Long-Term Sequelae and Health-Related Quality of Life Associated With Lyme Disease: A Systematic Review

Abstract

Lyme disease (LD) is the most commonly reported vector-borne disease, but its clinical consequences remain uncertain. We conducted a systematic review of the long-term sequelae and health-related quality of life (HRQoL) associated with LD in North America and Europe. We performed searches in 6 electronic databases up to December 2018 following Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines, including observational studies reporting long-term sequelae, HRQoL, and prognostic factors. We included 46 studies, published between 1994 and 2019. Based on 21 studies reporting attributable outcomes, higher proportions of sequelae reported from exposed patients were: neck pain, myalgia, arthralgia, paresthesia, sleep disorder, poor appetite, and concentration difficulties. Patients with PTLDS reported impaired HRQoL compared to the general US population. Included studies were highly heterogeneous in terms of study design, settings, patient characteristics, and quality. Patients with LD are more likely to report nonspecific long-term sequelae, especially those experiencing persistent symptoms posttreatment. Opportunities exist for prospective longitudinal studies to better understand LD outcomes.

This systematic review summarizes the long-term consequences and quality-of-life associated with Lyme disease (LD) after treatment compared to normal populations. Nonspecific patient-reported symptoms are more common in patients with LD and posttreatment Lyme disease syndrome than in controls.

Lyme disease (LD), also known as Lyme borreliosis (LB), is a vector-borne disease caused by several genospecies of the Borrelia burgdorferi senso lato spirochete, which is transmitted through a tick bite. In North America (NA), B. burgdorferi is most commonly responsible for LD, whereas in Europe, the common spirochetes include B. burgdorferi, B. garinii, and B. afzelii [1–3]. LD is the most commonly reported vector-borne disease in North America and Europe, with peak rates occurring during the early to late summer months (ie, May to September) [3]. Notable areas at high-risk for LD include: Maine, United States of America (86.4 cases per 100 000 population) [4], southern Sweden (464 per 100 000), and the Netherlands (134 per 100 000) [5]. In Canada, the number of reported LD cases increased from 144 in 2009 (0.4 per 100 000) to 2025 (5.5 per 100 000) in 2017 [6], almost equivalent to a 14-fold increase over the past 8 years.

Patients presenting with clinical signs and symptoms of LD are typically categorized into stages at diagnosis: early localized, early disseminated, late disseminated, and posttreatment Lyme disease syndrome (PTLDS) [7]. A common sign of early localized LD is erythema migrans, a bulls-eye rash [1, 8, 9]. Diagnosis of LD is based on a combination of clinical signs and symptoms, exposure to blacklegged ticks, and laboratory confirmation. If early localized LD remains undiagnosed or untreated, it can progress to early disseminated LD with neurologic (eg, facial palsy and meningitis) and cardiac complications (eg, Lyme carditis from atrioventricular block) [10]. Further progression of the disease can lead to Lyme arthritis and neurologic manifestations such as peripheral neuropathy or encephalopathy [8]. Approximately 10–20% of patients with LD continue to experience persistent fatigue, musculoskeletal and cognitive symptoms after standard antibiotic treatment, for6 months or longer, termed PTLDS [11]. Although these sequelae are noted in the literature, the natural history of treated LD progression to PTLDS and untreated LD are still not completely understood.

Understanding the potential long-term sequelae and health-related quality-of-life (HRQoL) can inform clinical management, prevention, and treatment of LD and inform identification and management of PTLDS. To address this, we conducted a systematic scientific literature review of primary observational (eg, case-control, cross-sectional, cohort) studies in North America and Europe to investigate and characterize the following outcomes in patients with confirmed LD: incidence of sequelae, HRQoL, health-state utility values (HSUV), prognostic factors, mortality, and length of hospitalization. We included HSUVs as they can express HRQoL in an overall single value, typically from 0 (state equivalent to death) to 1 (perfect health) [12].

METHODS

Data Sources and Searches

We conducted a systematic review following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines (Supplementary Table 1) [13]. We conducted a scientific literature search for English language studies published in 6 electronic databases: Medline E-pub Ahead of Print, In-Process & Other Non-Indexed Citations (Ovid interface; January 1, 1946, to January 10, 2019), Embase (Ovid interface; January 1, 1974, to 2019 Week 1); PsycInfo (Ovid interface; 1806 to December Week 5 2018), CINAHL Plus (Ebsco interface; January 1, 1902, to January 10, 2019); Environment Complete (Ebsco interface; January 1, 1902, to January 10, 2019), and Scopus (January 1, 1999, to January 10, 2019). Our search used medical subject headings (MeSH) and text words related to the following concepts: “Lyme disease,” “Lyme borreliosis,” “sequelae,” “quality of life,” “health status,” “burden of illness,” “prognosis,” and “mortality.” All searches are current as of January 10, 2019. A complete list of search terms and search strategy are in Supplementary 2). We manually searched reference lists from relevant articles and systematic reviews.

Study Selection

The protocol and eligibility criteria for studies to be included are published on PROSPERO (CRD42017068765). Reviewers (S. M., S. B., F. S.) independently screened all records in duplicate against the eligibility criteria, and resolved differences through a third reviewer (B. S.). We included studies that met the following criteria: enrolled exposed patients (diagnosed with definite/confirmed LD) based on the Centers for Disease Control and Prevention (or European equivalent) case definitions in North America or Europe; English language; primary study (case-control, cross-sectional, cohort studies) with an unexposed cohort (patients without LD infection); and addressed sequelae, HRQoL, and/or HSUVs, prognostic factors, length of hospitalization, or mortality associated with LD in humans. We defined Europe using the 2013 List of EU countries [14].

We excluded studies not conducted from the list of EU countries or North America (Canada and US) because of the differences in baseline health statuses, health systems, and genospecies of Borrelia spirochetes causing LD [3, 15]. We excluded case reports and case series due to feasibility and lack of control cohorts associated with these study types. Studies that used an unclear diagnosis criteria, enrolled patients with probable-LD or coinfected patients were excluded.

Data Extraction and Quality Assessment

We extracted study and patient characteristics in duplicate: study year, country of study, study design, setting, sample size for exposed and unexposed, mean or median age, sex distribution, LD diagnosis criteria, LD stage of patients, time from diagnosis to study (or time to follow-up), previous antibiotic treatment, measurement tools (tests), and outcomes examined. For sequelae, we extracted frequency, proportions, and standardized test scores for sequelae reported by exposed and unexposed patients. For HRQoL, HSUV outcomes, and prognostic factors, we extracted all quantitative scores such as SF-36 subdomain scores, utility values ranging from 0 to 1, and ratio measures (eg, rate ratios and odds ratios), respectively.

Quality appraisal was completed by 2 independent reviewers (S. M., S. B., F. S.) and conflicts were resolved through consensus using checklists developed by the Joanna Briggs Institute (JBI) [16, 17]. Studies reporting HSUVs were assessed using seven modified questions assessing HSUV elicitation studies adapted from Yepes-Nunez et al [18]. (Supplementary Table 3). Studies were deemed to be good or high quality if they met at least 50% or 80% of their assessment criteria, respectively [19].

Data Synthesis and Analysis

We classified studies based on patients’ LD stage at diagnosis. Supplementary Table 4 provides full details of LD stage (and PTLDS) classification steps. In our quantitative analysis looking at attributable sequelae, we only included studies that reported proportion of sequelae for exposed and unexposed cohorts. We summarized quantitative estimates for prognostic factors, HRQoL scores, and HSUVs. Due to heterogeneity in study designs and LD stage of patients, we did not perform any meta-analysis.

RESULTS

Study and Patient Characteristics

Our systematic review identified 6,086 de-duplicated records, of which 46 studies met eligibility criteria and were included (Figure 1). [20–65] Studies were published between 1994 and 2019, mostly from the US (n = 31), Sweden (n = 4), Germany (n = 3), Netherlands (n = 3) and Slovenia (n = 2). Thirty-three studies were cohort designs (retrospective n = 21; prospective n = 12), eight were cross-sectional, and five were other designs (e.g. economic evaluations) reporting HSUVs. Sample size for observational studies ranged from 6 to 2,067 for the exposed group, and 10 to 20,760 for the unexposed group. Mean or median age ranged from 6 to 66 in the exposed group, and 7 to 56 in the unexposed group. Sex composition (% female) was between 30 and 81.8%, and 34.6 and 67% for the exposed and unexposed cohorts, respectively.

Figure 1.

Preferred Reporting Items for Systematic Reviews and Meta-Analyses flow chart and study selection.

Thirty-six studies reported sequelae in frequencies, proportions, or using standardized test scores (Supplementary Tables 5 and 6): 13 reported HRQoL, 7 reported prognostic factors, and 5 reported HSUVs. Hospitalization was reported by a total of 20 patients with LD from 2 studies [24, 47], but the length of hospitalization was not reported. Mortality was reported by Obel et al for patients with LB in Denmark using a population-based cohort study but was defined as all-cause and not attributable to LB [50]. Only 31 studies were categorized by LD stage at diagnosis (2 early localized, 12 early disseminated, 6 late disseminated, and 11 PTLDS). Study and patient characteristics are summarized in Table 1.

Table 1.

Study and Patient Characteristics

Author Year	Country	Study Design	Exposed				Unexposed			Outcomes Examined	Funding Sources
			LD Stage at Diagnosis	n	Age, years, mean (SD)	Sex, % Female	n	Age, years, mean (SD)	Sex, % Female
Adams 1994	USA	Prospective cohort	EL-LD (n=8) ED-LD (n = 11) LD-LD (n = 22)	41	11.0 (2.9)	32	23	10.6 (2.2)	38	SQ	Nemours Foundation Clinical Research Fund Grant
Adams 1999	USA	Prospective cohort	PTLDS	25	12.1 (0.5)	36	17b	11.2 (1.3)b 12.4 (0.4)	50b 36	SQ	Grant from Nemours Research Programs
Aucott 2013	USA	Prospective cohort	EL-LD (n = 43) ED-LD (n = 20) PTLDS (n = 21, from original cohort)	42m	48.9 (15.5)	44	21	NRi	NR	SQ	Lyme Disease Research Foundation of Maryland
Back 2013	Germany	Cross-sectional	ED-LD	11	62 (12)	45	NRc	NRc	NRc	SQ; QOL	NR
Backman 2018	Sweden	Prospective cohort	ED-LD	68	Median (range): 6 (2–15)	44	133	Median (range): 13 (1–17)	62	SQ; QOL	Regional Research Council Uppsala-Örebro, the Center for Clinical Research Dalarna – Uppsala University, the Swedish Society of Medicine
Berende 2018	Netherlands	Economic evaluation	PTLDS	271	48.6 (12.8) – 50.2 (9.7)	44 – 48	NR	NR	NR	SQ	Netherlands Organization for Health Research and Development (ZonMw).
Desrch 2015	Germany	Retrospective cohort	ED-LD	30	58.46 (12.6)	47	35	55.91 (13.3)	51	UT	NR
Doshi 2018	USA	Cross-sectional	LD-LD (PTLDS)	81	46.5 (12.2)	59.3	44	44.1 (9.8)	50	SQ; QOL	NIMH R01 MH072383 (JR), NINDS R01 NS38636 (BAF), by the Steven and Alexandra Cohen Foundation, Lyme & Tick-borne Diseases Research Center at Columbia University Medical Center established by the Lyme Disease Association and Global Lyme Alliance
Eckman 1997	USA	Economic evaluation	NR	NR	NR	NR	NR	NR	NR	SQ	Grant (Cooperative Agreement no. CCU110291) from the Centers for Disease Control and Prevention.
Ettinger 1998	USA	Cross-sectional	PTLDSa	38	42.4 (13.2)	55	924; 2,514d	46 (14.7)	49	UT	NR
Fallon 1999	USA	Cross-sectional	PTLDS	13	43.1 (NR), range 20–67	62	16	45.3 (NR), range 22–67	50	SQ	Grants from NIAMS (1RO1AR-4313) and NIAID (NO1-AI-65308)
Greenberg 1995	USA	Retrospective cohort	LD-LD	11	37.3 (15)	82	10	36.4 (10.9)	60	QOL	NR
Horowitz 1996	USA	Prospective cohort	EL-LD (n = 80) ED-LD (n = 35)	115	44.0 (14.3)	35	110	31.8 (12.3)	55	SQ	Cooperative Agreement no.U50/CCU210280-01 from the Centers for Disease Control and Prevention (G.P.W.) and by grants RO1-AR41508 and RO1-AR43135 from the National Institute of Arthritis and Musculoskeletal and Skin Diseases
Jares 2014	USA	Retrospective cohort	EL-LD (n = 51) ED-LD (n = 9)	45e	54.9 (13.9), range 20–83	47	60	55.7 (NR)	47	SQ	Gundersen Medical Foundation
Kalish 2001	USA	Retrospective cohort	EL-LD (n = 25) ED-LD (n = 31) LD-LD (n = 28)	84	EL-LD, 53 (29–77) j ED-LD, 50 (19–61) j LD-LD, 45 (17–76) j	Total 51 EL-LD 55j ED-LD 48j LD-LD 50j	30	51 (29–74) j	67	SQ: QOL	Cooperative Agreement (CCU110291) from the Centers for Disease Control and Prevention, the Eshe fund, and a gift from Jeanette Rohatyn
Kaplan 1999	USA	Retrospective cohort	ED-LD (n = 14) PTLDS (n = 18)	32	LD-LD, 43.3 (11.8) PTLDS, 43.6 (8.9)	LD-LD 21 PTLDS 67	15	37.2 (9.2)	60	SQ: QOL	Grants AR-20358 from the NIH and the Eshe Fund
Keilp 2006	USA	Retrospective cohort	LD-LD (PTLDSa)	81	47.9 (11.0)	54	39	43.9 (10.6)	54	SQ	NINDS grant no. NS 38636 and by grants from the Lyme Disease Association, Time for Lyme, Inc., and the National Research Fund for Lyme and Tick-borne Diseases
Keilp 2018	USA	Cross-sectional	LD-LD (PTLDS)	81	47.8 (11.3)	54.3	97	38.3 (12.6)	50.5	SQ	Lyme & Tick-borne Diseases Research Center at Columbia University Medical Center (established by Global Lyme Alliance, Inc., and Lyme Disease Association, Inc.), the Steven & Alexandra Cohen Foundation, MH62155, MH62185, and NS38636.
Nichol 1998	Canada	Economic evaluation	NR	NR	NR	NR	NR	NR	NR	SQ	NR
Obel 2018	Denmark	Retrospective cohort	ED-LD	2,067	Median (IQR): 45 (11–62)	56	20,670	Median (IQR): 45 (11–62)	56	UT	Danish Council for Independent Research (grant number: 6110-00173B)
Pollina 1999 [1]	USA	Retrospective cohort	PTLDS	16	45.4 (8.49)	56	15	41.3 (10.6)	53	MO; PF	NR
Pollina 1999 [2]	USA	Retrospective cohort	PTLDS	25	43.3 (9.0)	56	23	40.7 (8.6)	56.5	SQ	NR
Rebman 2017	USA	Cross-sectional	PTLDS	61	49.3 (NR)	52.5	26	54.7 (NR)	53.9	SQ	Global Lyme Alliance and the Steven and Alexandra Cohen Foundation.
Sangha 1998	USA	Retrospective cohort	PTLDS	176	47.8 (NR)	50	160	49.7 (NR)	47	SQ; QOL	NIH grants AR36308 and AR02033
Schmidt 2015	Germany	Retrospective cohort	ED-LD	60	46.9 (13.9)	33	30	42.3 (16.0)	40	SQ; PF	Grants from the Niedersachsisches Ministeium fur Wissenschaft and Kultur (grant no. 11.4-76221-12-1/06, Ministry of Science and Culture of Lower Saxony, Hannover, Germany) and R. N. is supported by a grant from Else-Kröner-Fresenius Foundation
Seltzer 2000	USA	Retrospective cohort	EL-LD (68.1%)f ED-LD (23.9%) LD-LD (8.1%)	88	NRf	49f	88	NRf	66f	QOL	Grant from the Patrick and Catherine Weldon Donaghue Medical Research Foundation
Shadick 1994	USA	Retrospective cohort	LD-LD	38	49 (17.9)	42	43	49.1 (17.5)	49	QOL	NIH grants AR36308, AR20358, and AR07530
Shadick 1999	USA	Cross-sectional	PTLDS	186	47.5 (13.4)	50	167	19.6 (14.1)	47	SQ; PF	NIH grant AR36308.
Shadick 2001	USA	Economic evaluation	NR	105	NR	NR	NRg	NRg	NRg	SQ; QOL; PF	Grants AR-43653, AR 02033 and AR-36308 from the NIH (National Institute of Arthritis and Musculoskeletal and Skin Diseases)
Shotland 2003	USA	Retrospective cohort	PTLDS	18	M 44 (NR)k F 47.4 (NR)	39	33	NR	55	UT	National Institute of Allergy and Infectious Diseases protocol 96-I-0052 and National Institute on Deafness and Other Communication Disorders protocol 80-DC-106.
Skogman 2008	Sweden	Prospective cohort	ED-LD	72	6 (1–14) j	39	177	NR (matched)	NR (matched)	SQ	The Health Research Council in the South East of Sweden (FORSS), The County Council in Östergötland, The Center for Clinical Research in Dalarna (CKF), The Lions Foundation and The Samariten Foundation.
Skogman 2012	Sweden	Retrospective cohort	ED-LD	84	7 (2–14) j	48	84	NR (matched)	NR (matched)	SQ; PF	Research Council in Southeast Sweden (FORSS), the County Council in Östergötland, the Center of Clinical Research in Dalarna (CKF), the Swedish Society of Medicine, the Lions Foundation, and the Holmia Foundation
Stupica 2018_JAC	Slovenia	Prospective cohort	ED-LD	200	Median (IQR): 47 (35–58)	53	192	Median (IQR): 52 (35–61)	55.7	SQ	Slovenian Research Agency (grant numbers P3-0296 and J3-6788)
Stupica 2018_JAMA	Slovenia	Prospective cohort	EL-LD (n = 403) ED-LD (n = 200)	603	Median (IQR): 52 (40–61)	54.2	544	Median (IQR): 52 (37–62)	57.2	SQ	Grants P3-0296 and J3-6788 from the Slovenian Research Agency.
Svetina 1999	USA	Retrospective cohort	ED-LD	44	36 (7.8)	45	43	33 (9.4)	55	SQ; PF	Grant from the Helen and Irving Schneider family
Tager 2001	USA	Retrospective cohort	LD-LD (PTLDSa)	20	13.83 (2.41)	65	20	13.53 (2.67)	65	SQ	New York State Psychiatric Institute Research Support Grant, and by a grant from the Lyme Disease Association of New Jersey
Tjernberg 2010	Sweden	Retrospective cohort	ED-LD	20	66 (57–85)j,l 59 (34–77) j,l	63, 33l	23	49 (22–70) j	52	SQ	Medical Research Council of Southeast Sweden (FORSS)
van den Wijngaard 2015	Netherlands	Cross-sectional	EL-LD (n = 88)h ED-LD (n = 96) PTLDS (n = 476)	660	52 (NR)	50	NRg	NRg	NRg	SQ	Dutch Ministry of Health, Welfare, and Sport
Vazquez 2003	USA	Cross-sectional	ED-LD	43	9 (2–18) j	30	86	14 (3–26) j	50	UT	Grant from the NIH (K24AI01703), and a grant to the Yale Children’s Clinical Research Center from the General Clinical Research Centers, Program of the National Center for Research Resources (MO1-RR06022).
Wang 1998	USA	Retrospective cohort	PTLDS	21	NR	NR	22	NR	NR	SQ	NIH grant no. AR36308
Weinstein 2018	USA	Prospective cohort	EL-LD (n = 122) EL-LD (PTLDS, n = 6)	122	49.9 (NR)	47.5	26	54.7 (NR)	53.8	SQ	Global Lyme Alliance and the Lyme Disease Research Foundation, Inc.
Wills 2016	USA	Prospective cohort	EL-LD (n = 35) ED-LD (n = 46) LD-LD (n = 20)	101	49 (13)	51	NRc	NRc	NRc	SQ; QOL	Intramural Research Program of the National Institute of Allergy and Infectious Diseases at the NIH. This project was funded, in part, with federal funds from the National Cancer Institute, Center for Cancer Research, NIH (contract no. HHSN261200800001E).
Wormser 2015	USA	Prospective cohort	EL-LD	100	49.1 (10.9)	48	NRc	NRc	NRc	SQ; QOL; PF	Centers for Disease Control and Prevention (RO1 CK 000152 to G. P. W.).
Wormser 2019	USA	Prospective cohort	EL-LD (n = 31) ED-ED (n = 21)	52	50.2 (15.7)	34.6	104	50.4 (15)	34.6	QOL	RO1 CK 000152 from the Centers for Disease Control and Prevention (CDC) to G. P. W.
Zomer 2018	Netherlands	Retrospective cohort	EL-LD (n = 175) ED-LD (n = 125)	300	Median (range): 51 (18–87)	51.2	437	Median (range): 51 (18–87)	51.2	SQ	NR
Zotter 2013	Austria	Retrospective cohort	ED-LD	20	10.85 (2.87)	45	20	10.4 (3.28)	50	SQ	NR

Abbreviations: CLD, chronic Lyme disease; ED-LD, early disseminated Lyme disease; EL-LD, early localized Lyme disease; F, female; LD-LD, late disseminated Lyme disease; M, male; NIH, National Institutes of Health; NR, not reported; PF, prognostic factors; PTLDS, posttreatment Lyme disease syndrome; QOL, quality of life; SQ, sequelae; USA, United States of America; UT, utilities.

^aOriginally identified as chronic Lyme disease in studies.

^b6/17 were disease control patients, and 11/17 sibling control patients.

^cUsed the general population to compare health-related quality of life.

^dNormal healthy adults used for brief symptoms index and Center for Epidemiologic Studies depression scale were 924 and 2514, respectively. These unexposed (control) cohorts were from the literature and not recruited by the study.

^eOnly 45/60 were initially infected with LD (ie, single infection). And 15/60 were reinfected patients; their outcomes were not considered in this review.

^fThese percentages were given for the overall study cohort. The Review used a subgroup of 88/678 patients that met our diagnosis criteria for quality-of-life outcomes. The exact number of patients in each stage was indistinguishable.

^gNo control group was used for this health-state utility values elicitation study.

^hA total of 660 patients were included in the EQ-5D cross-sectional study (from different stages) to elicit disability weights for LD stages. Disseminated Lyme borreliosis did not distinguish between early or late stages.

ⁱThe mean age was given for the entire Lyme cohort of 63 patients. Age for patients with PTLDS (n = 42) or without PTLDS (n = 21) was not reported but the authors concluded there were no significant differences in demographic characteristics.

^jReported as median (range).

^kStudy only reported mean age for male ages 33 to 54, and female ages 41 to 53.

^lStudy reports Lyme neuroborreliosis with persistent symptoms (8/20) and Lyme neuroborreliosis without symptoms (12/20) characteristics, separately.

^m42 patients ultimately were considered to have PTLDS.

Quality Appraisal

Thirty-six studies (78%) met at least 50% of the quality assessment criteria; 17 (37%) met at least 80% of their respective criteria. Results from the quality appraisal are shown in Figure 2. For cohort studies, the domain most commonly unmet was whether patients were free of the outcome at the time of exposure measurement, whereas cross-sectional studies did not account for confounding factors (Supplementary Figure 7).

Figure 2.

Quality appraisal results.

Attributable Sequelae

Based on our eligibility criteria, 21 studies (14 NA, 7 EU) reporting sequelae outcomes for exposed and unexposed patients were included in our analysis [20, 22–24, 26, 37, 41, 42, 46, 47, 51, 52, 54–56, 58–60, 63–65]. We stratified by region and summarized the range in the proportion of sequelae reported for exposed versuss unexposed patients in Figure 3A–C. In North America (Figure 3A), weakness, neck pain, myalgia, fatigue, arthralgia, numbness/tingling, sleep disorder, memory difficulties, emotional fluctuation, word-finding, depression, and concentration difficulties are reported in higher proportions in exposed patients. However, the range in the proportion from the exposed and unexposed cohorts overlap for all sequelae except weakness and neck pain. In Europe (Figure 3B), neck pain, myalgia, arthralgia, paresthesia, facial nerve palsy, sleep disorder, poor appetite, and concentration difficulties are reported in higher proportions in exposed patients with overlap in all sequelae, except for paresthesia and sleep disorder (Supplementary Table 8). All sequelae were summarized in Supplementary Tables 5 and 6.

Figure 3.

A. Proportion of sequelae reported in eligible studies for Lyme disease cases and unexposed cohorts in North American countries. B. Proportion of sequelae reported in eligible studies for Lyme disease cases and unexposed cohorts in European countries. Each point represents a cohort of patients expressing the category of sequelae; the size of the points indicates the relative number of patients in that cohort (study). Raw data are in Supplementary Table 8. Abbreviation: LD, Lyme disease. C. Proportion of sequelae reported in eligible studies for exposed cohorts with posttreatment Lyme disease syndrome and unexposed cohorts in North America. Each point in the figure represents a cohort of patients expressing the category of sequelae; the size of the points indicates the relative number of patients in that cohort (study). Abbreviations: LD, Lyme disease; PTLDS, posttreatment Lyme disease syndrome.

Attributable Sequelae, by LD Stage at Diagnosis

Of the 14 NA studies reporting sequelae for both cohorts, LD stages at diagnosis were: early localized (n 1) [20], early disseminated (n 2) [22, 64], late disseminated (n 2) [46, 60], PTLDS (n = 6) [24, 41, 42, 47, 51, 65], and mixed between early localized and disseminated (n = 5, with 2 studies reporting PTLDS postdiagnosis) [24, 37, 51, 54, 59]. In patients with early localized LD, the sequelae reported with the highest proportion when compared to unexposed patients (ie, exposed vs unexposed patients) were fatigue (56% vs 43%), and body aches (36% vs 57%) [20]. In patients with early disseminated LD, sequelae reported with highest proportion were word finding issues (55% vs 14%) and joint or muscle pain (21% vs 5%) [22, 64], whereas patients with late disseminated LD reported highest proportions of frequent arthralgia (61% vs 16%), difficulty sleeping (47% vs 16%), and cardiac abnormalities (34% vs 35%) [46, 60].

Patients that were enrolled at the PTLDS stage reported the highest proportions of fatigue (79% vs 16%), depression (69% vs 20%), and arthralgia (67% vs 28%). Figure 3C depicts the ranges in proportion of sequelae reported. Neck pain, fatigue, coordination issues, sleep difficulties, memory difficulties, word-finding, and concentration difficulties are all reported in higher proportions in exposed patients and have no overlap in the range in proportion reported by the unexposed group.

Of the 7 European studies, LD stages at diagnosis were disseminated LB (n = 5) [24, 37, 51, 54, 59] and mixed between disseminated LB and early LB (n = 2) [52, 56]. Because the 2 mixed studies also contained patients with disseminated LB, we did not stratify any further.

Health-related Quality of Life (HRQoL)

Thirteen studies reported HRQoL associated with LD: 10 from the United States and 3 from Germany (Table 2) [20, 27, 30, 33, 35, 36, 39, 47, 48, 51, 54, 57, 62]. Time from diagnosis to study (follow-up duration) ranged from 8.4 days to 20 years. Most studies used the short-form 36 (SF-36) (n = 10).

Table 2.

Summary of Health-Related Quality-of-Life Associated With Lyme Disease

Study	Year	LD Stage at Diagnosis	Time from Diagnosis to Study (Follow-up)	Measurement Tool	Summary
North America
Fallon et al [39].	1999	PTLDS	3.8 (NR) years	Modified FIQ (VAS)	Total FIQ score (cases, 50.12 vs control, 14.02), total physical impairment (2.23 vs 0.07), and global well-being (5.28 vs 0.72) scores were significantly higher (P < .05) for PTLDS patients vs controls. VAS measures of joint pain (4.15 vs 0.85), fatigue (6.62 vs 2.98), depression (4.92 vs 0.90), ability to perform activities of daily living (4.62 vs 0.73), and impairment of memory and concentration (5.62 vs 0.76) were all significantly different from controls, suggesting greater impairment and reduced quality of life for PTLDS patients.
Shadick et al [47].	1999	PTLDS	6.0 (5.1) years	SF-36	PTLDS patients had poorer functional status than healthy controls on the SF-36 subscale for PF (85.4 vs 91), RP (68.3 vs 83.1), SF (69.1 vs 81.2), BP (69.5 vs 84.8) and global SF-36 (72.4 vs 80.6) score, suggesting that long-term impairment of functional status can occur in PTLDS patients.
Seltzer et al [62].	2000	Mix	4.25 (1.25 – 11.25) years	SF-36	There were no statistically significant differences in any of the 8 domains (PF, RP, BP, GH, VT, SF, RE, and MH) between 88 LD patients and healthy controls.
Kalish et al [20].	2001	Mix	10–20 years	SF-36	The standardized PCS and MCS for all stages of LD patients ranged from 50 to 52, and 51 to 54, respectively. These scores compared favorably with normative data scores from the general US population’s (control group) PCS of 50 (SD, 8) and MCS of 51 (SD, 9). There were no significant differences in these scores, or the 8 domains of the SF-36 questionnaire to suggest a different quality of life between LD patients and controls.
Aucott et al [35].	2013	Mix	7.9 (6.2) days (Follow-up: 6 months)	SF-36	At 6-month follow-up after treatment, the PTLDS group reported a lower quality of life for the RP domain (40.81 vs 49.13) and VT domain (45.22 vs 53.13) compared to the non-PTLDS control group. All other subdomains did not meet the predefined criteria of minimally important change. The study concluded that patients who develop PTLDS had significantly lower life functioning compared to those without PTLDS.
Jares et al [54].	2014	Mix	3.6 (NR) years	SF-36	There was a difference in one SF-36 domain score (VT) between the Lyme controls, 54.5 (NR) and non-Lyme controls, 64.5 (NR) (P = .02). The remaining seven SF-36 domains showed non-significant differences across when compared to the general US population.
Wormser et al [30].	2015	Early localized	16 (11, 20) years	SF-36	For the 8 domains measured by SF-36, early localized LD patients scored between 50.9 (RE) and 54.3 (VT). The control group used was the general US population which scored 50 for all domains. The long-term HRQoL of patients with early localized LD (PCS 51.9, MCS 52.8) was on average similar to that of the general United States population (PCS and MCS 50).
Wills et al [27].	2016	Mix	3.9 (0.5 – 11.3) years	SF-36	Overall mean quality of life was lower for LD patients (PCS 45.6, MCS 47.3) compared to the US population mean (control group, PCS and MCS 50). Overall scores for LD patients (PCS, 50.7; MCS, 50.1) were comparable to the US average after 36 months. HRQoL was lower for disseminated stages of LD compared to early localized LD (PCS, 42.7 vs 51.7; MCS, 44.5 vs 52.3) at baseline but similarly increased and were comparable to each other, and to the US population mean after 36 months. Multivariate analysis determined that comorbidities unrelated to LD was the only factor associated with lower HRQoL scores.
Rebman et al [51].	2017	PTLDS	3.6 (0.69 – 27.7) years	SF-36	Patients with PTLDS reported a poor health-related quality of life, PCS and MCS scores of 33.9 (9.7) and 42.9 (10.1), compared to the US population mean of 50.
Weinstein et al [57].	2018	Mix	8.4 (NR) days (Follow-up: 1 year)	SF-36	Overall, participants with LD scored lower on the SF-36 than controls, 52.31 (6.93) vs 55.09 (4.45). PTLDS patients at 1-year follow up reported greater functional impact resulting from their symptoms.
Europe
Back et al [36].	2013	Early disseminated	5.2 (3.6) years	EQ VAS; EQ-5D	HRQoL measured by the EQ VAS in patients with NB-associated vasculitis, was lower, 51.0 (24.8), than the age-matched general German population, 77.4 (19.0). The EQ-5D measure was also decreased for patients with NB-associated vasculitis at 0.55 (SD, 0.19). Quality of life seems to be worse in this subset of early disseminated LD patients than in common stroke patients without LD.
Dersch et al [33].	2015	Early disseminated	4.9 (2.74) years	SF-36	Quality of life estimates using SF-36 did not significantly differ between LNB patients (PCS, 51.12; MCS, 48.08) and healthy controls (PCS, 54.72; MCS, 51.77). However, LNB patients with residual symptoms have lower PCS quality of life scores (PCS, 46.33; MCS, 48.02) compared to LNB patients without residual symptoms (PCS, 57.48; MCS, 48.17).
Schmidt et al [48].	2015	Early disseminated	4.5 (3.7) years	SWL	No significant differences were measured for HRQoL between LNB patients and healthy controls using the Diener’s quality of life score (20.4 vs 19.2, P = .19) for satisfaction with life.

Abbreviations: BP, bodily pain; EQ-5D; EuroQol 5 dimensions; EQ VAS, EuroQol visual analogue scale; FIQ, fibromyalgia impact questionnaire; GH, general health; HRQoL, health-related quality-of-life; LD, Lyme disease; MCS, mental component score; MH, mental health; NR, not reported; PCS, physical component score; PF, physical functioning; PTLDS; posttreatment Lyme disease syndrome; RE, role emotional; RP, role physical; SD, standard deviation; SF-36, short-form 36; SF, social functioning; SWL, satisfaction with life scale; US, United States; v2, version 2; VAS, visual analogue scale; VT, vitality.

All 3 studies from the European Union enrolled patients with disseminated LB with varying comorbidities resulting in heterogeneous findings [33, 36, 48]. Schmidt et al concluded that HRQoL was not different between patients with Lyme neuroborreliosis (LNB) and unexposed patients based on the satisfaction with life scores [48]. Back et al used the EuroQol 5 dimensions (EQ-5D) questionnaire and concluded that NB-associated vasculitis resulted in lower HRQoL than the common unexposed German patient with vasculitis [36]. Dersch et al used the SF-36 and reported that, overall, patients with LNB did not experience different HRQoL from unexposed patients. However, patients with LNB and residual symptoms had a lower physical component score (PCS) than those without residual symptoms [33].

From North America, 1 study enrolled patients with early localized LD [30], 3 studies enrolled patients with PTLDS [39, 47, 51], and the remaining enrolled patients from various stages [20, 27, 35, 54, 57, 62]. Wormser et al concluded that the long-term HRQoL of patients with early localized LD was, on average, similar to that of the general US population [30]. For patients with PTLDS, all 3 studies concluded that HRQoL was partially or fully impaired for exposed patients in comparison to healthy controls or the general US population [39, 47, 51]. All HRQoL data are available in Supplementary Table 9.

All HSUV and prognostic factor outcomes are reported in Supplementary Tables 10–12.

Discussion

We systematically reviewed 46 studies for long-term sequelae, HRQoL, and prognostic factors associated with Lyme disease in North America and Europe. Over two-thirds of the studies included were conducted in North America. The higher number of publications from North America also suggests that disease progression remains of high interests for these Borrelia strains.

The overall quality of the studies was heterogeneous and not considered high. Over half of the studies did not meet 80% of their respective JBI checklist criteria in addition to many studies not reporting proportion of sequelae for both exposed and unexposed cohorts. Therefore, we were only able to include 21 studies that reported sequelae outcomes for both cohorts in our analysis, stratified by geography. Comparison of patients from the 2 regions showed that a higher proportion of exposed patients from North America reported depression, memory difficulties, and numbness, whereas a higher proportion of exposed patients from the European Union reported paresthesia and facial nerve palsy. Other sequelae that were reported in higher proportions in exposed patients but were common between the 2 regions included: neck pains, myalgia, arthralgia, sleep disorder, and memory difficulties. Differences of sequelae experienced between the 2 regions are similar to previous reports suggesting that the North American strains of Lyme disease are more cognitive in nature, whereas the European strains evolve into neurological sequelae [15]. There is a possibility that a reporting bias exists, as there were multiple cases of facial nerve palsy reported in North America that were excluded from analysis [28, 44, 49, 65].

Most of the sequelae presented in Figure 3 are self-reported and can be considered subjective with the exception of facial nerve palsy, cardiac abnormalities, and seizures. In Supplementary Table 6, we summarized studies reporting differences in some of these subjective sequelae (eg, depression, concentration difficulties, fatigue) using standardized tests such as the Beck depression index, Wechsler memory scale, and fatigue severity scale. However, these studies did not meet the criteria set forth for the attributable sequelae analysis and were summarized separately. Based on quality appraisal results, confounding was not identified or adjusted for in 50% of cross-sectional studies, whereas <20% of cohort studies had determined whether the patient was free of the outcomes (eg, sequelae) at the time of exposure.

Patients with PTLDS reported depression 18%–69% compared to 0–20% for unexposed patients and arthralgia 23%–67% compared to 7%–28%. Although these sequelae are considered nonspecific and common to many health conditions for primary care patients [66], our review suggests that patients with PTLDS do experience long-term subjectively reported sequelae. Current hypotheses explore the idea that patients thought to have PTLDS may in fact have symptoms from another health condition that are misattributed to PTLDS; there is a possibility of anchoring bias.

A common finding was that patients with PTLDS experienced a lower HRQoL than unexposed cohorts (eg, healthy controls). This is plausible and further supports our conclusions that patients with PTLDS may experience long-term subjective sequelae. Patients from Europe with LNB, despite having objective residual findings were not so different from unexposed patients, as was similarly found in 1 study in patients with early localized LD. We could not draw a definitive conclusion on the HRQoL in these patients given the range in follow-up times, different measurement tools, and sample sizes. This variation in HRQoL was also reflected in studies reporting HSUVs, where the utilities differed depending on the study population and preference elicitation tool.

Our review is subject to several limitations. In an attempt to report attributable sequelae, we excluded case report and case series study designs, which inadvertently excluded less common clinical outcomes. The overall quality of the included studies was not high as many studies did not adjust for all possible confounding factors (eg, age, sex, education level, comorbidities), ensure patients were free of the outcome (eg, persistent symptoms related to Lyme disease) at the exposure measurement, or supplement sequelae results with statistical test results or rate statistics. In 12 prospective studies, only 25% (3/12) determined that participants were free of outcomes at the start, and 50% (6/12) reported attributable sequelae. Therefore, it was inappropriate to further stratify our analysis for prospective studies only. We recognize that cross-sectional and retrospective studies have their inherent limitations and present lower quality evidence but decided against removing these studies from our analysis to also avoid potential publication bias.

In a Canadian study using retrospective LD registry data, arthritic sequelae were reported in 100% of patients with late disseminated LD (93% arthralgia and 7% arthritis) [67]. Although arthritis was reported by 26%–100% of patients in 7 different studies [20, 25, 28, 40, 42, 44, 49], all of these studies did not compare to an unexposed cohort and were thus ineligible and excluded arthritis from being reported in our attributable sequelae analysis. The high proportion of reported arthritis in the 7 studies result from misreporting of transient arthralgia as arthritis [67]. As such, we need to emphasize that this list of attributable sequelae be interpreted with caution; additional sequelae arising from LD are possible even though they were not reported in our analysis. Because the LD stage at diagnosis was indistinguishable in some studies, we were not able to include all studies into our analysis for a larger sample size. Due to the heterogeneity of the sample sizes, LD stages at diagnosis and follow-up, we determined it was inappropriate to perform a meta-analysis to estimate an overall incidence of sequelae or HRQoL associated with various LD stages. Furthermore, we were unable to definitively conclude that exposed patients and controls were not used in multiple studies by the same group of authors and so acknowledge that there may be an impression of a larger patient size or a possibility of double counting. This review may be subject to language bias due to exclusion of non-English studies, reducing the number of European studies and subsequently underestimating the sequelae reported by European patients with LB.

To our knowledge, this is the first systematic review to provide an overall summary of the clinical outcomes of LD and serve as an updated confirmatory study using a strict eligibility criteria to improve our understanding of PTLDS. Due to the controversial nature of LD diagnosis, our eligibility criteria for diagnosis was consistent to ensure that patients included in our analysis were all confirmed cases of LD. This study provided a sense of the sequelae and HRQoL experienced by LD patients in North America and Europe, which can help in clinical management and further understanding of the natural disease progression of LD. Secondary outcomes (prognostic factors, HSUVs) identified in this review can be used in decision-analytic modeling to determine national health and economic burden of Lyme disease in addition to the value of potential intervention or prevention strategies. Although the conclusion of this review was not unexpected, it provides timely evidence to the understanding of LD and its consequences and promotes the idea that we need well-designed prospective longitudinal studies.

Conclusions

Our review suggests that patients with Lyme disease are more likely to report a nonspecific long-term, mainly subjective, sequelae, especially in patients experiencing persistent symptoms posttreatment. The clinical outlook presented in this review can provide some evidence to support health policy decision makers. However, opportunities exist for future prospective longitudinal studies to understand additional risk factors and disease progression for the disseminated Lyme disease and PTLDS.

Supplementary Data

Supplementary materials are available at Clinical Infectious Diseases online. Consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or comments should be addressed to the corresponding author.

Notes

Acknowledgements. The authors thank Public Health Ontario’s (PHO's) Library Services’ Susan Massarella and Lindsay Harker for their assistance in developing search strategies and managing results, and Juan Pablo Diaz Martinez for his assistance with development of the figures.

Disclaimer. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Financial support. This work was supported by the Canadian Institutes of Health Research (CIHR) project grant PJT149087 held by B. S.; S. M. is supported by a CIHR Frederick Banting and Charles Best Canada Graduate Scholarship Doctoral Award GSD-159274.

Potential conflicts of interest. The authors: No reported conflicts of interest. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest.