Figure 1

Figure 1. Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) (10) flow diagram of the 2 literature searches performed in review of seropositivity for Borrelia burgdorferiin China,...

Figure 3

Figure 3. Forest plot illustrating seropositivity estimates for Borrelia burgdorferi,by province, China, 2005–2020. The red horizontal line indicates the summary estimate based on the primary analysis; error bars indicate 95%...

Figure 2

Figure 2. Estimated seropositivity for Borrelia burgdorferi, by province, China, 2005–2020. Ixodes persulcatusticks, among the most frequently identified ticks in China, have been found across the northeastern and...

Our literature review identified 3,657 articles that focused on China; 48 articles met the selection criteria (Figure 1), of which 42 articles met the criteria for the primary analytical strategy. In total, these 42 articles provided 72 estimates of seropositivity that we extracted for analysis. Some articles produced seropositivity estimates for multiple provinces or years (Appendix Table 2). Six articles did not meet the criteria for the primary analytical strategy; thus, they contributed data to only the 2 diagnostic sensitivity analyses. From the included studies, we compiled a description of estimates by subgroup (Table 1), by exposure group and province (Table 2), and by province (Figures 2, 3).

For the primary analytical strategy, the reported IgG seropositivity estimates based on a single-tier test (EIA or IFA) ranged from 0 to 37%; the fixed-effects modeled summary estimate was 9.1% (95% CI 7.5%–10.7%) (Table 1). When the random-effects model was used for the primary analytical strategy, neither estimate nor variance differed. The total sample size producing this summary estimate was 34,719 (Appendix Table 2).

Fewer articles and estimates were available for the diagnostic sensitivity analyses. For the sensitivity analysis based on 35 estimates (sample size of 9,446 obtained from 5 articles) that did not distinguish between IgG and IgM results, seropositivity was 14.5% (95% CI 11.8%–17.2%). For the sensitivity analysis that used a 2-tier testing system (16 estimates obtained from 6 articles with a sample size of 8,837), seropositivity was 1.8% (95% CI 0.9%–2.7%) (Table 1).

Seropositivity for the clinical suspicion sample was 7.1% (95% CI 6.4%–8.0%). Lyme disease seropositivity estimates by exposure risk populations were 4.5% (95% CI 3.9%–5.1%) for low risk, 6.1% (95% CI 5.4%–6.7%) for medium risk, and 10.0% (95% CI 9.6%–10.4%) for high risk (Table 1). The odds ratio of high exposure risk seropositivity compared with low exposure risk seropositivity was 2.4 (95% CI 2.1–2.7) and of moderate exposure risk seropositivity compared with low exposure risk seropositivity was 1.4 (95% CI 1.2–1.6).

Variation by province was substantial; the highest seropositivity estimates were 23.1% for Heilongjiang Province and 16.2% for Neimenggu (Inner Mongolia) Province (Figure 2). Moreover, variation across provinces was substantial (Figure 3). There was no discernable trend over time for seropositivity (data not shown).

Our systematic literature review of B. burgdorferi seropositivity in China generated summary estimates by diagnostic test, exposure risk, sex, age group, province, and year. Depending on the testing algorithm applied, the seropositivity ranged from 1.8% to 14.5%, reflecting Lyme disease endemicity in the population. Combined with the widespread distribution of Ixodes ticks, specifically I. persulcatus ticks in many provinces, this analysis reinforces that Lyme disease is a public health problem in China.

The summary estimates of 9.1% among EIA/IFA positive samples and 1.8% among samples confirmed with WB fell within the range identified in Europe. In Germany, the nationwide, population-based cross-sectional KiGGS study estimated seropositivity among children and adolescents of 4.8% by single-tier ELISA testing (4% when confirmed by line blot) (14). A similar nationwide, population-based, cross-sectional study among adults in Germany (DEGS) reported overall seropositivity of 9.4%, confirmed by line blot (15). A cross-sectional health survey among a representative sample of adults in Finland reported seropositivity of 3.9% according to 2-tier testing (16). A representative sample of healthy blood donors from the Tyrol region of Austria reported a seropositivity range of 1.5%–7.2% from samples confirmed by line blot (17). A regional study in Turkey among healthy volunteers revealed seropositivity of 4.1% by single-tier testing with ELISA and 2.2% confirmed by WB (18).

In China, results of the diagnostic sensitivity analyses were consistent with expectations based on Lyme diagnostic testing limitations. Several studies did not adequately delineate the results by IgM or IgG positivity, and this joint numerator resulted in a substantially higher estimate than IgG seropositivity alone. IgM responses to Borrelia may persist over time, although IgM reactivity alone without isotype switching to IgG may reflect a false-positive result (12,19). False-positive results may provide an explanation for the higher seropositivity found when including results that did not distinguish between IgM and IgG. Testing results were further complicated in many studies by assays that used a whole-cell sonicate; such a lysate generates multiple antigens that can increase the likelihood that cross-reactive IgG or IgM creates a false-positive result compared with newer EIAs that focus on a reduced set of well-defined purified antigens specific to B. burgdorferi genospecies (12).

In many countries in Europe and in the United States, a 2-tier testing system for Lyme disease is used for clinical diagnosis and seroprevalence assessments. The sensitive first-tier test uses an ELISA, or less often an IFA, followed by a highly specific, second-tier WB if the ELISA is positive or equivocal. Diagnostic sensitivity can vary widely, with estimates ranging from as low as 14% during the early stages of disease to 100% as symptoms and manifestations evolve (12). The value of this standard 2-tier testing approach is improved specificity compared with ELISA or IFA alone (9,20). Specificity in all clinical phases is robust at >99% after the second-tier test. Recently, a modified 2-tier testing system based on 2 EIAs, which substitutes an EIA for the second-tier WB, has been implemented. Third-generation EIAs focus on select antigens; as a result, pairing with different EIAs enables substantial orthogonality that improves sensitivity while maintaining specificity (12). Neither of those 2-tier testing approaches has been widely adopted in China (13), where studies reporting the seropositivity of antibodies to B. burgdorferi relied primarily on the first-tier EIA or IFA and less often on the confirmatory, specific WB.

Furthermore, diagnostic performance data are not readily available from China, particularly across the range of testing procedures used. Some studies acknowledged use of reagents provided by the China Centers for Disease Control and Prevention, others used local commercial test kits, a few used nondomestic test kits, and several others did not describe the assays used. The China Centers for Disease Control and Prevention described criteria for standardization of a WB based on lysates of B. garinii bacteria; the assay used a single band for IgG or IgM, a criterion that differs from guidelines in Europe and the United States that require at least 2 of 3 bands by IgM or 5 of 10 bands by IgG to be classified as positive by WB (12,13). In addition, interpretation criteria were not consistent across associated articles. The few articles from our review that reported seropositivity with a confirmatory WB result led to a summary estimate substantially lower than the single-tier test (1.8% vs. 9.1%). The limited information on quality of assays, accreditation, and validation remains a major limitation of analytic interpretability. Nevertheless, the sampling for the 2-tier testing strategy occurred in 4 provinces across all types of exposure categories, providing some evidence of a representative seroprevalence estimate (Appendix Table 2).

Seropositivity was higher for populations that had been assessed as having a higher risk for exposure to a natural foci of Lyme disease, either by occupation or by location. These data were consistent with targeted samples of higher risk occupational groups from other countries. For instance, IgG seropositivity among farmers was 5.5%–9.7% in Belgium (21) and 10%–13.7% in Poland (22,23). Forestry workers are among the most sampled high exposure risk occupational groups; reported seropositivity was 14% in Lithuania (24), 31% in Hungary (25), 7.8% in Italy (26), 11.8% in Serbia (27), 10.9% in Turkey (28), 21.6% in Belgium (29), and 14%–34% in Poland (22,30). Likewise, increased seropositivity seen in medium to high exposure risk populations in China and consistency with data from Europe provide initial confidence in the overall estimates provided in this report.

The summary estimate in the clinical suspicion sample, which reflected a mixture of studies focused on history of a suspected tick bite or clinical suspicion of Lyme disease, was 7.1%. The TBD STING study in Sweden and Finland reported seroconversion after a tick bite for 3.5% of participants (31). Nonetheless, seroconversion does not necessarily reflect clinical infection because Lyme disease manifestation during the 3-month follow-up period did not develop for 57.6% of the Borrelia-infected persons who seroconverted. Among the seroconversions that resulted in clinical manifestations, these included erythema migrans (85%), Borrelia lymphocytoma (3%), nervous system disease (6%), or nonspecific symptoms of Lyme disease (6%). A study conducted in China documenting clinical manifestations after tick bite reported a lower proportion of erythema migrans (69%) and a higher proportion of nervous system disease (21%) and arthralgia (21%), among other manifestations (32). In the United States, a randomized controlled trial of a vaccine candidate documented an asymptomatic proportion of <10%, potentially arising from the shorter duration of follow-up for symptoms compared with that in Europe (33). Notwithstanding, differences in asymptomatic proportions and clinical manifestations between studies and countries may reflect differences in study design, duration of follow-up, diagnostic quality, timing of postinfection treatment, antibody waning, and circulating genospecies.

The highest summary seropositivity estimates were detected in 2 provinces in northeastern China (Heilongjiang Province, 23.1%) and western China (Neimenggu Province [Inner Mongolia], 16.2%). I. persulcatus ticks are among the most frequently identified ticks in China and have been found across the northeastern and select western, central, and eastern, provinces (6). Several other provinces that border Heilongjiang and Neimenggu Provinces have frequently reported the presence of I. persulcatus ticks, although the reported seropositivity has been lower (7%–10%). The lower calculated estimates within these provinces could reflect differences in the sampled exposure groups because studies from Heilongjiang and Neimenggu Provinces largely focused on populations for which higher infection prevalence was expected (e.g., forest residents and forestry workers). In addition, no samples from Heilongjiang and Neimenggu Provinces were tested by using the more conservative 2-tier testing strategy, which probably would have resulted in lower seropositivity estimates. Alternatively, lower seropositivity in border regions could reflect true differences in exposure risk. Other regions of China, particularly those in the southern and eastern regions, reported somewhat lower seropositivity (2%–10%). The distribution of I. persulcatus ticks is limited in these regions, although I. sinensis and I. granulatus ticks have been reported and are considered to be the main vectors in these provinces. However, demonstration of vector competency and efficiency of I. sinensis and I. granulatus ticks as vectors of B. burgorferi is unclear (5,7).

Other caveats to consider include the possibility that seropositivity may be driven by persons who become infected in higher incidence regions but reside in regions without efficient local transmission, because their limited awareness of Lyme disease precludes appropriate personal prevention measures. Another consideration is the local prevalence of Lyme disease for such persons. Despite good specificity of the test, a low a priori probability of disease will lead to a lower positive predictive value for true disease (12). Regardless of these caveats, increasing tick distributions across China has been attributed to planned reforestation and changing land use patterns leading to suitable environments to maintain the tick enzootic cycle and ultimately Borrelia transmission (3).

Among the study limitations, there were substantial variations in populations sampled (including persons seeking clinical care, or convenience samples), in risk exposure population targeting, and in varying study designs (none of the studies were designed to be nationally representative). More than half of the studies were conducted among a higher risk exposure population that probably elevated the summary estimate. With additional information on the percentage of the country’s population at different levels of risk, a weighted average could be obtained, although this type of data is difficult to quantify. Second, the included studies used a range of testing methods that may not be comparable. For example, IFA was widely used to estimate seropositivity, but this traditional, manual method relies on the experience of the technician, leading to potentially lower specificity compared with enzyme immunoassay (EIA or ELISA) (9,20). In addition, 2-tier testing was not uniformly used, which could have resulted in potentially higher rates of false-positive results. Third, exposure risk may have been misclassified because an accurate description of specific testing populations may be missing from source manuscripts. Fourth, although B. garinii and B. afzeli have been reported as the predominant circulating genospecies, other circulating nonpathogenic genospecies may cause a false-positive result (3,4). Last, seropositivity estimates reflect exposure risk for infection regardless of clinically apparent disease; therefore, these summary estimates should not be interpreted as reflecting risk for clinical disease. Collectively, these limitations portend to overestimate seropositivity compared with other results as noted in Europe.

In conclusion, the results from this meta-analysis demonstrate seropositivity to B. burgorferi in China over the past several decades, particularly in certain provinces and in high exposure risk populations. By itself, however, the utility of this information for driving public health policy is limited because it gives no indication of clinical burden, either overall or by severity, and may not accurately represent geographic variations in risk. The expanding geographic range of infected ticks and increased likelihood of contact with humans will continue to present a public health challenge for China.