The Klempner Article

By Robert Bransfield, M.D.


Stephen Brand, Ph.D.



The recent article in the NEJM, “Two Controlled Trials of Antibiotic Treatment in Patients with Persistent Symptoms and a History of Lyme Disease,” by Klempner, et. al., provides some interesting data, but the proper interpretation of this data is of little relevance to both clinical practice and guidelines related to chronic Lyme disease. It does, however, provide some interesting insight about the significance of PCR and IgG testing in Lyme disease and it demonstrates a poor quality of life is associated with a history of Lyme disease. It is also an excellent example of the failure of objectivity in a peer reviewed article in a prestigious medical journal followed by misperceptions of the significance of this article by the media, which results in the improper use of this article by the insurance industry. An objective review of this article is one issue, while the role of this article towards perpetuating false beliefs about Lyme disease is an additional and more significant issue.

Medicine today is much like law. Both have adversarial qualities. Information is presented in a manner that is most effective in arguing a point of view. When this technique is effective, guilt is proven to be innocence; black is proven to be white, etc. The technique is always the same. There is a series of statements and conclusions. Each statement is a 10-degree twist on the truth, so small that it can go below the radar of our ability to detect deception. After a series of complicated and related statements, there is a 180-degree twist to the truth.

The twists to this article began many years ago, when a few rheumatologists involved in Lyme disease incorrectly attempted to apply their observations of acute Lyme arthritis to explain late-stage Lyme disease and Lyme neuroborreliosis. The problem was compounded when Lyme disease was conceptualized from the perspective of an acute infection rather than recognizing the persistent relapsing nature of this microbe. These and many other similar twists set the stage for biasing the researchers to confirm their prior views on the subject. A failure to recognize and comprehend that tick-borne diseases were often complex interactive infections involving multiple stealth and persistent pathogens as well as an excessive reliance upon molecular mimicry hypotheses further added to the problem. As a result of these and other errors, a twenty-first century problem is approached with a nineteenth century view of infectious disease. The bias was further solidified as patent issues, competing diagnostic systems, research grants, reputations, lawsuits, and a need to defend prior consultant decisions provided to insurance companies by high paid consultants further clouded scientific objectivity. Anyone coming from this perspective would be highly motivated to design a study arguing that persistent infection was not a significant issue in Lyme disease, and this is exactly what happened.

The study design guaranteed the study would fail. The design of the research protocol is the most critical step in the project. This study had very restrictive inclusionary and exclusionary criteria. The vast majority of patients with chronic Lyme disease do not meet the criteria for inclusion into the study, therefore the findings and stated conclusions have no relevance to most patients with chronic Lyme disease (CLD.) The other two NIH sponsored trials had much better study designs. I recall sitting in a meeting several years ago at NIH when the study design of the NIH intramural study was discussed. A pyramid was used as an analogy. At the bottom of the pyramid were the patients who had some of the features of Lyme disease, with disagreement about whether or not these patients truly had Lyme disease. The intent was to select the patients at the tip of the pyramid who clearly had Lyme disease. This strategy was NOT used in the Klempner study. Patients who demonstrated DNA evidence for the presence of Borellia burgdorferi by PCR testing were excluded from the study. There was only one short sentence in the journal article that discussed this, and this was the most significant sentence in the entire article. PCR testing has been considered absolute proof in the legal system. Why would a PCR positive patient ever be excluded? It was stated that it would have been unethical to give a placebo rather than the antibiotic treatment to a PCR positive patient. This is a clear endorsement that positive PCR testing is considered confirmation of persistent infection with Bb. Depression, a major symptom of the neuropsychiatric manifestations of Lyme disease, was also an exclusionary criterion. Even though both PCR and depression were exclusionary criteria, the status of the study participants was measured by PCR testing on the spinal fluid and the Beck Depression Scale. Other means of evaluating the status of the patients were the SF-36 and other testing. Basically, the wrong tests and tests that lacked objectivity were used to evaluate the status of the patients. Although there were 12 authors, none were psychiatrists. Why? Neuropsychiatric symptoms are a major part of the later manifestations of Lyme disease. How can a study, which is supposed to study late stage disease, not have the input of a psychiatrist with experience in working with Lyme disease? The neuropsychological scales that were implemented in the study were notably insufficient to adequately assess the psychiatric symptoms and the executive functioning impairments, which are associated with persistent Lyme disease infections. The SF-36 scale was used. The scores were quite low, reflecting a poor quality of life for these patients. In addition, the SF-36 is only a subjective scale based solely upon patient opinion and contains no objective criteria for the assessment of patients’ status.

Rather than PCR, a positive IgG Western Blot test based upon five bands at only one laboratory was used as a reference point. No studies have ever demonstrated whether the absence of IgG reactivity had diagnostic significance in late-stage disease.

The study was designed to provide patients with “intensive antibiotics.” Although one month of ceftriaxone and two months of doxycycline would be considered intensive for the treatment of many acute infectious diseases, few physicians who shoulder the responsibility of treating chronic Lyme disease would consider this to be an adequate retreatment for a patient with late-stage Lyme disease who had failed prior courses of significant treatment. Results from recent scientific studies on dog models by Strubinger of chronic Lyme disease indicate persistent infection even after six months of antibiotic treatment. Furthermore, experience in the treatment of human infections caused by other persistent bacteria, such as Treponema pallidum, Mycobacterium tuberculum, and Helicobacter pylori demonstrates that prolonged antibiotic therapy and/or different combinations of antibiotics would result in greater improvement than was observed in this study. Additionally, while the study tested patients for tick-borne coinfections like babesiosis or ehrlichiosis, it failed to offer any treatment strategies.

The placebo and treatment groups appeared to start with significantly different scores on the primary outcome measures. In the group of seropositive patients, those in the placebo group had significantly better scores on the MOS Cognitive measures and exhibited fewer neurocognitive symptoms at base-line than the patients in the antibiotic group. In addition, within the seronegative group patients in the placebo group had significantly poorer scores on the SF-36 Mental Component, the MOS Pain scale, and the Fibromyalgia Impact Questionnaire. The categorization of patients into Improved, Unchanged, and Worse on each of the outcome components seems to lose a certain amount of information about the magnitude of change in functioning from base-line until post-treatment and to be lacking in statistical power compared with other procedures for analyzing these data. In view of these two points, it would seem advisable to utilize Analysis of Covariance procedures to measure the impact of treatment on the outcome measures after adjusting for patients’ baseline scores on the measures of functioning. This procedure would take into consideration the baseline differences between groups, would treat change in the outcome measures as a continuous rather than a categorical variable, and would afford more statistical power in testing differences between the placebo and antibiotic groups. Since the symptoms of chronic Lyme disease are different in different patients, some patients presented impairments primarily in specific areas of functioning and change from treatment would be expected in the specific areas that were problematic for each patient. An analysis of change in the most problematic items on each of these scales would provide a more sensitive assessment of potential treatment effects. In addition, the small sample size raises concerns about statistical power and further undermines the validity of the study.

The 35% improvement threshold when combined with the other study design constraints would predict a failure even before initiating the study. It should have been necessary to adjust the status assessment to make allowances for the Jarish-Herxheimer rection, but this was not a part of the study design. It would be important to know more details of the NIH NIAID Data Safety and Monitoring Board’s termination of the study.

Since the patients with seronegative Lyme disease showed better improvement on antibiotics than their seropositive counterparts, it appears that negative IgG reactivity is not a useful criterion to deny the presence of persistent infection. If the investigators feel persistent infection was not a cause of the patient’s current symptoms, then what else is a truly more plausible explanation?

A disclaimer reviewing the potential conflicts of interests of the authors was absent from the article and would have been most appropriate in light of the significant political, financial, and ethical controversies surrounding chronic Lyme disease. This should have included disclosures of the authors’ Lyme disease-related insurance consultations, roles as defendants in Lyme-related lawsuits, and their roles as paid expert witnesses as well as their ownership of Lyme disease-related patents, commercial diagnostic systems, and their financial vaccine interests.

Along with three other related articles, the Klempner article was distributed on the Internet a month before the date of publication. The reason for the early release is a subject of considerable speculation. When the article was released, there was a well-orchestrated flood of articles in the media, which mostly interpreted the significance of the article out of context. Since the release, insurance companies have already used the article as proof to deny coverage for the treatment of Lyme disease.

In summary, the design of the research and the manner in which the article was written and promoted in the media even before it was published reflects many of the biases that have obstructed scientific objectivity and medical progress surrounding Lyme disease for decades. The study demonstrates that patients with a history of exposure to Lyme disease in the past who probably do not have active infection at the time of the study as demonstrated by negative PCR’s and a history of significant antibiotic treatment in the past, do not respond to inadequate antibiotic treatments as measured by incorrect and highly subjective criteria that are not representative of symptoms associated with CLD administered by physicians who appear to not know how to treat Lyme disease, some of whom have strong conflicts of interest to the contrary in a study that was never completed. In short, the study proves very little. It does, however, demonstrate that it is very easy to waste large amounts of Federal money for something of little benefit to taxpayers. In a backhanded way, it validates PCR testing as proof of persistent infection. It suggests that IgG testing by Dearborn criteria is not a valid criterion to either confirm or refute the presence or absence of active infection. Most importantly, it demonstrates that false beliefs can be self-perpetuating, information in the peer reviewed medical literature can be highly inaccurate and deceptive, and it is very easy for medical information to be improperly interpreted by the media, insurance industry, and potentially by the legal system.