September 2, 2010




International Research Foundation for RSD / CRPS, Inc.



Joseph Walsh, PhD

Vice President for Research

Northwestern University


This memorandum is submitted by representatives of the International Research Foundation for RSD/CRPS.  Our mission is to promote excellence in the treatment and research of RSD/CRPS.  In furtherance of that mission, our organization reviews and evaluates relevant research studies in order to identify and publicize quality research that will further patient care.  At times, however, we identify research studies that are not based on sound research practices and which, if relied upon, result in substandard patient care.  A recent review of several studies conducted by a member of your faculty has raised serious concerns about their validity, and we feel obligated to report our findings to you.

R. Norman Harden, M.D., a member of your faculty, is an advocate for the use of diagnostic criteria for complex regional pain syndrome (CRPS), known as the Budapest diagnostic criteria, which he has been instrumental in developing.  Dr. Harden has promoted these criteria based in part on research conducted at your institution:

S. Bruehl, R, N. Harden et al. / Pain 81 (1999) 147–154

R.N. Harden, S, Bruehl et al. / Pain 83 (1999) 211-219

S. Bruehl, R. N. Harden et al. / Pain 95 (2002) 119–124

R.N. Harden, S. Bruel et al. / PAIN 150 (2010) 268–274


However, our review of these studies has identified several discrepancies and misstatements.  The following observations, summarized by Anthony Kirkpatrick, MD, PhD, a member of our Board of Directors, raise serious questions about the scientific integrity of the data upon which Harden's Budapest diagnostic criteria for CRPS is based:


1. As is true for all human subjects research, studies conducted at Northwestern University must receive IRB approval.  However, there is no evidence that the 1999 Harden validation study published in the journal PAIN (upon which the Budapest Criteria are based) received approval from your institution’s IRB before subjects at Northwestern University facilities were enrolled.  The same is true for a subsequent clinical study in 1999 and another clinical study in 2002 performed by Harden and published in the journal.

The failure to obtain necessary IRB approval demonstrates serious noncompliance with basic principles designed to ensure the integrity of the research and the protection of the rights and welfare of human subjects.




2. Subsequent to Harden's first validation paper in 1999 for the diagnosis of CRPS, a group of CRPS experts in Japan received a copy of Harden's CRPS data checklist and instructions from Harden. The investigators then attempted to reproduce the results of the 1999 Harden study by applying his diagnostic criteria to a larger group of research subjects with CRPS. Like the Harden study, they compared the CRPS group with non-CRPS research subjects. Harden’s first validation study in 1999 left out 15% of patients with CRPS (i.e., sensitivity 0.85). In contrast, the larger Japanese validation study published in the journal PAIN in 2010 showed even worse sensitivity. When Harden's diagnostic criteria were applied by the Japanese investigators, 55% of research subjects with CRPS were left out (i.e., sensitivity 0.45).

The poorer results for diagnostic sensitivity in the larger Japanese study, raises the question about a difference in CRPS in Americans and Japanese. The poorer outcome of the Harden diagnostic criteria when tested independently overseas also raises the question about the scientific integrity of Harden's first validation study. In addition, the results from these studies suggest that Harden’s application of his diagnostic criteria to defeat a patient’s claim of CRPS in court could miss the diagnosis of CRPS when it is actually present. Failing to diagnose CRPS will lead to denying or delaying treatments that are known to alleviate pain and suffering due to this neurological disorder. 

[ CRPS: Current Diagnosis and Therapy, 2005, p. 217 ]   


3. Even though the investigators of the larger Japanese study concluded that their diagnostic criteria were more effective than Harden’s criteria in making the diagnosis of CRPS, they caution that their criteria are not completely perfect (i.e., sensitivity 1.0, specificity 1.0) and therefore hope that their criteria will not be used in court to replace health professionals in determining if CRPS is present. John D. Loeser MD, Chair, IASP Taxonomy Working Group, published a comment particularly apropos the Harden situation: “We must be careful in our pursuit of diagnostic criteria, for it is far too easy to make definitions whose major purpose is to validate what we do to earn our living". [ CRPS: Current Diagnosis and Therapy, 2005, p. 4 ]

On September 12, 2007, in a civil lawsuit styled Robert Hogan, Plaintiff vs. Baptist Medical Center – Nassau, Inc., a Florida Corporation, Defendant and Third-Party Plaintiff vs. American Cancer Society and Gray Gable, Nassau Village Volunteer Fire Department, Inc., Third Party Defendants, 4th Judicial Circuit, Nassau County, Florida (Case No. 06-CA-44), Harden testified under oath in 2 depositions.

Deposition #1

Deposition #2


View a 28-minute video of Harden's sworn deposition testimony:



4. In the August 2010 issue of PAIN, Harden is the lead author in a second validation study of his diagnostic criteria for CRPS. He refers to these criteria as the "Budapest Criteria" because they were based on a "consensus workshop" held in Budapest, Hungary. It is not clear how this group was constituted. Harden lists in his August 2010 paper the following persons as participants in the workshop held in Budapest.



Harden testified under oath that the above persons agreed that the Budapest criteria should be adopted as the criteria for making the clinical diagnosis of CRPS.

[ See deposition #2, page 153, line 13 ]

Harden: "And they sat down and very critically assessed what the statistics were telling us and said does this make sense in terms of reality? Does this make sense in terms of really diagnosing these people that we see in our clinics all the time? And to make a long story short after four days of hard work and a lot of jumping up and down and yelling and screaming, if you will, but -- but -- but, finally, at the end of the day we were in agreement and we said yes. These statistically derived criteria are good. They do apply and we should embrace these. We should adopt these as the criteria for making the diagnosis."


In the same deposition, Harden contradicts his testimony that the participants were in agreement that his diagnostic criteria should be adopted. The head of the IASP Taxonomy Committee, John Loeser, MD, participated in the workshop. Loeser's committee is the official body within the IASP that is responsible for adopting diagnostic criteria for CRPS. Loeser recommended further validation studies before adoption:


[ See deposition #2, page 238, lines 7 - 13 ]


Harden: "In fact, the head of the IASP Taxonomy Committee specifically has told us if we accomplish this goal, which we did, and another goal, which we're doing, that he will publish this new criteria in the next book."


In the 2010 paper, Harden reports that the CRPS database checklist for the second validation study is "similar" to the database checklist used for the first validation study. It was not possible for Dr. Kirkpatrick to confirm how similar the data collection processes for the two validation studies were because Dr. Harden refused to provide a copy of the CRPS database checklist for his first validation study upon which the Budapest Criteria are based.  ( See exchanged emails ). Harden's refusal to allow a direct comparison of data collection forms for both studies casts a cloud on his conclusions.

Harden's second validation study was performed on a international population of research subjects. The sensitivity for diagnosing CRPS improved dramatically, reaching nearly 100%, i.e., 0.99 sensitivity. The second validation study for the clinical diagnosis of CRPS left out virtually no patients with CRPS. Although his new result for sensitivity showed dramatic improvement, the results are inconsistent, as noted above, with his first validation study with a lower sensitivity (85% sensitivity) and dramatically different from the larger study from Japan where his criteria were tested and showed only 45% sensitivity for the diagnosis of CRPS.

5. The second author (Stephan Bruelh, PhD) of the 2010 Harden paper wrote a Commentary in PAIN suggesting the discrepancy in the results in Harden's smaller international study compared to the larger Japanese study may be due to random sampling variability and differing clinical referral patterns. Careful examination of the facts, however, reveals a more troubling explanation:

  • Over half of the coauthors listed in Harden's 2010 study published in PAIN come from institutions represented at the August 2003 "consensus workshop" on diagnostic criteria for CRPS held in Budapest. Harden testified under oath that the group attending Budapest workshop agreed that his statistically derived criteria for diagnosing CRPS are good and that they should be "adopted".


[ See deposition #2, page 153, lines 13 - 24 and page 154, 1- 4 ]


  • The fact that some of the study investigators represented in Harden's 2010 paper were individuals already biased in favor of replacing the IASP criteria with the Budapest criteria, together with the fact that the determination of many of the clinical signs of CRPS is subject to interpretation by the investigators (instead of by objective measurement), creates a scenario conducive to corruption in clinical research.


Bruehl wrote:


"All patient-reported symptoms used in CRPS diagnosis are by definition subjective, and even supposedly objective diagnostic signs, such as allodynia, hyperalgesia, and range-of-motion impairments on clinical examination, rely on patient reports of pain and thus are at best only semi-objective."


Dr. Harden describes under oath how the determination of a sign by an investigator is left to the interpretation of the investigator:

See deposition #2, page 163, lines 19 - 25  ]




Dr. Harden, you had just set out for us the Budapest criteria, and I want to make sure I -- I understand -- or the jury understands. Can you tell us what you mean by sensory component of a patient report?


This is subjective presentation by the patient; when it's symptoms it mean things like I have pain, it's burning, it's stinging, it's electrical. If it is signs, it is more in the hands of the investigator such as they will -- they will use a light touch or a brush or a light touch of their hand to see if that causes pain now which is allodynia. They will use a pin and actually do light pin pricks, and the patient will say, you know, that hurts, of course, with a pinprick.



Harden asserts under oath how, using the Budapest clinical diagnostic criteria, the physician's bias might contaminate the findings on physical examination of the patient.


[  See deposition #2, page 169 lines 10 - 16  ]:


 ... although it's a trained medical professional, and they should be able to observe and record accurately, they're, of course, biased at this point by what the patient has told them because we get the history first. So they're already biased before they even go in to do the exam.


The risk of bias being introduced into Harden's studies by study investigators was made worse because the diagnosis was already given to the investigator as either CRPS or non-CRPS prior to examining the research subject. Here is how it works:


1. The study investigator checks the correct box at the top of the data collection form for the diagnosis as follows:

2. The investigator is now committed to a diagnosis prior to identifying signs / symptoms in the research subject. The Budapest Criteria and IASP Criteria are available to the investigator as the subject is examined. Thus the investigator has the opportunity to contaminate the results with their personal bias by simply checking the appropriate box on the checklist form for the sign and symptom that best match, in their opinion, the Budapest or IASP diagnostic criteria. As noted, adding further to the risk of introducing bias is the fact that determining many signs of CRPS are left to the interpretation of the study investigator.


Harden is concealing information that would permit a critical analysis of the integrity of his research. Kirkpatrick sent an email to Harden on August 17, 2010, requesting a copy of the "instructional video" that was referenced in the METHOD section of his 2010 paper in PAIN. The paper claims this video was provided to investigators at study sites around the world to "maximize uniform assessment across sites." Harden refused to provide a copy of the video. (One coauthor of Harden's 2010 paper informed Kirkpatrick that he did not recall ever receiving an instructional video for the study).

Kirkpatrick also requested that Harden provide a blank copy of the data collection form used in the first validation study (the validation scheme as discussed in the Budapest workshop). Harden refused to provide a copy of this data collection form.

When Kirkpatrick asked Harden in an email if it was possible to introduce bias into the data collection process by study investigators, he refused to answer the question. Instead Harden sent Kirkpatrick an email stating, " Please do not email me again."

(The emails exchanged between Kirkpatrick and Harden August 17-18, 2010, are available by clicking here)

Given that Harden's research methodology is subject to investigator bias, given his financial interest in legal cases aimed almost exclusively at showing absence of CRPS based on his diagnostic criteria  [ deposition #2, page 230, lines 1 -20  ], given his 1 day a week spent on patient care [ deposition #2, page 226, lines 8-12 ], given the apparent lack of IRB approval for many of his studies, given Harden’s efforts to conceal information that would permit a critical analysis of the integrity of his research (see exchange of emails with Kirkpatrick) and given the results from his studies over the past decade, it is difficult not to  question the existence of bias, fabrication, falsification and/or manipulation of clinical data.

Harden's first validation study in 1999 left out 15 % (0.85 sensitivity) of subjects with CRPS from the diagnosis of CRPS. When it was attempted to replicate his findings independently in a larger population of research subjects, over half of the subjects with CRPS were left out of the diagnosis of CRPS (0.45 sensitivity). Under these circumstances it is difficult to accept the credibility of his second validation study published in 2010 where virtually no subjects with CRPS are left out of the diagnosis of CRPS (0.99 sensitivity).


Harden's Clinical Diagnostic Criteria




Harden 1999

Replication of Harden's work by Sumitani et al, 2010

Harden 2010



6. Kirkpatrick has reviewed a copy of Harden's 2010 paper as written prior to submission to the journal PAIN. The data in Table 5 of the manuscript, which shows characteristics of CRPS as a function of its spread beyond the initially affected area, was not included in the published version. Omitting the data suggests manipulation to show a more favorable outcome for Harden's Budapest Criteria in making the diagnosis of CRPS.

It is known that the longer the duration of pain with CRPS, the more obvious the diagnosis of the disease. The current diagnostic criteria adopted by IASP generally requires that a certain amount of time elapses after an inciting event before a diagnosis of CRPS can be established. See criteria #2 below.


IASP diagnostic criteria for Complex
Regional Pain Syndrome

(1) The presence of an initiating noxious event, or a cause of immobilization.
(2) Continuing pain, allodynia, or hyperalgesia with which the pain is disproportionate to any inciting event.
(3) Evidence at some time of edema, changes in skin blood flow, or abnormal sudomotor activity in the region of pain.
(4) This diagnosis is excluded by the existence of conditions that would otherwise account for the degree of pain and dysfunction.

Merskey, H. and Bogduk, N., Classification of Chronic Pain: Descriptions of Chronic Pain Syndromes and Definitions of Pain Terms, 2nd edn., IASP Press, Seattle, WA, 1994.


In fact, Harden’s 2010 paper points to the data in Table 5 and concludes that the research subjects with spreading CRPS had a longer duration of pain, and were more likely to demonstrate a very specific objective sign for CRPS, e.g., asymmetrical sweating. Thus the longer the duration of pain, the more likely the research subjects with CRPS will meet the diagnostic criteria for CRPS as required by Harden's diagnostic criteria and thus it becomes more likely the CRPS group will show 100% sensitivity for the diagnosis of CRPS. However, the longer the duration of pain in research subjects with CRPS, the more irrelevant Harden's diagnostic criteria become in making an early diagnosis of CRPS. Early diagnosis of CRPS is essential so that treatments can be administered as soon as possible to alleviate pain and suffering in patients.

One way to manipulate the results of a study in order to show a more favorable outcome for sensitivity in making the diagnosis of CRPS is to recruit only patients with longstanding pain due to CRPS, e.g. mean duration pain of three years. Then the results are manipulated by reporting a shorter duration of pain in the final publication. This appears to be precisely what was done. The data published in Table 1 show a median duration of pain of 14.2 months (N=113) in research subjects with CRPS. The CRPS subjects in Table 5 (not published or disclosed) are from the same population of research subjects as in Table 1. The research subjects in Table 5 were divided into two groups, i.e., "Distal CRPS Only" and "Spreading CRPS". The mean duration of pain when both groups are combined (N=111) is nearly three years or 33 months.

There is further evidence of manipulation of data:

(a) In all three previous publications in PAIN by Harden, including the data in Table 5 that was omitted from Harden's 2010 publication, the duration of pain is presented as a mean in months rather than as a median in months as shown for the first time in Table 1. In addition, the legend to Table 1 states the duration of pain is presented as a mean but the actual data in Table 1 is in fact presented as a median. This manipulation of data in Table 1 suggests an attempt to obscure a direct comparison with previous data and creates the impression of a more favorable outcome, than is justified, for Harden's diagnostic criteria.

(b) In the 2010 publication in PAIN, Harden assessed "acute pain sensitivity" in CRPS and non-CRPS groups by measuring pain due to heat stimulus applied to skin with a probe. The CRPS subjects in Table 1 of the manuscript (not published or disclosed) showed no significant increase in pain due to heat stimulus compared to the non-CRPS subjects. Therefore, the authors concluded that the pain due to heat stimulus was "statistically comparable across groups."

Perhaps recognizing that this conclusion contradicted the data published in Table 2, which shows pain due to any tactile stimulus applied to skin (allodynia) in the CRPS group, the data in Table 1 of the manuscript was altered for publication to show an increase in pain due to heat stimulus applied to the skin in the CRPS group. In fact, the altered data shows a mean difference between the CRPS and non-CRPS groups for increased pain due to heat stimulus more than 4-fold the original value. This manipulation of data results in a statistically significant difference (p < .01) which is a complete fabrication.

Further evidence of manipulation of data in Table 1, as well as concealing this manipulation from others, is an email from the second coauthor Stephen Bruehl PhD to coauthor Timothy Lubenow MD. Lubenow was told by Bruehl not to worry if Harden had left him out of the peer review process because "only a few minor revisions" in the manuscript would be made prior to submission to the journal PAIN for publication.

The addition of two new coauthors (see below), the deletion of Table 5, the manipulation of data to justify conclusions favorable to the study, and the manipulation of data to change conclusions adverse to the study are hardly minor revisions.


 7. On October 16, 2009, a coauthor of Harden's 2010 paper in PAIN was informed by second coauthor Bruehl that only "minor" revisions would be made to the manuscript prior to submitting the document to the journal PAIN for review. One would assume that the study was completed and that the manuscript was simply undergoing minor revisions prior to submission to the journal. However, the final publication reveals that two additional research subjects were added to the CRPS group. This raises the possibility that data was fabricated to show a more favorable outcome using Harden's diagnostic criteria.

8. The journal PAIN requires the following in order to insure peer review prior to submission:


"Manuscripts submitted under multiple authorship are reviewed on the assumption that all authors listed concur with the submitted version of the manuscript and with the listing of the authors"


One of the coauthors of the 2010 Harden paper informed Kirkpatrick that he had not seen the version of the manuscript submitted to PAIN nor was he aware of two additional co-authors added to the paper.

9. The journal PAIN requires the following in order to insure scientific integrity prior to submission:


Acknowledgments. Acknowledgments should be placed at the end of the text before the Reference List and should specify: (1) contributions that need acknowledging but do not justify authorship; (2) acknowledgments of technical help; (3) acknowledgments of financial and material support, specifying the nature of the support; (4) financial arrangements that may represent a possible conflict of interest. This would also include any of the following arrangements, if any of the authors:

      • have a financial relationship to the work;
      • have received any government or company grants or research support;
      • are employees of a company;
      • are consultants for a company;
      • are stock holders of the company;
      • are members of a speakers bureau;
      • have received any other form of financial support.

    A Conflict of Interest statement must be included for all manuscripts within the Acknowledgements section. If there are no conflicts of interest, please explicitly state this.


Harden sees patients only once per week.  He makes himself available in legal matters almost exclusively as a defense expert who is hired to disprove a patient has CRPS. It would seem appropriate, if not mandatory, that he should have made these disclosures in the publication.


Dr.  Harden is advocating that his questionable diagnostic criteria become a standard of care.  However, the criteria are based on flawed and misrepresented research, and their adoption would lead to substandard patient care.  We believe the facts as we have described them demonstrate that they are based on human subject research conducted without required approvals and that the published results were based in part on falsified or fabricated data. In the interests of public safety and scientific integrity, our organization urges that Northwestern University investigate the research referenced above that was conducted by Dr. Harden at your institution.


Our concern for the public safety, the ethical treatment of patients, and the integrity of medical research is such that we must strongly consider sharing our concerns with the public media. 



Additional Issues Brought to the Attention Northwestern University:


On September 12, 2007, in a civil lawsuit styled Robert Hogan, Plaintiff vs. Baptist Medical Center – Nassau, Inc., a Florida Corporation, Defendant and Third-Party Plaintiff vs. American Cancer Society and Gray Gable, Nassau Village Volunteer Fire Department, Inc., Third Party Defendants, 4th Judicial Circuit, Nassau County, Florida (Case No. 06-CA-44), Harden testified under oath in 2 depositions.


Deposition #1Transcript

Deposition #2 Transcript

Deposition #2 (28-Minute Video)




Dr. Harden's goal is to have the International Association for the Study of Pain (IASP) approve his mathematical formula as the international criteria for the diagnosis of CRPS. He uses his formula in court to bolster his claim that the Plaintiff in the litigation does not need treatment for CRPS.

Harden’s formula for diagnosing CRPS appears to be biased and based on questionable data.  Harden testified that he sees patients only once a week, and makes himself available for hire in legal matters primarily as a defense expert to prove a patient does not suffer from CRPS.  The diagnostic formula advanced by Harden makes his job as an expert for the defense much easier, to his personal and financial benefit. 



Northwestern University has been asked to explain how Dr. Harden could be promoted to Associate Professor without Board Certification in a medical specialty.


According to Marcie B. Weiss, Assistant Dean for Faculty Affairs, Northwestern University policy requires that faculty obtain Board Certification in the field associated with the faculty member’s clinical department before being promoted from Instructor to Assistant Professor.

Dr. Harden stated falsely under oath (page 121) that he is "Board Certified". He is not Board Certified in any field of medicine.

If Dr. Harden was granted promotion to Associate Professor of Physical Medicine and Rehabilitation based on false pretense, his ethics should be questioned. Northwestern University should be questioned for giving him recognition before the public for clinical competency that he does not deserve. Clinical competency is at issue in this complaint.



International Research Foundation for RSD / CRPS, Inc.


Richard Hoffman, PhD


Jerry Trachtman, ESQ

Secretary & General Counsel

Anthony Kirkpatrick, MD, PhD



ADDENDUM: On January 21, 2011, Dr. Kirkpatrick received an email from Kathy Kreiter, Executive Director, IASP, stating:

"The Council has concluded that Dr. Harden did not misrepresent IASP and that there is no good cause to revoke Dr. Harden’s membership."