ME/CFS Skeptic
Senior Member (Voting Rights)
I’ve decided to take a closer look at the evidence for chronic Lyme disease. I’ve chosen to skip issues relating to the persistence of the bacterium Borrelia burgdorferi and the use of prolonged antibiotic therapy, to focus on basic epidemiology. The main question I wanted to answer is whether there is an increased incidence of chronic debilitating symptoms such as fatigue, pain and cognitive difficulties following Lyme disease compared to a control group that didn’t experience Lyme disease. In my view, if the answer to this question is no, all other discussions on the etiology and treatment of chronic Lyme disease become redundant as the diagnosis might not be the correct one. For those who might not be able to read through the whole text, my conclusion is that the current epidemiological evidence for chronic Lyme disease is conflicting and unconvincing.
Post-treatment Lyme disease syndrome (PTLDS): the question not the answer
When I explain my question people usually refer me to studies on Post-Treatment Lyme Disease Syndrome (PTLDS) such as the one by Rebman and colleagues [1]. But this isn’t exactly what I’m looking for. These studies often show that patients who receive a PTLDS diagnosis are severely disabled, but don’t demonstrate that this disability is related to Lyme disease. The problem with many studies on PTLDS is that they start by defining the syndrome and then recruit patients to study it, which results in selection bias. It’s a bit like defining ‘post-vaccine autism syndrome’, then recruiting persons who developed autism following vaccination and declaring that the syndrome is real if persons meet the definition. What I would like to know is whether having Lyme disease increases the risk of developing the type of severe disability seen in those PTLDS-studies. To answer that question we need proper epidemiological studies that look at representative samples of patients with Lyme disease and that follow these patients in a longitudinal design to see if they develop more disabling symptoms than a matched control group.
The Nancy Shadick studies in Boston: some evidence suggestive of PTLDS
There are a few studies like that and the early ones did suggest a connection between experiencing Lyme disease and subsequent chronic symptoms. The first prominent study was published in 1994 in Annals of Internal Medicine by Nancy Shadick and colleagues. [2] It was a study conducted in Boston, Massachusetts where they followed up on 38 Lyme patients and matched controls for a period of more than 6 years. The authors reported that patients who had Lyme disease experienced more symptoms than the controls, including arthralgia, concentration difficulties and difficulty sleeping.
There were some interpretation problems though. No correction for multiple comparisons was performed and it’s possible that the difference for prominent symptoms such as fatigue (26% compared with 9%; P = 0.04) would no longer be statistically significant if such corrections were applied. Another problem is recall bias: the tendency to report symptoms could be affected by the history of patients. The patients in this study contracted Lyme disease in the 1980s when the recognition of the illness was poor. Few received the recommended antibiotic therapies and some had months to years of disease before treatment. As one commenter noted: “patients in this study, informed that they received treatment for Lyme disease before optimal treatment protocols were established, might recall aches and pains that others might dismiss.” Finally, the result focused on the prevalence of reported symptoms in the Lyme versus control group, without indicating the impact of those symptoms. The symptoms were corroborated by neurocognitive testing (the Lyme group had more neurocognitive deficits than the control group) but the SF-36 global health scores seem to indicate differences in functional capacity were rather small (although I find it hard to interpret those scores, which I assume are raw scores on a scale of 30?)
The sample size of this first study was also quite small. In 1999 Shadick et al. published a larger study where they followed up on 186 persons who had a history of Lyme and 167 controls that did not. [3] Once again the former reported more symptoms than the latter. The main problem with this study, however, is that it is retrospective: patients weren’t examined clinically and then followed up, instead the authors mailed random persons to find a sample that did and didn’t experience Lyme disease based on medical records. As the authors note, this design “lacked the clinical certainty of case status that is seen in a prospective analysis.”
And that’s a bigger issue than it might seem because blood tests for Lyme disease are often false positive, especially before 1995 when U.S. Public Health Service recommended both two-tier testing and the use of evidence-based criteria for interpreting immunoblots. So what might have happened is that some patients with chronic fatigue syndrome, fibromyalgia or other conditions that resemble Lyme disease were misdiagnosed because of a false positive blood test.
The study recruited patients from Nantucket Island, one of the highest reported incidences of Lyme disease in the United States. So it’s not unlikely that physicians thought of Lyme disease when a patient presents with unexplained symptoms of fatigue, pain or cognitive dysfunction. In some cases, this could have resulted in a false positive serology. These patients will not have gotten better with antibiotic treatment and thus give a false impression of persistent Lyme disease.
Although the two Shadick studies are frequently cited, they aren’t convincing evidence that more people remain significantly disabled following Lyme disease. At the time there was also the report by Wang et al. that 26 children with prior Lyme disease did not have a higher prevalence musculoskeletal or neurological symptom than controls. [4] So what was needed were larger and better-conducted studies.
The Eugene Shapiro study in Connecticut
The next important study was published in 2000 in JAMA by the research team of Eugene Shapiro. [5] It followed-up on a random subsample of 212 patients selected from a large sample of all reports of Connecticut residents with suspected Lyme disease submitted to the Department of Public Health from 1984-1991. There is a risk of selection bias because for a large proportion contact information was not available and there’s a problem with the control group having more females (66% versus 49%). Despite these shortcomings, this study seems like one of the best in its kind. The results showed that 9.0% of patients believed they were still not cured of Lyme disease. When researchers made the comparison with the healthy group, however, there were practically no differences in impairment. The authors report:
So the results of the highest quality study suggested that while there might be lingering symptoms these did not lead to a clear increase in disability compared to a control group, as one would expect.
Allen Steere and the patients from Lyme
Following reports of chronic disability after Lyme Disease, Allen Steere, the man who gave the disease its name, decided to contact some of the original study patients from the Lyme, Connecticut, region. He randomly selected 84 patients so that there were three Lyme groups characterized by either facial palsy, Lyme arthritis or confirmed erythema migrans. These were then compared to a control group of 30 healthy persons. There were some mild residual deficits in facial nerve function and bodily/joint pain in the facial palsy group but overall, the three Lyme groups and the control group did not differ significantly in current symptoms or neuropsychological test results. On the 8 scales of mental and physical health of the SF-36 questionnaire, the three patient groups and the control group had similar results that compared favorably with normative data scores from the general US population. [6]
The Gary Wormser study in New York
The next important study comes from the research team of Gary Wormser in New York. What is remarkable about this study is that it had very strict selection criteria: patients not only had to have erythema migrans, the characteristic rash associated with Lyme disease but they also took skin or blood samples to culture the B. burgdorferi and make sure the spirochetes were really there. This way the problem with false-positive antibody testing is out of the way.
The first results were published in 2003. [7] Only 8 of the 81 cases who were followed for 1 year were symptomatic at their last visit. The patients were described as follows:
So while this study did not have a sex- and age-matched control group, it does provide some evidence against the hypothesis that persons who experienced Lyme disease are at increased risk of long-term disability as usually seen in patients diagnosed with PTLDS or chronic Lyme disease.
In 2015 results were reported of a 15.4 years follow-up of 100 patients of the same cohort. In separate publications, the authors explain that there was no increased incidence of severe fatigue [8] or fibromyalgia [9] and that the SF-36 scores were similar to those of the general population. [10] One might think that the patients who developed severe fatigue and disability due to Lyme disease might be too ill to participate in the study but patients with severe fatigue due to other reasons than Lyme did show up at assessments, so I’m not sure that this could explain the results.
The Wisconsin study by Kowalski and colleagues:
Then there’s the study by Todd Kowalski and colleagues in Wisconsin, published in 2010. [11] This was the biggest study with 607 patients involved, but just like the Shadick et al. 1999 study it was also retrospective: the authors simply looked at the charts for ICD-9 diagnosis of Lyme disease and then sent up SF-36 questionnaires to see how these patients were doing. Patients with late Lyme disease (eg, arthritis and late neurologic disease) were excluded from this study. The results showed, in short, that treatment failure was rare and participants had a good prognosis.
In a separate publication (Jares et al. 2014) [12] where patients from this cohort with a reinfection were studied more closely, there was a comparison of a sample of 60 Lyme patients (15 reinfected and 45 infected only once) and a healthy control group, showing no major differences in the reported frequency of symptoms. Unfortunately, the comparison we are interested in is somewhat obscured because the authors had split up the Lyme group.
The large Slovenian treatment trial
Also published in 2010 was a large Slovenian treatment trial of Lyme disease where the authors were smart enough to include a healthy control group. [13] 285 patients with adult patients with erythema migrans were included in the study. The results showed that “at both 6 and 12 months, the frequency of new or increased symptoms in patients with erythema migrans did not exceed the frequency of such symptoms in a control group of individuals of similar gender and age without a clinical history of Lyme disease.” And: “Of the 5 patients with new or increased symptoms at 12 months, none qualified to have “post-Lyme disease syndrome,” because the reported symptoms were not severe enough to result in a reduction in previous levels of patient activities.”
Post-treatment Lyme disease syndrome (PTLDS): the question not the answer
When I explain my question people usually refer me to studies on Post-Treatment Lyme Disease Syndrome (PTLDS) such as the one by Rebman and colleagues [1]. But this isn’t exactly what I’m looking for. These studies often show that patients who receive a PTLDS diagnosis are severely disabled, but don’t demonstrate that this disability is related to Lyme disease. The problem with many studies on PTLDS is that they start by defining the syndrome and then recruit patients to study it, which results in selection bias. It’s a bit like defining ‘post-vaccine autism syndrome’, then recruiting persons who developed autism following vaccination and declaring that the syndrome is real if persons meet the definition. What I would like to know is whether having Lyme disease increases the risk of developing the type of severe disability seen in those PTLDS-studies. To answer that question we need proper epidemiological studies that look at representative samples of patients with Lyme disease and that follow these patients in a longitudinal design to see if they develop more disabling symptoms than a matched control group.
The Nancy Shadick studies in Boston: some evidence suggestive of PTLDS
There are a few studies like that and the early ones did suggest a connection between experiencing Lyme disease and subsequent chronic symptoms. The first prominent study was published in 1994 in Annals of Internal Medicine by Nancy Shadick and colleagues. [2] It was a study conducted in Boston, Massachusetts where they followed up on 38 Lyme patients and matched controls for a period of more than 6 years. The authors reported that patients who had Lyme disease experienced more symptoms than the controls, including arthralgia, concentration difficulties and difficulty sleeping.
There were some interpretation problems though. No correction for multiple comparisons was performed and it’s possible that the difference for prominent symptoms such as fatigue (26% compared with 9%; P = 0.04) would no longer be statistically significant if such corrections were applied. Another problem is recall bias: the tendency to report symptoms could be affected by the history of patients. The patients in this study contracted Lyme disease in the 1980s when the recognition of the illness was poor. Few received the recommended antibiotic therapies and some had months to years of disease before treatment. As one commenter noted: “patients in this study, informed that they received treatment for Lyme disease before optimal treatment protocols were established, might recall aches and pains that others might dismiss.” Finally, the result focused on the prevalence of reported symptoms in the Lyme versus control group, without indicating the impact of those symptoms. The symptoms were corroborated by neurocognitive testing (the Lyme group had more neurocognitive deficits than the control group) but the SF-36 global health scores seem to indicate differences in functional capacity were rather small (although I find it hard to interpret those scores, which I assume are raw scores on a scale of 30?)
The sample size of this first study was also quite small. In 1999 Shadick et al. published a larger study where they followed up on 186 persons who had a history of Lyme and 167 controls that did not. [3] Once again the former reported more symptoms than the latter. The main problem with this study, however, is that it is retrospective: patients weren’t examined clinically and then followed up, instead the authors mailed random persons to find a sample that did and didn’t experience Lyme disease based on medical records. As the authors note, this design “lacked the clinical certainty of case status that is seen in a prospective analysis.”
And that’s a bigger issue than it might seem because blood tests for Lyme disease are often false positive, especially before 1995 when U.S. Public Health Service recommended both two-tier testing and the use of evidence-based criteria for interpreting immunoblots. So what might have happened is that some patients with chronic fatigue syndrome, fibromyalgia or other conditions that resemble Lyme disease were misdiagnosed because of a false positive blood test.
The study recruited patients from Nantucket Island, one of the highest reported incidences of Lyme disease in the United States. So it’s not unlikely that physicians thought of Lyme disease when a patient presents with unexplained symptoms of fatigue, pain or cognitive dysfunction. In some cases, this could have resulted in a false positive serology. These patients will not have gotten better with antibiotic treatment and thus give a false impression of persistent Lyme disease.
Although the two Shadick studies are frequently cited, they aren’t convincing evidence that more people remain significantly disabled following Lyme disease. At the time there was also the report by Wang et al. that 26 children with prior Lyme disease did not have a higher prevalence musculoskeletal or neurological symptom than controls. [4] So what was needed were larger and better-conducted studies.
The Eugene Shapiro study in Connecticut
The next important study was published in 2000 in JAMA by the research team of Eugene Shapiro. [5] It followed-up on a random subsample of 212 patients selected from a large sample of all reports of Connecticut residents with suspected Lyme disease submitted to the Department of Public Health from 1984-1991. There is a risk of selection bias because for a large proportion contact information was not available and there’s a problem with the control group having more females (66% versus 49%). Despite these shortcomings, this study seems like one of the best in its kind. The results showed that 9.0% of patients believed they were still not cured of Lyme disease. When researchers made the comparison with the healthy group, however, there were practically no differences in impairment. The authors report:
“Although many patients who had been diagnosed as having Lyme disease reported increased symptoms or increased difficulties with typical daily activities, the proportions were similar to those reported by the matched controls, except that there were statistically significant differences between the groups in the categories of ‘joint or muscle pain’ and ‘ability formulating ideas.’ Likewise, there were no significant differences in the entire cohort between the cases and the controls in the results of the SF-36 or the CES-D.”
So the results of the highest quality study suggested that while there might be lingering symptoms these did not lead to a clear increase in disability compared to a control group, as one would expect.
Allen Steere and the patients from Lyme
Following reports of chronic disability after Lyme Disease, Allen Steere, the man who gave the disease its name, decided to contact some of the original study patients from the Lyme, Connecticut, region. He randomly selected 84 patients so that there were three Lyme groups characterized by either facial palsy, Lyme arthritis or confirmed erythema migrans. These were then compared to a control group of 30 healthy persons. There were some mild residual deficits in facial nerve function and bodily/joint pain in the facial palsy group but overall, the three Lyme groups and the control group did not differ significantly in current symptoms or neuropsychological test results. On the 8 scales of mental and physical health of the SF-36 questionnaire, the three patient groups and the control group had similar results that compared favorably with normative data scores from the general US population. [6]
The Gary Wormser study in New York
The next important study comes from the research team of Gary Wormser in New York. What is remarkable about this study is that it had very strict selection criteria: patients not only had to have erythema migrans, the characteristic rash associated with Lyme disease but they also took skin or blood samples to culture the B. burgdorferi and make sure the spirochetes were really there. This way the problem with false-positive antibody testing is out of the way.
The first results were published in 2003. [7] Only 8 of the 81 cases who were followed for 1 year were symptomatic at their last visit. The patients were described as follows:
“Their symptoms tended to be intermittent, with only 3 patients (4%) consistently symptomatic at each follow-up visit. Symptoms were also generally mild; in 5 (71%) of the 7 evaluable cases, individual symptoms at the last follow-up visit scored less than 3 (out of 8) on the visual analog scale.”
So while this study did not have a sex- and age-matched control group, it does provide some evidence against the hypothesis that persons who experienced Lyme disease are at increased risk of long-term disability as usually seen in patients diagnosed with PTLDS or chronic Lyme disease.
In 2015 results were reported of a 15.4 years follow-up of 100 patients of the same cohort. In separate publications, the authors explain that there was no increased incidence of severe fatigue [8] or fibromyalgia [9] and that the SF-36 scores were similar to those of the general population. [10] One might think that the patients who developed severe fatigue and disability due to Lyme disease might be too ill to participate in the study but patients with severe fatigue due to other reasons than Lyme did show up at assessments, so I’m not sure that this could explain the results.
The Wisconsin study by Kowalski and colleagues:
Then there’s the study by Todd Kowalski and colleagues in Wisconsin, published in 2010. [11] This was the biggest study with 607 patients involved, but just like the Shadick et al. 1999 study it was also retrospective: the authors simply looked at the charts for ICD-9 diagnosis of Lyme disease and then sent up SF-36 questionnaires to see how these patients were doing. Patients with late Lyme disease (eg, arthritis and late neurologic disease) were excluded from this study. The results showed, in short, that treatment failure was rare and participants had a good prognosis.
In a separate publication (Jares et al. 2014) [12] where patients from this cohort with a reinfection were studied more closely, there was a comparison of a sample of 60 Lyme patients (15 reinfected and 45 infected only once) and a healthy control group, showing no major differences in the reported frequency of symptoms. Unfortunately, the comparison we are interested in is somewhat obscured because the authors had split up the Lyme group.
The large Slovenian treatment trial
Also published in 2010 was a large Slovenian treatment trial of Lyme disease where the authors were smart enough to include a healthy control group. [13] 285 patients with adult patients with erythema migrans were included in the study. The results showed that “at both 6 and 12 months, the frequency of new or increased symptoms in patients with erythema migrans did not exceed the frequency of such symptoms in a control group of individuals of similar gender and age without a clinical history of Lyme disease.” And: “Of the 5 patients with new or increased symptoms at 12 months, none qualified to have “post-Lyme disease syndrome,” because the reported symptoms were not severe enough to result in a reduction in previous levels of patient activities.”
Last edited: