Climate Change Indicators in the United States Technical Documentation Lyme Disease

Lyme Disease

Identification

1. Indicator Description

This indicator looks at the incidence of Lyme disease in the United States since 1991. Lyme disease is a
tick-borne bacterial illness that can cause fever, fatigue, and joint and nervous system complications.
The spread of Lyme disease is affected by tick prevalence; populations and infection rates among host
species; human population patterns, awareness, and behavior; habitat; climate; and other factors. Lyme
disease may be useful for understanding the long-term effects of climate change on vector-borne
diseases because shorter-term variations in weather have less of an impact on ticks than on other
disease vectors such as mosquitoes. This is the case for several reasons (Ogden et a I., 2013):

• Ticks have a relatively long life cycle, including stages of development that take place in the soil,
where temperatures fluctuate less than air temperatures.

• Tick development rates have a delayed response to temperature changes, which minimizes the
effects of short-term temperature fluctuations.

• Ticks can take refuge in the soil during periods of extreme heat, cold, drought, or rainfall.

• Ticks are associated with woodland habitats, where microclimates are buffered from
temperature extremes that occur in treeless areas.

• Unlike other disease vectors such as mosquitoes, ticks do not have nonparasitic immature
feeding stages whose survival is susceptible to short-term changes in weather.

Consequently, in some locations in the United States, Lyme disease incidence would be expected to
increase with climate change.

Components of this indicator include:

• Annual incidence of Lyme disease in the United States (Figure 1).

• Change in reported Lyme disease incidence in the Northeast and Upper Midwest (Figure 2).

• Change in incidence and distribution of reported cases of Lyme disease in the Northeast and
Upper Midwest (1996 and 2014 maps).

2. Revision History

May 2014: Indicator published.

June 2015: Updated indicator on EPA's website with data through 2013.

August 2016: Updated indicator with data through 2014.

Technical Documentation: Lyme Disease

-------
Data Sources

3. Data Sources

This indicator is based on annual numbers of confirmed Lyme disease cases, nationally and by state,
compiled by the Centers for Disease Control and Prevention's (CDC's) Division of Vector-Borne Diseases
(DVBD). Incidence was calculated using the most recent mid-year population estimates for each year
from the U.S. Census Bureau. The 1996 and 2014 comparison maps also came from CDC.

4. Data Availability

All of the data for this indicator are publicly available on CDC and Census Bureau websites.

EPA obtained the data for this indicator from CDC's website. Prior to 2008, CDC compiled only confirmed
cases, but in 2008 it also began to track probable (but unconfirmed) cases. CDC's database allows users
to query just the confirmed cases, which EPA used for this indicator.

Although data are available for 1990, this indicator starts in 1991 because Lyme disease did not become
an official nationally reportable disease until January 1991. In 1990, some states reported Lyme disease
incidence using the standardized case definition that went into effect nationwide in 1991, but other
states did not.

CDC's national and state-level data are available online. Through the years, these data have been
published in CDC's Morbidity and Mortality Weekly Reports (MMWR), which are available at:
www.cdc.gov/mmwr/mmwr nd/index.html. Data from 2003 onward are also available in tabular form
at: www.cdc.gov/lyme/stats/mmwr.html. Underlying county-level data are not available publicly—or
they are combined into multi-year averages before being made publicly available—because of concerns
about patient confidentiality. Annual maps of reported cases of Lyme disease, as shown in the
1996/2014 comparison for this indicator, are posted online at: www.cdc.gov/lvme/stats/index.html.

Following CDC's standard practice, incidence has been calculated using population estimates on July 1 of
each calendar year. These population estimates are publicly available from the U.S. Census Bureau's
Population Estimates Program. Pre-2010 data are available at:

www.census.gov/popest/data/intercensal/index.html. Data for 2010 and later are available at:
www.census.gov/popest/data/index.html.

Methodology

5. Data Collection

This indicator is based on the annual reported number of Lyme disease cases as compiled by CDC.

State and local health departments report weekly case counts for Lyme disease following CDC's case
definitions through the National Notifiable Diseases Surveillance System (NNDSS). The NNDSS is a public-
health system for the reporting of individual cases of disease and conditions to state, local, and
territorial health departments, which then forward case information to CDC. The provisional state-level

Technical Documentation: Lyme Disease

-------
data are reported in CDC's MMWR. After all states have verified their data, CDC publishes an annual
surveillance summary for Lyme disease and other notifiable diseases.

Health care providers nationwide follow a standardized definition for what constitutes a "confirmed"
case of Lyme disease, but this definition has changed over time (see Section 8). The first standardized
surveillance case definition was established in 1990 by the Council of State and Territorial
Epidemiologists (CSTE). In January 1991, Lyme disease became a nationally notifiable disease in the
United States, using the CSTE's 1990 definition. As such, state and local health departments work with
health care providers to obtain case reports for Lyme disease based upon the CSTE case definition.

6. Indicator Derivation

Figure 1. Reported Cases of Lyme Disease in the United States, 1991-2014

National incidence of Lyme disease was calculated using the total number of confirmed Lyme disease
cases and the national population for each year from 1991 through 2014. EPA calculated incidence by
dividing the number of confirmed cases per year by the corresponding population on July 1 in the same
calendar year. EPA then multiplied the per-person rate by 100,000 to generate a normalized incidence
rate per 100,000 people. This is CDC's standard method of expressing the incidence of Lyme disease.

Figure 2. Change in Reported Lyme Disease Incidence in the Northeast and Upper Midwest, 1991-2014

EPA used ordinary least-squares linear regression to determine the slope of the trend over time for each
state. Of the 50 states plus the District of Columbia, 36 have a long-term linear trend in Lyme disease
incidence that is statistically significant to a 95-percent level, and 31 have trends that are significant to a
99-percent level. Many of these trends, however, have a very small slope. Taking the regression slope
(the annual rate of change) and multiplying it by 23 years (the length of the period of record) to
estimate total change, more than half of the states had a total change of less than 1 case per 100,000 in
either direction.

In this analysis, 14 states stand out because they have Lyme disease rates more than 10 times higher
than most of the other states, average rates of more than 10 cases per 100,000 per year during the most
recent five years of data, and in all but three of these states, statistically significant total increases of 10
to 100 cases per 100,000 between 1991 and 2014. These 14 states are:

• Connecticut

• Delaware

• Maine

• Maryland

• Massachusetts

• Minnesota

• New Hampshire

• New Jersey

• New York

• Pennsylvania

• Rhode Island

• Vermont

Technical Documentation: Lyme Disease

-------
•	Virginia

•	Wisconsin

Together, these 14 states account for about 95 percent of the nation's reported cases of Lyme disease in
2014, as the map in Figure TD-1 indicates.

Figure TD-1. Reported Cases of Lyme Disease in the United States, 2014

1 dot placed randomly within county of residence tor each confirmed case

Data source: CDC: www.cdc.gov/lyme/stats/maps.html. Accessed December 2015.

Figure 2 shows the total change (annual rate of change [regression slope] multiplied by 23 years) for the
14 states listed above. Trends are not shown for Connecticut, New York, and Rhode Island in Figure 2
because of too much year-to-year variation in reporting practices to allow trend calculation (see Section
12).

Reported Lyme Disease Cases in 1996 and 2014

This comparison uses two maps—one for the year 1996 and one for the year 2014—to illustrate changes
in the incidence and distribution of reported cases of Lyme disease in the United States over time. CDC
created these maps. Each dot on the maps represents an individual case placed randomly within the
patient's county of residence, which may differ from the county of exposure.

Indicator Development

In the course of developing and revising this indicator based on peer review and comments from CDC
experts, EPA considered several ways to present the data. For example:

Technical Documentation: Lyme Disease

4

-------
• The incidence of a disease can be tracked with total case counts or with incidence rates that are
normalized by population size. EPA chose to display rates for this indicator so as to eliminate
state-to-state population differences and changes in population overtime as confounding
factors. This approach is also consistent with data for EPA's Heat-Related Deaths indicator,
which is displayed using incidence rates.

• EPA considered focusing the analysis of reported Lyme disease on a subset of states. One
approach was to consider "reference states" as defined by CDC

(www.cdc.gov/mmwr/pdf/ss/ss5710.pdf). Upon clarification from CDC, however, this set of
reference states has not been used operationally since CDC's Healthy People 2010 effort, which
concluded in 2010, and they do not necessarily represent a consistent baseline from which to
track trends. EPA chose to use more objective, data-driven thresholds for selecting states to
show readers all change in reported Lyme disease incidence as in Figure 2. However, there is
scientific evidence (e.g., Diuk-Wasser et al., 2012; Stromdahl and Hickling, 2012) that notes the
geographic differences in Ixodes scapularis (the deer tick or blacklegged tick) in North America—
and that increases in Lyme disease cases in many states south of 35°N latitude are likely due to
non-climate-related expansion of northern I. scapularis tick genotypes. Analyzing data for a set
of states in the northern part of the range of I. scapularis might lead to better understanding of
changes in Lyme disease cases as related to a warming climate. Thus, future work on this
indicator will attempt to reflect the effects of climate change on expansion in the range of /.
scapularis, increasing abundance of I. scapularis where it already occurs, increases in the
prevalence of Borrelia burgdorferi (the bacteria that actually cause Lyme disease) in host-
seeking ticks, and/or updated understanding of other known environmental drivers, such as
deer density and changes in landscape, habitat, and biodiversity.

• EPA considered mapping rates or trends by county; however, county-level case totals are only
publicly available from CDC in five-year bins, in part because of the very low number of cases
reported in many counties.

7. Quality Assurance and Quality Control

Each state has established laws mandating that health providers report cases of various diseases
(including Lyme disease) to their health departments. Each state health department verifies its data
before sharing them with CDC. The NNDSS is the primary system by which health surveillance data are
conveyed to CDC for national-level analyses.

Starting in 1990, CDC launched the National Electronic Telecommunications System for Surveillance
(NETSS), replacing mail and phone-based reporting. In 2000, CDC developed the National Electronic
Disease Surveillance System (NEDSS) Base System (NBS). This central reporting system sets data and
information technology standards for departments that provide data to CDC, ensuring that data are
submitted quickly, securely, and in a consistent format.

Using CSTE case definitions, CDC provides state and local health departments and health providers with
comprehensive guidance on laboratory diagnosis and case classification criteria, ensuring that all health
providers and departments classify Lyme disease cases consistently throughout the United States.

State health officials use various methods to ascertain cases, including passive surveillance initiated by
health care providers, laboratory-based surveillance, and "enhanced or active surveillance" (Bacon et al.,

Technical Documentation: Lyme Disease

-------
2008). State officials check the data and remove duplicate reports before submitting annual totals to
CDC.

CDC has undertaken a review of alternative data sources to see how closely they align with the disease
counts captured by the NNDSS. These alternative sources include medical claims information from a
large insurance database, a survey of clinical laboratories, and a survey that asks individuals whether
they have been diagnosed with Lyme disease in the previous year. Preliminary results from this review
suggest that the NNDSS may be undercounting the true number of cases of Lyme disease (CDC, 2013).
See Section 10 for further discussion about this possible source of uncertainty.

Analysis

8. Comparability Over Time and Space

Lyme disease data collection follows CDC's case definition to ensure consistency and comparability
across the country. The national case definition for Lyme disease has changed twice since Lyme disease
became a notifiable disease, however: first in 1996 and again in 2008. Prior to 1996, a confirmed case of
Lyme disease required only a skin lesion with the characteristic "bulls-eye" appearance. In 1996, CDC
expanded the definition of confirmed cases to include laboratory-confirmed, late-manifestation
symptoms such as issues with the musculoskeletal, nervous, and cardiovascular systems. In 2008, the
case classifications were expanded again to include suspected and probable cases.

These definition changes necessitate careful comparisons of data from multiple years. While it is not
possible to control for the case definition change in 1996, CDC provides the numbers of confirmed cases
and suspected and probable cases separately. The granularity of the data enables EPA to use confirmed
cases in the incidence rate calculation for all years and exclude the probable cases that have been
counted since 2008, ensuring comparability over time.

In addition to the national changes, several state reporting agencies have changed their own definitions
at various times. These state-level changes include California in 2005, Connecticut in 2003, the District of
Columbia in 2011, Hawaii in 2006, New York in 2007, and Rhode Island in 2004. The extent to which
these changes affect overall trends is unknown, but it is worth noting that Connecticut and Rhode Island
both have apparent discontinuities in their annual totals around the time of their respective definitional
changes, and these two states and New York all have statistically insignificant long-term trends (see
Section 12), despite being surrounded by states with statistically significant increases. Because of these
state-level uncertainties, Figure 2 shows only state-level trends that are statistically significant. In this
case, the p-value for each displayed state is less than 0.01.

9. Data Limitations

Factors that may have an impact on the confidence, application, or conclusions drawn from this
indicator are as follows:

1. For consistency, this indicator includes data for only confirmed cases of Lyme disease. However,
changes in diagnosing practices and awareness of the disease overtime can affect trends.

Technical Documentation: Lyme Disease

-------
2. CDC's national Lyme disease case definitions have changed twice since Lyme disease became a
notifiable disease. As discussed in Section 8, it is not possible to control for the case definition
change in 1996, which adds some uncertainty to the indicator. State agencies have also changed
their definitions at various times, as described in Section 8.

3. As described in Section 10, public health experts believe that many cases of Lyme disease are
not reported, which means this indicator underestimates the true incidence of the disease (CDC,
2013). The reporting rate may vary over time and space as a result of differences in funding and
emphasis among state surveillance programs. In addition, Lyme disease can be difficult to
diagnose. Cases in locations where Lyme disease is not endemic are at particular risk of being
unidentified or misdiagnosed.

4. As an indicator of climate change, Lyme disease is limited due to several confounding factors:

• Pest extermination efforts and public health education may counteract the growth of
confirmed cases expected due to warming climates.

• Importantly, there are several factors driving changes in incidence of Lyme disease other
than climate. Several of these factors have not been well-quantified or studied. Possible
factors include range expansion of vector ticks, which is not always climate-related;
proximity of hosts; changes in deer density; changes in biodiversity; and the effects of
landscape changes such as suburbanization, deforestation, and reforestation.

• Pathogen transmission is affected by several factors including geographic distribution,
population density, prevalence of infection by zoonotic pathogens, and the pathogen load
within individual hosts and vectors (e.g., Cortinas and Kitron, 2006; Lingren, 2005; Mills et
al., 2010; Raizman, 2013).

• Human exposure depends upon socioeconomic and cultural factors, land use, health care
access, and living conditions (Gage et al., 2008; Gubler et al., 2001; Hess et al., 2012;

Lafferty, 2009; Wilson, 2009).

5. Lyme disease surveillance data capture the county of residence, which is not necessarily the
location where an individual was infected.

10. Sources of Uncertainty

The main source of uncertainty for this indicator stems from its dependence on surveillance data.
Surveillance data can be subject to underreporting and misclassification. Because Lyme disease is often
determined based upon clinical symptoms, lack of symptoms or delayed symptoms may result in
overlooked or misclassified cases. Furthermore, surveillance capabilities can vary from state to state, or
even from year to year based upon budgeting and personnel.

Although Lyme disease cases are supposed to be reported to the NNDSS, reporting is actually voluntary.
As a result, surveillance data for Lyme disease do not provide a comprehensive determination of the
U.S. population with Lyme disease. For example, it has been reported that the annual total number of
people diagnosed with Lyme disease may be as much as 10 times higher than the surveillance data

Technical Documentation: Lyme Disease

-------
indicate (CDC, 2013). Consequently, this indicator provides an illustration of trends over time, not a
measure of the exact number of Lyme disease cases in the United States.

Another issue is that surveillance data are captured by county of residence rather than county of
exposure. Reports of Lyme disease may therefore occur in states with no active pathogen populations.
For example, a tourist may be infected with Lyme disease while visiting Connecticut (an area with high
incidence of Lyme disease) but not be identified as a Lyme disease case until the tourist returns home to
Florida (an area where blacklegged ticks cannot survive). This may result in underreporting in areas of
high Lyme disease incidence and overreporting in areas of low Lyme disease incidence.

For a discussion of the uncertainties associated with the U.S. Census Bureau's intercensal estimates, see:
www.census.gov/popest/methodology/intercensal nat meth.pdf.

11. Sources of Variability

The incidence of Lyme disease is likely to display variability over time and space due to:

• Changes in populations of blacklegged ticks and host species (e.g., deer, mice, birds) over time.

• Spatial distribution of blacklegged ticks and changes in their distribution over time.

• The influence of climate on the activity and seasonality of the blacklegged tick.

• Variability in human population over time and space.

This indicator accounts for these factors by presenting a broad multi-decadal national trend in Figures 1
and 2. EPA has reviewed the statistical significance of these trends (see Section 12).

12. Statistical/Trend Analysis

Based on ordinary least-squares linear regression, the national incidence rate in Figure 1 increases at an
average annual rate of +0.22 cases per 100,000 people (p < 0.001).

Of the 14 states shaded in Figure 2, 11 had statistically significant increases in their annual incidence
rates from 1991 to 2014 (all p-values substantially less than 0.01), based on ordinary least-squares linear
regression. The shading in Figure 2 shows the magnitude of these trends. The other three states did not:
Connecticut (p = 0.74), New York (p = 0.11), and Rhode Island (p = 0.23). A broader analysis described in
Section 6 found that more than half of the 50 states had significant trends in their annual incidence rates
from 1991 to 2014, but most of these states were excluded from Figure 2 because their overall
incidence rates have consistently been at least an order of magnitude lower than the rates in the 14 key
Northeast and Upper Midwest states where Lyme disease is most prevalent.

References

Bacon, R.M., K.J. Kugeler, and P.S. Mead. 2008. Surveillance for Lyme disease—United States, 1992-
2006. Morbidity and Mortality Weekly Report 57(SS10):l-9.

CDC (U.S. Centers for Disease Control and Prevention). 2013. CDC provides estimate of Americans
diagnosed with Lyme disease each year, www.cdc.gov/media/releases/2013/p0819-lvme-disease.html.

Technical Documentation: Lyme Disease

-------
Cortinas, M.R., and U. Kitron. 2006. County-level surveillance of white-tailed deer infestation by Ixodes
scapularis and Dermacentor albipictus (Acari: Ixodidae) along the Illinois River. J. Med. Entomol.
43(5):810-819.

Diuk-Wasser, M.A., A.G. Hoen, P. Cislo, R. Brinkerhoff, S.A. Hamer, M. Rowland, R. Cortinas, G. Vourc'h,
F. Melton, G.J. Hickling, J.I. Tsao, J. Bunikis, A.G. Barbour, U. Kitron, J. Piesman, and D. Fish. 2012. Human
risk of infection with Borrelia burgdorferi, the Lyme disease agent, in eastern United States. Am. J. Trop.
Med. Hyg. 86(2):320-327.

Gage, K.L., T.R. Burkot, R.J. Eisen, and E.B. Hayes. 2008. Climate and vector-borne diseases. A. J. Prev.
Med. 35(5):436-450.

Gubler, D.J., P. Reiter, K.L. Ebi, W. Rap, R. Nasci, and J.A. Patz. 2001. Climate variability and change in the
United States: Potential impacts on vector- and rodent-borne diseases. Environ. Health. Perspect.
109:223-233.

Hess, J.J., J.Z. McDowell, and G. Luber. 2012. Integrating climate change adaptation into public health
practice: Using adaptive management to increase adaptive capacity and build resilience. Environ. Health.
Perspect. 120(2):171-179.

Lafferty, K.D. 2009. The ecology of climate change and infectious diseases. Ecology 90(4):888-900.

Lingren, M., W.A. Rowley, C. Thompson, and M. Gilchrist. 2005. Geographic distribution of ticks (Acari:
Ixodidae) in Iowa with emphasis on Ixodes scapularis and their infection with Borrelia burgdorferi.
Vector-Borne Zoonot. 5(3):219-226.

Mills, J.N., K.L. Gage, and A.S. Khan. 2010. Potential influence of climate change on vector-borne and
zoonotic diseases: A review and proposed research plan. Environ. Health. Perspect. 118(11):1507-1514.

Ogden, N.H., S. Mechai, and G. Margos. 2013. Changing geographic ranges of ticks and tick-borne
pathogens: Drivers, mechanisms, and consequences for pathogen diversity. Front. Cell. Infect. Microbiol.
3:46.

Raizman, E.A., J.D. Holland, and J.T. Shukle. 2013. White-tailed deer (Odocoileus virginianus) as a
potential sentinel for human Lyme disease in Indiana. Zoonoses Public Hlth. 60(3):227-233.

Stromdahl, E.Y., and G.J. Hickling. 2012. Beyond Lyme: Aetiology of tick-borne human diseases with
emphasis on the south-eastern United States. Zoonoses Public Hlth. 59 Suppl 2:48-64.

Wilson, K. 2009. Climate change and the spread of infectious ideas. Ecology 90:901-902.

Technical Documentation: Lyme Disease

9

-------