Due to the imprecise definition of Chronic Lyme Disease and the inherent self-selection bias, how could psychological factors affect your results?
Viewed 236 times

I read with great interest this article by Johnson et al. and have the following comment.

As mentioned in this section, by selecting participants from Lyme Disease support forums you may only be finding the sickest individuals. But, due to condition of study and the methods employed, individuals with underlying psychological issues may also be more likely to be identified and sampled. The subjective nature of the HRQoL makes this a potential explanation for the observed results. reference Specifically, one could alternatively interpret this study as showing that individuals self-identified as Chronic Lyme Disease patients perceive a reduced quality of life and greater symptom burden. Hence, the principal findings may be more a consequence of the sample identified than reflecting the actual role of Lyme Disease in these processes. Such bias would also greatly affect the comparisons made with other populations.

I feel that the potential role of this different kind of selection bias should have been included in the limitations section.

Thank You

waiting for moderation

I want to make a point of clarification regarding my previous comment. In this study specifically, due to the way in which patients were examined, through use of a web-based questionnaire, there are several potential reasons for the observation of symptoms persisting among individuals previously treated for Lyme Disease. This is true, even with test documentation. 1) misdiagnosis of initial Lyme Disease; 2) illness unrelated to Lyme Disease or initial treatment; 3) post-treatment Lyme Disease Syndrome (PTLDS); 4) persistent spirochetal infection. These groups are extremely difficult to differentiate and I do not pretend to know the actual distribution, but in the given sample it is possible that each of these is represented. For those suffering from psychological illnesses (groups 1 and 2 particularly), it could be difficult to ascertain a reliable estimate of quality of life and symptom burden, especially through use of a web-based survey. As noted by the authors, quality of life for PTLDS patients has been previously examined and this group does indeed suffer impairment, but it is the other groups for which the accuracy of this scale is uncertain.

Regardless of the underlying etiology, these patients are suffering greatly and are in dire need of advocacy, as they represent an extremely marginalized population. Despite the limitations mentioned, these results do suggest that quality of life impairments for these individuals may be more dramatic than previously appreciated. I feel it is extremely damaging to make the assertion that symptoms are all "in one's head" as this detracts from the actual issues at hand, and that is not what I meant to communicate. In closing, the point of my original comment was simply to discuss a potential limitation, not to comment beyond the scope of the article.

Thank You

waiting for moderation
1 Answer
Accepted answer

We appreciate Dr. Donnelly’s interest in our study of patients with chronic Lyme disease (CLD). The problem of selection bias in internet surveys is discussed in some depth in the limitations section of our article. It seems likely that most patients heard about the survey via the internet through various Lyme support groups throughout the nation. We are not aware of any studies indicating that patients on the internet are more likely to have psychological issues.

On-line data gathering involving internet-based communities is becoming common. For example, PatientsLikeMe is an internet-based community that gathers disease-related data from patients for analysis and publication. Many patient organizations, particularly those with diverse populations or less common conditions, have formed data registries just for this purpose. The use of patient-driven “big data” has also caught the attention of the government. For example, the government-sponsored Patient Centered Outcomes Research Institute (PCORI) is funding a $100 million research network comprised of 18 patient powered research groups to investigate diseases such as emphysema, multiple sclerosis, epilepsy, and Crohn’s disease.

Patient-centered research relies on patient self-reported outcomes such as the HRQoL metric used in our survey. Despite its subjective nature, HRQoL has been found to be a more powerful predictor of mortality and morbidity than many objective measures of health (1). Studies have found that CDC HRQoL measures are consistent in terms of content, construct and criterion validity, and test-retest reliability (2).

Although patients who are no longer ill or only mildly ill are not as likely to be involved in on-line communities or be motivated to take this type of survey, our results included a broad spectrum of patients from around the country, as indicated by the demographics in Table 2. It should be noted also that because the normal distribution of CLD is unknown and has not been studied, all sample populations used to study this disease have potential for reflecting a segment rather than the full spectrum of the disease. The full spectrum of Lyme disease is not captured by randomized controlled trials, insurance pools, or laboratory tests. For example as we point out, one randomized controlled trial screened 3,368 people to yield a sample population of 36. Samples this small are not generalizable and do not permit subgroup analysis.

The CDC recently raised its estimate of the annual number of new Lyme disease cases diagnosed in the US from 30,000 to 300,000 using survey data, insurance data, and laboratory data. Their surveillance system captures only 1 in 10 cases. We do not know what the other 9 cases look like. The point is that, until we know the normal distribution of patients with CLD, all sample populations used to study CLD will remain problematic. Further, we cannot know the normal distribution of the disease without the type of research we have conducted here that seeks to characterize this population despite the inherent sample limitations in all studies of Lyme research today.

We are quite sensitive to the possibility that patients may be misdiagnosed with Lyme disease and actually have some other disease, and nobody knows how commonly this occurs. Equally unknown is the number of patients who are diagnosed with other diseases but who actually have Lyme disease. The lack of a test that is both highly sensitive and highly specific makes a definitive diagnosis of Lyme disease difficult even for the clinician, who frequently must rely on clinical judgment in diagnosing tickborne illness.

To address this problem, we asked a number of questions intended to narrow the range of respondents included in the survey results. Hence, patients had to be clinically diagnosed and have positive Lyme serology to be included in the study. In addition, we compared an even narrower group of clinically diagnosed patients who also had an erythema migrans rash or two-tiered CDC-positive Lyme serology with those who were clinically diagnosed with other supporting laboratory tests. We did not find a substantial difference between the groups, as shown in Table 4 of our report.

Dr. Donnelly questions the impact of physical versus psychological symptoms in our web-based survey. Patients with chronic physical diseases frequently have psychological manifestations. Indeed, it is known that patients with psoriasis, leprosy, multiple sclerosis, diabetes and rheumatoid arthritis, for example, have a greater prevalence of depression than the general population. A recent article explains that proinflammatory cytokines can cause depression in those who suffer chronic illness (3). However, a dominance of physical distress over mental distress has been noted in a number of chronic diseases including asthma, obesity, diabetes and arthritis (4). We observed the same dominance of physical distress over mental distress in this sample of patients with CLD.

Although not included in the publication, we looked at how patients with emotional problems and an activity limitation (“Emotional Patients”) from the Richardson study (5) compared with our population and found substantial differences between the two populations. While poor mental days dominate in Emotional Patients, poor physical days dominate in CLD. Patients with emotional problems reported only 8 days of poor physical health per month, while those with CLD reported 20 days. Patients with CLD had increased numbers of pain days, activity limitation days, and unrested days compared to Emotional Patients. Conversely, Emotional Patients experienced more depressed and anxious days compared to CLD patients. This observation conforms with other chronic diseases rooted in physical illness.

Fair/Poor General Health Status: 72% CLD patients vs. 40.5% Emotional Patients Poor physical health: 20 days CLD patients vs. 8 days Emotional Patients Poor mental health: 15.5 days CLD patients vs. 17.9 days Emotional Patients Activity limitations: 16.8 days CLD patients vs. 11.9 days Emotional Patients Pain limited activity: 16.5 days CLD patients vs. 8.8 Emotional Patients Sad, blue, depressed: 12.4 days CLD patients vs. 16.9 days Emotional Patients Worried, tense or anxious: 15.8 days CLD patients vs. 19.7 Emotional Patients Unrested: 20.3 days CLD patients vs. 15 days Emotional Patients Full of energy: 3.5 days CLD patients vs. 7.7 days Emotional Patients

We appreciate Dr. Donnelly’s comment regarding the unproductiveness and harm caused by marginalizing and stigmatizing sick patients. Unfortunately, there is substantial medical specialty bias in Lyme disease, with differing viewpoints held by the Infectious Diseases Society of America (IDSA) and the International Lyme and Associated Diseases Society (ILADS). Both groups have published guidelines regarding the diagnosis and treatment of Lyme disease. IDSA does not believe that infection can persist in Lyme disease patients despite a substantial number of studies to the contrary and despite the profound impact on quality of life caused by tickborne disease. Because IDSA denies the possibility of persistence, some members take the position that the symptoms are most likely “all in one’s head”. This viewpoint has led to marginalization of many patients with significant physical and emotional quality of life issues related to Lyme disease, as outlined in our survey. We agree that it is important to improve the quality of life in these patients.


  1. http://www.health.ny.gov/statistics/disabilities/chart/docs/2007_disability_chartbook.pdf
  2. http://www.cdc.gov/hrqol/measurement_properties/andresen1.htm
  3. http://www.nature.com/nrn/journal/v9/n1/full/nrn2297.html
  4. http://www.cdc.gov/pcd/issues/2011/jan/pdf/09_0234.pdf
  5. http://onlinelibrary.wiley.com/doi/10.1002/cncr.23291/pdf
waiting for moderation