In the JournalsPerspective

Lyme disease diagnosis based on clinician suspicion often inaccurate

Clinician suspicion had minimal accuracy in differentiating between children with and without Lyme disease who presented to the ED, according to findings published in Pediatrics.

“Clinicians use clinical prediction rules to combine available demographic, clinical and laboratory factors to estimate the probability of an outcome and assist clinical decision-making. However, the considerable overlap between Lyme disease and its mimics limits the clinical applicability of these predictive models,” Lise E. Nigrovic, MD, MPH, from the division of emergency medicine at Boston Children’s Hospital, and colleagues wrote. “The ability of clinician suspicion to accurately identify children at either high or low risk for Lyme disease has not been rigorously evaluated.”

To examine the accuracy of clinician suspicion for Lyme disease, researchers conducted a prospective cohort study of children aged 1 to 21 years who were examined for Lyme disease in EDs. They asked treating clinicians to estimate the probability of Lyme disease on a 10-point scale, with 1 being “not likely to be Lyme disease” and 10 being “very likely.”

 They constructed receiver operating characteristic curves then used the area under the curve (AUC) to quantify the discriminative ability of clinicians, with AUC less than 0.7 considered poor discriminatory value, AUC between 0.7 and 0.8 as minimally accurate, AUCs of 0.8 to 0.9 as good accuracy and AUCS of more than 0.9 as excellent accuracy.

Of 1,021 children enrolled in the study, 238 (23%) had Lyme disease.  Clinician suspicion had a minimal ability in determining whether children had Lyme disease or did not (AUC, 0.75; 95% CI, 0.71-0.79).

Of 554 children who the treating physicians thought were unlikely to have the disease (score 1-3), 12% had Lyme disease. Out of 127 patients who the treating physicians thought were very likely to have Lyme (score 8-10), 31% did not have the disease. These findings suggest that although clinician suspicion of Lyme disease can assist with initial management, final decisions and diagnoses should not be made until laboratory confirmation arrives.

“New approaches for the diagnosis of Lyme disease that include more rapid and more accurate diagnostic tests are needed,” Nigrovic and colleagues wrote. “Although 2-tiered Lyme disease serology takes several days to return results in more clinical settings, final patient management decisions should await confirmatory test results to avoid both over- and underdiagnosis of Lyme disease.” – by Savannah Demko

 Disclosure: The authors report no relevant financial disclosures.

 

Clinician suspicion had minimal accuracy in differentiating between children with and without Lyme disease who presented to the ED, according to findings published in Pediatrics.

“Clinicians use clinical prediction rules to combine available demographic, clinical and laboratory factors to estimate the probability of an outcome and assist clinical decision-making. However, the considerable overlap between Lyme disease and its mimics limits the clinical applicability of these predictive models,” Lise E. Nigrovic, MD, MPH, from the division of emergency medicine at Boston Children’s Hospital, and colleagues wrote. “The ability of clinician suspicion to accurately identify children at either high or low risk for Lyme disease has not been rigorously evaluated.”

To examine the accuracy of clinician suspicion for Lyme disease, researchers conducted a prospective cohort study of children aged 1 to 21 years who were examined for Lyme disease in EDs. They asked treating clinicians to estimate the probability of Lyme disease on a 10-point scale, with 1 being “not likely to be Lyme disease” and 10 being “very likely.”

 They constructed receiver operating characteristic curves then used the area under the curve (AUC) to quantify the discriminative ability of clinicians, with AUC less than 0.7 considered poor discriminatory value, AUC between 0.7 and 0.8 as minimally accurate, AUCs of 0.8 to 0.9 as good accuracy and AUCS of more than 0.9 as excellent accuracy.

Of 1,021 children enrolled in the study, 238 (23%) had Lyme disease.  Clinician suspicion had a minimal ability in determining whether children had Lyme disease or did not (AUC, 0.75; 95% CI, 0.71-0.79).

Of 554 children who the treating physicians thought were unlikely to have the disease (score 1-3), 12% had Lyme disease. Out of 127 patients who the treating physicians thought were very likely to have Lyme (score 8-10), 31% did not have the disease. These findings suggest that although clinician suspicion of Lyme disease can assist with initial management, final decisions and diagnoses should not be made until laboratory confirmation arrives.

“New approaches for the diagnosis of Lyme disease that include more rapid and more accurate diagnostic tests are needed,” Nigrovic and colleagues wrote. “Although 2-tiered Lyme disease serology takes several days to return results in more clinical settings, final patient management decisions should await confirmatory test results to avoid both over- and underdiagnosis of Lyme disease.” – by Savannah Demko

 Disclosure: The authors report no relevant financial disclosures.

 

    Perspective
    Eugene Shapiro

    Eugene Shapiro

    The conclusion of the authors of this study could be a bit misleading. I do not agree that the clinical suspicion of the clinicians was only minimally accurate in patients without eythema migrans (EM). Most of the other manifestations of Lyme disease, such as arthritis or facial nerve palsy, are not specific for Lyme disease, so laboratory confirmation is necessary.

    The authors only included patients for whom antibody tests had been ordered. It is likely that there were many more patients (not included in the study) for whom Lyme disease was considered but was ruled out based on clinical judgment. The area under the ROC curve for the estimates of the clinicians was 75%, which the authors classified as minimally accurate. The standard of accuracy used by the authors is arbitrary and usually is applied to classification of the accuracy of laboratory (or radiographic) tests to confirm a clinical diagnosis, not to classification of the accuracy of the clinical diagnosis alone. If the estimates were no more accurate than would be expected by chance, the area under the curve would be 50%. Thus, the estimates of the clinicians were 50% better than would be expected by chance alone — not that bad.

    Moreover, Lyme disease was confirmed in 11.7% and 69.3% of patients in whom the clinicians thought that Lyme disease was either “unlikely” or “very likely,” respectively (OR = 17; P < 0.0001) — pretty good! Is this good enough to forgo laboratory tests to make a diagnosis of Lyme disease? As the authors conclude, of course not! The accuracy of these estimates are probably comparable to those of most clinicians, and as good as or better than estimates (without laboratory test results) for a number of other conditions.

    My conclusion is that in patients with manifestations of Lyme disease other than EM, laboratory confirmation is necessary.

    • Eugene Shapiro, MD
    • Professor of pediatrics and epidemiology Yale School of Medicine

    Disclosures: Shapiro reports no relevant financial disclosures.