Medical Claims Data Reveals Lyme Disease Underreported

Lyme disease (LD) is a zoonotic infection, a disease that normally exists in animals but that can infect humans, transmitted by Ixodes spp. ticks and caused by the spirochete Borrelia burgdorferi. Signs and symptoms of infection range in severity and can include erythema migrans (LD skin lesion), arthritis, facial palsy, disease of the spinal nerves, arrhythmia, and meningitis. Most patients recover fully after antimicrobial treatment. However, serious illness and even deaths have been reported, although rarely. In the United States, LD is the fifth most commonly reported nationally notifiable disease; about 36,000 confirmed and probable cases were reported in 2013. U.S. cases are concentrated heavily in the Northeast and upper Midwest.


Surveillance for LD in the United States is based on reports submitted by laboratories and health care providers to state and local health departments. These reports provide valuable insight into the age and sex distribution of patients with LD and the seasonality and geographic distribution of cases, and they enable monitoring of disease trends over time. Unfortunately, underreporting and variation in surveillance practices limit the ability of routine surveillance to capture the true overall frequency of LD within the population. Studies conducted during the 1990s in high-incidence states suggest that LD cases are underreported by a factor of 3 to 12. These studies were limited to specific states and do not necessarily reflect underreporting nationwide.


Medical claims data provide an additional source of information about the epidemiology and public health importance of LD. Because these data are based on billing records submitted by clinicians for reimbursement, they are less prone to underreporting than are routine surveillance data that require additional documentation. The researchers used information from a large, nationwide medical claims database to (1) describe the epidemiology of LD diagnosed by clinicians, (2) identify similarities and differences with surveillance data, and (3) estimate the number of LD cases per year in the United States.


The Minnesota Department of Health found the LD billing code was present in 145 (56.4%) of 257 charts for which a clinician documented a new case of LD. A Maryland Department of Health and Mental Hygiene study found the LD billing code in 45 (44.6%) of 101 charts from patients in whom LD was diagnosed and reported by clinicians or clinical centers. Furthermore, the New York State Department of Health found the LD billing code in 114 (41.8%) of 273 charts from patients in whom LD was diagnosed. Finally, the Tennessee Department of Health found the LD billing code listed at least once in 9 (37.5%) of 24 charts from patients with Blue Cross Blue Shield insurance in whom LD was diagnosed and who were reported to the Department of Health. Thus, of 655 collective charts from LD patients, 313 charts had LD billing codes. Therefore, to account for patients in whom LD was diagnosed but whose charts were not coded with the LD billing code, the researchers multiplied the estimated number of cases with the LD billing code by a correction factor calculated as follows: 313/655 = 1/x, where x = 2.09.


The 15 states and district with the highest average incidence represented 80.6% of clinician-diagnosed LD and were as follows, in descending order: Connecticut, Rhode Island, Maryland, New Jersey, Massachusetts, New York, New Hampshire, Pennsylvania, Maine, Delaware, Virginia, Vermont, Wisconsin, District of Columbia, and Minnesota. These same 15 states and district were seen in surveillance data, although the rank order differed slightly, and they constituted a significantly greater proportion (96.3%) of reported cases.


Direct standardization of clinician-diagnosed LD and addition of estimated cases in persons under 65 years of age produced an estimate of 157,137 cases per year, which was multiplied by 2.09 to correct for omission of the LD billing code in patient charts. This calculation yielded a national estimate of 329,000 LD cases per year during 2005–2010.


On the basis of this number, the estimated incidence of clinician-diagnosed LD in the United States during this period was 106.6 cases per 100,000 persons per year. In comparison, average US incidence according to surveillance data during this period was 9.4 cases per 100,000 persons per year.


Using medical claims data, it was estimated that 329,000 (95% credible interval 296,000–376,000) LD cases occur annually in the United States, which emphasizes the substantial public health effect of this disease. This estimate is consistent with findings from a recent study of diagnostic laboratories that yielded an estimate of 288,000 (range 240,000–444,000) infections among patients for whom a laboratory specimen was submitted in 2008. As expected, the estimate is slightly higher because it also includes LD cases diagnosed without laboratory testing (i.e., clinical diagnosis based on presence of erythema migrans, the classic bull’s-eye rash that appears on most people infected with LD, after exposure in a Lyme-endemic area).


Presence of a diagnostic code in the chart or a clinician diagnosis of an infectious condition does not necessarily signify a true infection. Possible reasons include rule-out diagnoses, codes for medical history but not incident infections, and overdiagnosis (incorrect diagnosis of LD when the patient has a different condition). Rule-out diagnoses and medical history codes most likely were reduced—but not completely eliminated—by including only outpatients treated with an antimicrobial drug recommended for LD. Overdiagnosis of LD is not uncommon given that, in some circumstances, the differential diagnosis for symptoms of LD can be broad.


Epidemiologic patterns of clinician-diagnosed LD were similar to patterns among cases reported through national surveillance. For example, incidence was highest among boys 5 to 9 years of age and persons 60 to 64 years of age of both sexes, which is believed to be attributable partially to behavioral factors and increased exposure to ticks in these age groups. However, some discrepancies were also noted. Specifically, incidence of clinician-diagnosed LD was higher than expected among women 15 to 44 years of age. A study of records with the LD billing code using Maine’s statewide electronic database of inpatient and outpatient encounters also found a higher percentage of female patients compared with surveillance data. This finding might be attributable to differential overdiagnosis of LD in these groups, variations in insurance coverage and health care–seeking behavior, or other factors. Studies in Europe have found sex discrepancies in risk for tick bites and clinical presentation of LD that should be explored further in US research studies.


The estimated number of clinician-diagnosed LD cases in the United States is higher than the number reported through routine surveillance and consistent with previous estimates of LD underreporting. Underreporting occurs with other notifiable conditions and should not be confused with lack of treatment. Indeed, our study confirms that many LD cases not formally reported are nevertheless diagnosed and treated by clinicians. Furthermore, underreporting aside, the general concordance in LD epidemiology seen in MarketScan and surveillance data underscores that LD surveillance serves its central purpose: to identify and track patterns of disease.


Primary advantages of this study are the large sample size, ability to circumvent the obstacles and biases inherent in routine reporting mechanisms, detailed information about clinical and prescription data, and the ability to follow patient data over time. The findings from these analyses were generally consistent with each other and with results expected on the basis of public health experience.


The study highlights the need for continued coding research, particularly as health departments explore the feasibility of using electronic medical records to facilitate LD reporting. Additional information about LD coding practices will enable robust comparisons of ICD codes related to actual cases and facilitate future research using medical databases. In addition, ongoing research using the MarketScan databases and other sources will elucidate detailed epidemiologic and clinical aspects of LD that are not apparent in standard surveillance data.


In conclusion, the findings underscore that LD is a considerable public health problem, both in terms of number of cases and overall health care use. Furthermore, as with other conditions, underreporting in the national surveillance system remains a challenge. Continued research and education are necessary to enhance prevention efforts and improve diagnostic accuracy to reduce the effects of this disease.


See the CDC Report


See also Medical Law Perspectives, April 2014 Report: Danger and Controversy: Lyme Disease Liability Risks


See the Medical Law Perspectives July 27, 2015, Blog: Problems of Proving a Lyme Disease Diagnosis