyway? Henry J. McQuay, DM, Jodie Barden, BA, and R. Andrew Moore, DSc Pain Research, University of Oxford, Churchill Hospital, Oxford, United Kingdom
It is very easy to become muddled by the phrases clinically important difference, clinically meaningful difference, minimal detectable change or minimal clinically important difference. We badly need to organize our thoughts because this is an important research topic. Unfortunately we have at least two camps of people using the same words to describe different things. The aim of this editorial is to stimulate discussion so that we move forward. The two camps start in different fields. Researchers want to anchor their research methods into clinical reality. How much change on a visual analogue scale of pain intensity represents a worthwhile change to all patients? The clinicians want to know if last week’s change in medication produced a worthwhile result for that patient. The aspirations of both camps clearly share some common ground, but the questions are subtly different, and the methods used to achieve an answer are different, too. We need to have an agreed vocabulary so that each of us knows what the other is trying to say. One starting point is experimental pain, where historically two terms, threshold and tolerance, were used. Threshold meant the time (after immersion) at which you became aware of the pain in your arm due to the icy water around it. Tolerance was the time at which you could no longer tolerate the pain, and had to withdraw your arm from the icy water. These two old-fashioned terms, threshold and tolerance, are important if we are to be clear. Threshold, we think, is the minimal detectable
Address reprint requests to: Henry J. McQuay, DM, Oxford Pain Relief Unit, Churchill Hospital, Headington, Oxford OX3 7LJ, United Kingdom. © 2003 U.S. Cancer Pain Relief Committee Published by Elsevier. All rights reserved.
change. Tolerance is the clinically important or meaningful term. Table 1 shows the possible relationship between these terms and the researcher and clinician aspirations, using words and avoiding pejorative acronyms. The paper by Farrar et al.1 uses a cancer pain dataset (open titration phase) to confirm a previous finding that the substantive effect (Table 1)2 occurred at about 30% pain relief. The endpoint they used for substantive effect was the decision to remedicate. This is an important confirmation, but there are a number of caveats that the authors acknowledge fairly. The first is that the original finding and this confirmation came from cancer pain data, and we do not know if these findings are valid in other pain settings. Put simply, can we extrapolate these findings to other pain settings with confidence? Second is the limitations of the data. Two-thirds of the placebo patients in the randomized phase of the trial did not remedicate, so that the remedication endpoint may not be robust if the pain is self-limited. Nonetheless these two papers do move us forward. The hazard from these papers is that there is a reflex among pain clinicians to jump to the simplistic thought that “30% relief” is all we need to know, just as in the past we jumped to the thought that placebo always had a 30% fixed effect in 30% of the people. Funny how it is always 30%. Really all we can say at the moment is that in these datasets remedication was significantly more likely if there was less than 30% relief. Our group has used 50% relief as the outcome to derive league tables of relative efficacy of analgesics. We chose this time-honored endpoint from Beecher’s work because of its clinical grounding. We have subsequently analyzed 0885-3924/03/$–see front matter doi:10.1016/S0885-3924(03)00099-X
396
Editorial
Table 1 Terminology Experimental Pain
Research
Clinic
Threshold
Perceptible effect? Substantive effect?
Any discernible effect? Satisfactory effect?
Tolerance
See Reference 2 for a more comprehensive table published by rheumatology researchers.
at other “cut-off” points, including 30% relief, and have shown that the choice of cut-off point makes no difference to the relative efficacy.3 Farrar et al. are asking a different question “below what level of relief is the patient likely to need remedication,” and at the moment the 30% figure is that number.
Vol. 25 No. 5 May 2003
Now we need to know whether or not we can generalize that number to other pain contexts, and we need to agree on a vocabulary for these threshold and tolerance endpoints in research and in the clinic.
References 1. Farrar JT, Berlin JA, Strom BL. Clinically important changes in acute pain outcome measures: a validation study. J Pain Symptom Manage 2003;25(5): 406–411. 2. Beaton DE, Bombardier C, Katz JN, et al. Looking for important change/differences in studies of responsiveness. J Rheumatol 2001;28:400–405. 3. McQuay HJ, Moore RA. An evidence-based resource for pain relief. Oxford: Oxford University Press, 1998.
Journal of Pain and Symptom Management
395
Editorial
Clinically Important Changes—What’s Important and Whose Change Is It Anyway? Henry J. McQuay, DM, Jodie Barden, BA, and R. Andrew Moore, DSc Pain Research, University of Oxford, Churchill Hospital, Oxford, United Kingdom
It is very easy to become muddled by the phrases clinically important difference, clinically meaningful difference, minimal detectable change or minimal clinically important difference. We badly need to organize our thoughts because this is an important research topic. Unfortunately we have at least two camps of people using the same words to describe different things. The aim of this editorial is to stimulate discussion so that we move forward. The two camps start in different fields. Researchers want to anchor their research methods into clinical reality. How much change on a visual analogue scale of pain intensity represents a worthwhile change to all patients? The clinicians want to know if last week’s change in medication produced a worthwhile result for that patient. The aspirations of both camps clearly share some common ground, but the questions are subtly different, and the methods used to achieve an answer are different, too. We need to have an agreed vocabulary so that each of us knows what the other is trying to say. One starting point is experimental pain, where historically two terms, threshold and tolerance, were used. Threshold meant the time (after immersion) at which you became aware of the pain in your arm due to the icy water around it. Tolerance was the time at which you could no longer tolerate the pain, and had to withdraw your arm from the icy water. These two old-fashioned terms, threshold and tolerance, are important if we are to be clear. Threshold, we think, is the minimal detectable
Address reprint requests to: Henry J. McQuay, DM, Oxford Pain Relief Unit, Churchill Hospital, Headington, Oxford OX3 7LJ, United Kingdom. © 2003 U.S. Cancer Pain Relief Committee Published by Elsevier. All rights reserved.
change. Tolerance is the clinically important or meaningful term. Table 1 shows the possible relationship between these terms and the researcher and clinician aspirations, using words and avoiding pejorative acronyms. The paper by Farrar et al.1 uses a cancer pain dataset (open titration phase) to confirm a previous finding that the substantive effect (Table 1)2 occurred at about 30% pain relief. The endpoint they used for substantive effect was the decision to remedicate. This is an important confirmation, but there are a number of caveats that the authors acknowledge fairly. The first is that the original finding and this confirmation came from cancer pain data, and we do not know if these findings are valid in other pain settings. Put simply, can we extrapolate these findings to other pain settings with confidence? Second is the limitations of the data. Two-thirds of the placebo patients in the randomized phase of the trial did not remedicate, so that the remedication endpoint may not be robust if the pain is self-limited. Nonetheless these two papers do move us forward. The hazard from these papers is that there is a reflex among pain clinicians to jump to the simplistic thought that “30% relief” is all we need to know, just as in the past we jumped to the thought that placebo always had a 30% fixed effect in 30% of the people. Funny how it is always 30%. Really all we can say at the moment is that in these datasets remedication was significantly more likely if there was less than 30% relief. Our group has used 50% relief as the outcome to derive league tables of relative efficacy of analgesics. We chose this time-honored endpoint from Beecher’s work because of its clinical grounding. We have subsequently analyzed 0885-3924/03/$–see front matter doi:10.1016/S0885-3924(03)00099-X
396
Editorial
Table 1 Terminology Experimental Pain
Research
Clinic
Threshold
Perceptible effect? Substantive effect?
Any discernible effect? Satisfactory effect?
Tolerance
See Reference 2 for a more comprehensive table published by rheumatology researchers.
at other “cut-off” points, including 30% relief, and have shown that the choice of cut-off point makes no difference to the relative efficacy.3 Farrar et al. are asking a different question “below what level of relief is the patient likely to need remedication,” and at the moment the 30% figure is that number.
Vol. 25 No. 5 May 2003
Now we need to know whether or not we can generalize that number to other pain contexts, and we need to agree on a vocabulary for these threshold and tolerance endpoints in research and in the clinic.
References 1. Farrar JT, Berlin JA, Strom BL. Clinically important changes in acute pain outcome measures: a validation study. J Pain Symptom Manage 2003;25(5): 406–411. 2. Beaton DE, Bombardier C, Katz JN, et al. Looking for important change/differences in studies of responsiveness. J Rheumatol 2001;28:400–405. 3. McQuay HJ, Moore RA. An evidence-based resource for pain relief. Oxford: Oxford University Press, 1998.