Since the Oswestry Disability Index (ODI) was first published in 1980 it has become an internationally recognized, gold-standard outcome measure for research on spinal disorders.
It is important to first establish what the ODI is trying to measure. The standard version of the ODI asks the respondent about information concerning the effects of back and leg pain on the day he/she is actually filling out the questionnaire. Other measures use “this week” or even “this month” as the frame of reference.
The ODI is divided into ten sections, the first of which relates to pain intensity. The structure of this section is repeated in all the other sections. Each section has six possible responses, which are scored from 0 to 5. In theory, only a single response in a single section is required to give an ODI score. In practice, however, we rarely find responses to fewer than eight sections. If a respondent checks two boxes in a section, the highest scoring box is taken as the true response.
The other sections cover such issues as personal care, lifting, walking, sitting, standing, sleeping, sex life (if applicable), social life, and traveling. Each section has a maximum score of five points. If all sections are completed and the responses are totaled together the maximum possible score is 50 points. We then double this score thus giving the Index a total range from 0 to 100. Though the original concept was to give a percentage score, most users just use points to express the total ODI’s score.
Meaning of scores
In our original publication we suggested that a score of 0% to 20% represented minimal disability; 20% to 40%, moderate disability; 40% to 60%, severe disability; 60% to 80%, crippled. We felt scores of 80% to 100% reflected either an exaggerated response or the respondent was almost certainly bedbound.
In 2000, I published a paper2 with Paul B. Pynsent, PhD, Director of Research at the Royal Orthopaedic Hospital, Birmingham, UK. We reviewed over 300 publications that had used the ODI as a Patient-Reported Outcome (PRO) measure. Investigators had used the ODI to assess a wide variety of conditions. With such a large sampling we were able to determine a plausible mean score for each condition and the likely effects of these conditions on perceived disability.
More recently in 2013, three Dutch researcher Miranda van Hooff developed a series of publications covering an intensive rehabilitation program in which she used the ODI to measure outcomes in large cohorts of back pain patients. She, and others, had sought to establish the criteria for designating success and failure of treatment using the ODI.
For an individual patient, a 15-point reduction between baseline and post-treatment is widely accepted as a clinically relevant change. In fact, such a figure has been adopted by the U.S. Food and Drug Administration for the assessment of spinal surgical interventions. However, in my view, there are some unsatisfactory features of this approach. First it depends on the initial pre-intervention score, which, if it is already a low number, may not allow a 15-point clinically significant improvement.
On the other hand, if the patient has a very high baseline score, a 15-point change may reflect only a slight improvement, since the subject would still have a high level of disability, even though it was better than before the intervention. Furthermore, all this assumes that the ODI and other PRO measures behave in a linear fashion. We feel that is impossible to prove or disprove.
For that reason, we decided to try to find an absolute score equivalent to a satisfactory symptom state. You may recall the ODI defines five groups of scores reflecting different levels of disability. We suggested that people with a score falling in the 0 to 20% group would have “minimal disability.”
Miranda went back to our 2000 publication and identified the references for four cohorts of back pain free individuals4-7 and found they had a mean ODI score of 10. That led her to define a normal range as being mean plus two standard deviations, thus making the normal range for the ODI of less than 22 points.
Figure 1. Functional status as measured with the ODI (0-100) in the RealHealthNL study sample (n=524)
In Figure 1, plotting pre-treatment scores against post-treatment scores, you can see that in this cohort of 524 patients who underwent an intensive rehabilitation and fitness program in the Netherlands, that the vast majority showed an improvement in ODI scores at the one-year follow-up point. Those falling below the neutral line are shown in green. By drawing a line parallel to and 15 points below this mark we have the FDA’s criterion of success. If we then draw a horizontal line at 22, the patients with scores falling below this line are considered to have a satisfactory symptom state at 12 months. About 40% of Miranda’s cohort fell into this group following intensive rehabilitation.
Anne Mannion, of the Schulthess Klinic in Zurich, took a different approach. She looked at a cohort of 532 patients undergoing lumbar spinal surgery. All these patients had completed the ODI and COMI (Core Outcome Measures Index) scores were taken at various times up to four years after surgery. One of the COMI items asks: “If this is how you had to spend the rest of your life with the symptoms you have right now, how would you feel about it?” There were five available responses, of which the first and the second responses were “very satisfied” and “somewhat satisfied.” These responses were related to the ODI score. The ODI was able to respond well and discriminate between these two groups. Anne plotted Receiver Operating Characteristic (ROC) curves, the cut-off indicating a “somewhat satisfied state” at 29 points and a “very satisfied state” at 14 points.
I believe that these change scores reflect an improvement after treatment. This method may give a more optimistic view than when you look at the proportion of patients reporting a “satisfactory state.” I believe that investigators should look at the concept of the proportion of patients reaching a satisfactory symptom state in the clinical and the research setting.
At this point, I favor the 22-point criterion as being the best choice, but I cannot prove this definitely, and more work on larger cohorts is needed.
Translations and “rogue” ODI versions
The ODI, though originally developed in English, has been used worldwide in many dialects of English and other languages for over three decades. In fact, the Mapi Research Trust (MRT) now has more than 60 translations available for licensing—and new ones are appearing almost yearly.
I have chosen MRT to handle the global licensing of the ODI because I have had increasing problems with investigators using what I call “rogue versions” of the Index. Unfortunately, this is not a problem unique to the ODI. In fact, many PRO measures are altered without the knowledge or accord of the developers every year. Many of these investigators believe that they can improve on the original instrument but by altering it they make it impossible to compare their results with those who had used the original concepts.
After some 35 years of scrutiny and worldwide use, the ODI has proven itself to work extremely well. That said, I prefer all those who use it choose its latest iteration: version 2.1a. This does not preclude future development but any alterations to it should be done in conjunction with the authors and the Mapi Research Trust.
I’d like to provide an example of what can happen when a “rogue version” is used rather than an approved and validated version ODI. In a number of recent studies of chronic back pain and surgical treatment by spinal fusion or disc replacement, the mean and standard deviation baseline ODI scores ranged between 40 and 55 points. This was an expected variation. However there were two outliers. It took me a long time to find out why the results of these two studies were so different from the others.
It turned out that the investigators had found a version of the ODI that had been developed by chiropractors8 without my knowledge or consent. There were some fundamental differences in the structure of some of the questions. In fact, these questions addressed completely different concepts from the original ODI. This “chiropractors’ version” was, of course, never validated and it was only when used in a large, otherwise well-designed study that its deficits were revealed.
Around the same time that I learned of this “rogue version,” Megan Davidson, an Australian physiotherapist, had been performing a Rasch analysis of three different versions of the ODI. She found that this rogue chiropractic version behaved in a completely different way from the other versions. In my view, using an altered and non-validated version of a PRO instrument discredits any results the study may have found.
I have written a number of articles about this, most recently in the Journal of Neurosurgery Spine.9 My experience with “rogue versions” was one of the reasons why I asked the Mapi Research Trust to license the use of the ODI and to identify where such versions of it have been used or developed.
Electronic versions of the ODI
We are working on eversions, and some of them will be available soon. For more information, please visit ePROVIDE™.
Availability of the ODI
Clinical and academic users can obtain a license to use the fully validated ODI version 2.1a from the Mapi Research Trust at no charge. Commercial investigators will be asked to pay an affordable license fee. By contacting the Mapi Research Trust before starting any new study, investigators can be assured that they will have the latest and fully approved version of the ODI.
For more information about the Oswestry Disability Index and how to get a license to use it in a future study, please go to: https://eprovide.mapi-trust.org/instruments/oswestry-disability-index or simply send an email request to email@example.com
If your study calls for the ODI in languages other than English, chances are it has already been translated and linguistically validated in those languages. In the rare case that it has not, the experts at Mapi’s Linguistic Validation Department can provide you with a cost and time estimate to have it linguistically and culturally validated into any language you may need.
- Fairbank JC, Couper J, Davies JB. The Oswestry Low Back Pain Questionnaire. Physiotherapy 1980; 66: 271-273.
- Fairbank J, Pynsent P. The Oswestry Disability Index. Spine 2000;25:2940-53
- Fairbank J, Van Hoof ML, Mannion AF. Determination of the Oswestry Disability Index (ODI) score equivalent to a “satisfactory symptom state. Presented at the Europsine meeting, 2013, Liverpool. Available at http://eposterliverpool.eurospine.org/cm_data/eposter/QF77.PDF
- Kankaanpaa M, Taimela S, Laaksonen D, Hanninen O, Airaksinen O. Back and hip extensor fatigability in chronic low back pain patients and controls. Archives of Physical Medicine and Rehabilitation 1998;79:412-7.
- Kankaanpaa M, Taimela S, Webber CL, Airaksinen O, Hanninen O. Lumbar paraspinal muscle fatigability in repetitive isoinertial loading: EMG spectral indices, Borg scale and endurance time. European Journal of Applied Physiology and Occupational Physiology 1997;76:236-42.
- Mayo NE. Letter. Spine 1995;20:1535-6
- Deo S, Wanders L, Makan P, et al. Outcome measures for neurogenic claudication. North American Spine Society 1998, San Francisco.
- Zigler, J; Delamarter, R; Spivak, J; et al. Results of the Prospective, Randomized, Multicenter Food and Drug Administration Investigational Device Exemption Study of the ProDisc(R)-L Total Disc Replacement Versus Circumferential Fusion for
- the Treatment of 1-Level Degenerative Disc Disease. Spine 2007;32(11):1155-1162
- Fairbank J. Letter to the Editor: Oswestry Disability Index. Journal of Neurosurgery Spine. 2014 Feb 2014;20(2):239- 42.