Skip to main content


Springer Nature is making SARS-CoV-2 and COVID-19 research free. View research | View latest news | Sign up for updates

Public health impact of strain specific immunity to Borrelia burgdorferi



Lyme disease, caused by Borrelia burgdorferi, is the most common tick-borne infection in the United States. Although humans can be infected by at least 16 different strains of B. burgdorferi, the overwhelming majority of infections are due to only four strains. It was recently demonstrated that patients who are treated for early Lyme disease develop immunity to the specific strain of B. burgdorferi that caused their infection. The aim of this study is to estimate the reduction in cases of Lyme disease in the United States that may occur as a result of type specific immunity.


The analysis was performed based on three analytical models that assessed the effects of type specific immunity. Observational data on the frequency with which different B. burgdorferi strains cause human infection in culture-confirmed patients with an initial episode of erythema migrans diagnosed between 1991 and 2005 in the Northeastern United States were used in the analyses.


Assuming a reinfection rate of 3 % and a total incidence of Lyme disease per year of 300,000, the estimated number of averted cases of Lyme disease per year ranges from 319 to 2378 depending on the duration of type specific immunity and the model used.


Given the assumptions of the analyses, this analysis suggests that type specific immunity is likely to have public health significance in the United States.

Peer Review reports


Lyme disease, caused by Borrelia burgdorferi, is the most common tick-borne infection in the United States. Additionally, recent evidence suggests that the 30,000 reported cases actually represent only 10 % of the true incidence in the United States [1]. The majority of reported Lyme disease cases originate in the Northeastern (82 %) and Midwest (9 %) United States [2], although the geographic range is currently expanding [35]. Despite the fact that more than 16 B. burgdorferi lineages exist in these areas, the majority of disseminated human infections are caused by just four. Similarly, the subsets of B. burgdorferi lineages that cause infection in wild animals vary among animal species [68].

B. burgdorferi lineages have been categorized by their allele at the highly variable outer surface protein C (ospC) locus, and are often referred to by their ospC genotype. While there is ample evidence that the surface exposed OspC lipoprotein is necessary for infection, and that there is a strong and protective immune response against OspC [911], there is currently limited direct evidence that the variability at ospC directly affects differential infectivity of humans or other animal species. Nevertheless, the variability at the ospC locus, along with the high linkage among all genetic loci [1215], makes ospC a good marker to delineate different B. burgdorferi lineages.

Based on a unique data set of culture-confirmed initial and repeat episodes of early Lyme in antibiotic-treated patients with erythema migrans, two conclusions were made. One was that recurrence of early Lyme disease in such patients is due to reinfections rather than relapses of the original infection [16]. Secondly, based on multiple modeling techniques and using observed data on the frequency with which different strains of B. burgdorferi were cultured from initial versus recurrent episodes of early Lyme disease, it was demonstrated that type specific immunity lasting at least five years is likely to develop following infection [17].

In this analysis, we attempt to estimate the reduction in cases of Lyme disease that may occur as a result of the recently demonstrated type specific immunity. The analyses suggest that the presence of type specific immunity is likely to have public health significance.


Several assumptions were made to perform the analyses. The probability of exposure to various strains of B. burgdorferi due to a bite from an infected tick was estimated based on the frequency with which different strains of B. burgdorferi were recovered from skin biopsy samples of patients with early Lyme disease associated with erythema migrans in the Lower Hudson Valley region of New York State (Table 1) [17, 18]. These published data were derived from the distribution of strains cultured from the skin of a sample of 200 patients (from a total of 298) with no prior history of Lyme disease that were diagnosed between 1991 and 2005, as described previously [17, 18]. B. burgdorferi lineages were previously classified according to the allele at the ospC gene as previously described e.g., [15, 18, 19]. The present analysis was performed in 2014 and is exempted from Institutional Review Board (IRB) approval as it involved the analyses of existing published data (category 4), as defined by the Office for Human Research Protections (

Table 1 Frequency of the different OspC types that were cultured from the skin of 200 patients with erythema migrans [17, 18].

Additional assumptions that were made included that type specific immunity follows infection with all strains of B. burgdorferi and is absolute, lasts ≥ 5 years, and is restricted to a single OspC genotype. Based on our prior observations we assumed that 3 % of Lyme disease cases are due to reinfections, but also assessed the impact of type specific immunity if 1 % or 5 % of infections occur as a result of a reinfection.

Number of B. burgdorferi infections averted due to strain specific immunity

The potential impact that strain-specific immunity could have on human Lyme disease incidence was assessed using three different mathematical modeling structures. Analytical models are necessary to estimate what would have occurred in the absence of a factor such as strain-specific immunity, a process known as precluding. The models employed account for the total number of human cases as well as the number of averted cases. In this study, the term “averted cases” refers to cases that would have occurred in the absence of strain-specific immunity, holding all other model variables constant.

The three analytical models have different underlying assumptions, different sets of limitations, and different degrees of realism, and are thus useful to assess varying aspects of the effects of strain-specific immunity (Table 2).

Table 2 Summary of the characteristics of the three models used to estimate the number of averted cases of Lyme disease due to type specific immunity

Deterministic probability model

With the deterministic probability model the proportion of human cases that are averted due to strain-specific immunity is equivalent to the proportion of patients that are exposed to a B. burgdorferi strain with which the patient has been previously infected. This proportion can be calculated as

$$ A=\varSigma {f_i}^2 $$

where f i is the frequency of the ith strain (Table 1) and A is the proportion of cases averted due to strain-specific immunity; eg., the proportion of patients that are exposed to strain K on two separate occasions occurs at a frequency of f K 2 = 0.3652 = 0.1332 (Table 1). Based on the ~30,000 Lyme disease reported cases in the United States and an estimated 3 % reinfection rate the number of averted cases can be calculated as

$$ N=\left(0.03\ *A\right)/\left(1 - A\right)\ *\ 30,000 $$

where N is the total number of cases averted in relationship to the total number of cases in the United States.

Equilibrium dynamic model

In the equilibrium dynamic model, humans can change from susceptible (S) to infected (I) to immune (R) and back to susceptible (S). The effect of strain-specific immunity on human Lyme disease incidence is the sum of number of averted cases from exposure to each individual B. burgdorferi strain.

Patients transition from susceptible to strain i (S i ) to infected by strain i (I i ) at rate β i , equivalent to the incidence rate of strain i.

$$ {I}_i={\upbeta}_{\mathrm{i}}*\ {S}_i $$

The incidence rate for each strain i is scaled to the total incidence rate for all B. burgdorferi strains using the frequency at which each strain is found in human infections such that

$$ \beta = \varSigma \kern0.5em {\beta}_i $$

Patients transition from the infected state (I i ) to the immune state (R i ) at a rate of 1, such that the transition occurs with each iteration of the model (1 year) with 100 % probability. Patients are assumed to transition from the immune state (R i ) back to the susceptible state (S i ) in either 5 years (γ = 0.2) or 30 years (γ = 0.0333). These 2 limits were selected based on prior estimates of the minimum duration of strain specific immunity [17] or a maximum approximating life-long immunity. Therefore, the number of averted cases of Lyme disease can be estimated as the number of individuals in the infected or immune states (I i  + R i ) that are exposed to strain i.

$$ A={\beta}_i*\ \left({I}_i+{R}_i\right) $$

where A is the number of averted cases as a function of the incidence rate of Lyme disease (β i ).

Individual-based stochastic simulation model

In the stochastic simulation model, we simulated 100,000 individuals over a 30 year period in which they could be exposed to an infectious tick bite. Each individual has a set per year probability of exposure to an infectious tick bite which is drawn from an exponential frequency distribution such that most individuals have a very low per year probability of exposure. The rate parameter of the exponential distribution was chosen such that the reinfection rate would equal 3 % of total infections at an incidence rate of 100 cases of Lyme disease per 100,000 people.

In each year that an individual is available to be exposed to an infectious tick bite, the simulation determines if the individual is exposed to an infectious tick bite given the per year probability of exposure of the simulated individual and a random number generator. If the individual is exposed to an infectious tick bite, the probability of exposure to a particular borrelial strain was assumed to equal the frequency at which that strain has been recovered from patients (Table 1). If the simulated individual has been infected by the chosen strain within the time frame of strain-specific immunity, the individual is considered immune and an averted case is recorded. If the individual has not been previously infected or if the duration of strain-specific immunity has ended, an infection with the chosen strain is recorded. The simulation records the number of averted and successful infections among 100,000 simulated individuals. For every combination of parameters, 200 simulations were run each with 100,000 individuals. Parameters included varying duration of strain-specific immunity (5 or 30 years) and varying incidence rates of Lyme disease (0–500 per 100,000 people).


If the reinfection rate is 3 % and the total number of Lyme cases per year is 300,000, the estimated number of averted cases of Lyme disease per year ranges from 319 to 2,378 depending on the model employed and the particular assumptions used for that model (Table 3). With different assumptions on the incidence of Lyme disease and on the frequency of reinfections, however, the number of averted cases per year might be as few as 11 or as many as 4,100 (Table 3 and Additional file 1: Figure S1).

Table 3 Estimation of the yearly number of averted Lyme diseases cases in the United States based on the number of Lyme disease cases reported (approximately 30,000) or estimated (around 300,000) using three different analyses. The number of averted cases was calculated using three different estimates of reinfection rates: 1 %, 3 %, and 5 %

Available published data suggest that 83.3 % of patients with blood culture-confirmed disseminated B. burgdorferi infection from the Lower Hudson Valley region of New York State are infected with one of just four OspC genotypes: A (accounted for 23.5 % of positive blood cultures), B (for 14.4 %), I (for 12.1 %) or K (for 33.3 %) [18]. Assuming that a patient was initially infected by one of these four strains and that patients are exposed to B. burgdorferi strains at rates consistent with published data on initial human infections (Table 1, Additional file 2: Table S1), the results of the deterministic model suggest that the probability of a positive blood culture in patients with a recurrence of erythema migrans would be reduced by approximately 25 % due to strain-specific immunity.


The number of cases of Lyme disease in the United States continues to rise over time and expand geographically [3, 5]. Prevention methods rely primarily on measures to avoid tick bites, which have had variable rates of adherence and success [20]. Unfortunately, no vaccine to prevent B. burgdorferi infections is currently available for humans.

The results of this study suggest, given the assumptions of an annual 3 % reinfection rate and 300,000 total cases per year, that were it not for the presence of type specific immunity there might be as many as 2,300 additional cases of Lyme disease per year in the United States. Although only recently recognized [17], the presence of type specific immunity to strains of B. burgdorferi is therefore likely to have public health significance. Furthermore, the frequency of Lyme disease cases with hematogenous dissemination of the spirochete would also be appreciably higher in the absence of type specific immunity. Our findings suggest a 25 % risk reduction in having a positive blood culture during a recurrence if the initial infection was caused by one the four genotypes most often associated with detectable spirochetemia. In a previously published study the frequency of a positive blood culture for B. burgdorferi in 41 patients with a prior history of Lyme disease was 31.7 % versus 46.5 % in 172 patients during a first episode of Lyme disease [18]. Thus, the observed 31.5 % reduction in the rate of positive blood cultures in patients with prior Lyme disease in that study is consistent with the existence and clinical relevance of type specific immunity.

The public health effects of type specific immunity have been previously observed with several pathogens. For example, patients infected with one variety of the influenza virus are protected from reinfection with the same strain, but remain susceptible to other strains. The public health effect of type specific immunity in human influenza has temporal limitations due to the rapid evolution of novel strains [21], which is not apparent in B. burgdorferi populations.

Limitations of our study center around the accuracy of several key assumptions: that Table 1 reflects the overall frequency of infection by different strains of B. burgdorferi in the United States, that type specific immunity is restricted to just a single genotype, and that type specific immunity is 100 % protective for that genotype. The reported frequencies at which humans are exposed to each B. burgdorferi strain (Table 1) may be inaccurate due to geographic variation in the frequencies of strains in ticks and due to differences in cultivability among strains. The frequencies of B. burgdorferi strains in ticks differs among northeastern, Midwestern, and California Lyme disease foci e.g.,[6, 15, 22, 23] implying that human exposure rates differ [24]. These differences would alter the calculations of the public health effect of type specific immunity for the 18 % of infections that occur outside of the northeastern foci [2], although it is difficult to predict if the calculated number of averted cases would increase or decrease from those reported here.

Inaccuracies in the reported frequencies at which humans are exposed to each B. burgdorferi strain (Table 1) may also be caused by culture biases. However, several lines of evidence suggest that it is unlikely that culture bias has resulted in the uneven detection of strains in human blood. First, there are differences in the frequencies of strains in cultures of skin versus blood of humans. Second, non-culture methodologies performed with various animal species produce similar types of frequency biases, suggesting that the biases are not the result of differences in cultivability. In addition to culture biases, we did not consider in the analyses that ticks may transmit more than one strain of B. burgdorferi simultaneously [6, 19], and thus even if the person bitten was immune to reinfection by one of the transmitted strains, that individual might still develop Lyme disease from the other transmitted strain(s).

The accuracy of the assessment of the public health impact of type specific immunity is also impacted by the assumption that type specific immunity is restricted to just a single genotype and that it is 100 % protective for that genotype. The analyses assumed that each strain elicited a protective response against infections with the same strain but afforded no cross protection for other strains. It is possible that partial or full protective immunity also develops against additional strains [25], which would elevate the estimate of the public health effect of this type of immunity. The analyses also assumed 100 % protection from reinfection. Relaxing this assumption would decrease the estimate of the public health impact of type specific immunity. Experimental infections or data sets with larger numbers of patients with recurrent infections are necessary to determine the extent of cross immunity among strains and the degree of protection.

Although none of these assumptions is likely to be completely accurate, making a precise estimate of the impact of type specific immunity impossible to determine, it is reasonable to conclude that the presence of some degree of type specific immunity should serve to reduce the number of recurrent infections. In addition, the impact of immunity on Lyme disease prevention is likely to be higher than we have estimated, since the immune response associated with later manifestations of Lyme disease, such as Lyme arthritis, was not considered in this analysis and is likely to be broader and longer lasting than would occur following erythema migrans [26].

The molecular mechanisms leading to type specific immunity in B. burgdorferi are not known. Immune responses targeting OspC have been the focus of considerable research, but evidence to support the hypothesis that an immune response to OspC is responsible for type specific immunity is indirect. For example, transient immunity to reinfection with the same strain of B. burgdorferi has been demonstrated in laboratory animals and mice immunized with one OspC protein are selectively protected against infection with only B. burgdorferi expressing the same OspC protein, not against strains expressing a different OspC protein [27, 28]. Strain-specific immunity is unlikely to be restricted to OspC and may develop against multiple B. burgdorferi strain-specific proteins that are in genetic linkage with OspC e.g., [13, 19, 29]. Further empirical studies that characterize multiple antigens are necessary to determine what antigens are responsible for strain-specific immunity.


Recent data have demonstrated the development of type specific immunity against the infecting strain of B. burgdorferi in patients treated for early Lyme disease. The current analysis suggests that this observation is likely to have public health significance.


  1. 1.

    Hinckley AF, Connally NP, Meek JI, Johnson BJ, Kemperman MM, Feldman KA, et al. Lyme disease testing by large commercial laboratories in the United States. Clin Infect Dis. 2014;59(5):676–81.

  2. 2.

    CDC. Accessed 26 Nov 2014

  3. 3.

    Khatchikian CE, Prusinski M, Stone M, Backenson PB, Wang I-N, Levy MZ, et al. Geographical and environmental factors driving the increase in the Lyme disease vector Ixodes scapularis. Ecosphere. 2012;3(10):art85.

  4. 4.

    Kelly RR, Gaines D, Gilliam WF, Brinkerhoff RJ. Population genetic structure of the Lyme disease vector Ixodes scapularis at an apparent spatial expansion front. Infect Genet Evol. 2014;27:543–50.

  5. 5.

    Khatchikian CE, Prusinski MA, Stone M, Backenson PB, Wang I-N, Foley E, et al. Recent and rapid population growth and range expansion of the Lyme disease tick vector, Ixodes scapularis, in North America. Evolution. 2015;69(7):1678–89.

  6. 6.

    Brisson D, Dykhuizen DE. OspC diversity in Borrelia burgdorferi: different hosts are different niches. Genetics. 2004;168(2):713–22.

  7. 7.

    Vuong HB, Canham CD, Fonseca DM, Brisson D, Morin PJ, Smouse PE, et al. Occurrence and transmission efficiencies of Borrelia burgdorferi ospC types in avian and mammalian wildlife. Infect Genet Evol. 2014;27:594–600.

  8. 8.

    Hanincova K, Ogden NH, Diuk-Wasser M, Pappas CJ, Iyer R, Fish D, et al. Fitness variation of Borrelia burgdorferi sensu stricto strains in mice. Appl Environ Microbiol. 2008;74(1):153–7.

  9. 9.

    Wilske B, Busch U, Eiffert H, Fingerle V, Pfister H-W, Rössler D, et al. Diversity of OspA and OspC among cerebrospinal fluid isolates of Borrelia burgdorferi sensu lato from patients with neuroborreliosis in Germany. Med Microbiol Immunol. 1996;184(4):195–201.

  10. 10.

    Schwan TG, Piesman J, Golde WT, Dolan MC, Rosa PA. Induction of an outer surface protein on Borrelia burgdorferi during tick feeding. Proc Natl Acad Sci U S A. 1995;92(7):2909–13.

  11. 11.

    Önder Ö, Humphrey PT, McOmber B, Korobova F, Francella N, Greenbaum DC, et al. OspC is potent plasminogen receptor on surface of Borrelia burgdorferi. J Biol Chem. 2012;287(20):16860–8.

  12. 12.

    Qiu WG, Bosler EM, Campbell JR, Ugine GD, Wang IN, Luft BJ, et al. A population genetic study of Borrelia burgdorferi sensu stricto from eastern Long Island, New York, suggested frequency-dependent selection, gene flow and host adaptation. Hereditas. 1997;127(3):203–16.

  13. 13.

    Bunikis J, Garpmo U, Tsao J, Berglund J, Fish D, Barbour AG. Sequence typing reveals extensive strain diversity of the Lyme borreliosis agents Borrelia burgdorferi in North America and Borrelia afzelii in Europe. Microbiology. 2004;150(Pt 6):1741–55.

  14. 14.

    Dykhuizen DE, Brisson D, Sandigursky S, Wormser GP, Nowakowski J, Nadelman RB, et al. The propensity of different Borrelia burgdorferi sensu stricto genotypes to cause disseminated infections in humans. Am J Trop Med Hyg. 2008;78(5):806–10.

  15. 15.

    Brisson D, Vandermause MF, Meece JK, Reed KD, Dykhuizen DE. Evolution of northeastern and midwestern Borrelia burgdorferi, United States. Emerging Infect Dis. 2010;16(6):911–7.

  16. 16.

    Nadelman RB, Hanincová K, Mukherjee P, Liveris D, Nowakowski J, McKenna D, et al. Differentiation of reinfection from relapse in recurrent Lyme disease. N Engl J Med. 2012;367(20):1883–90.

  17. 17.

    Khatchikian CE, Nadelman RB, Nowakowski J, Schwartz I, Wormser GP, Brisson D. Evidence for strain-specific immunity in patients treated for early Lyme disease. Infect Immun. 2014;82(4):1408–13.

  18. 18.

    Wormser GP, Brisson D, Liveris D, Hanincova K, Sandigursky S, Nowakowski J, et al. Borrelia burgdorferi genotype predicts the capacity for hematogenous dissemination during early Lyme disease. J Infect Dis. 2008;198(9):1358–64.

  19. 19.

    Wang IN, Dykhuizen DE, Qiu W, Dunn JJ, Bosler EM, Luft BJ. Genetic diversity of ospC in a local population of Borrelia burgdorferi sensu stricto. Genetics. 1999;151(1):15–30.

  20. 20.

    Phillips CB, Liang MH, Sangha O, Wright EA, Fossel AH, Lew RA, et al. Lyme disease and preventive behaviors in residents of Nantucket Island, Massachusetts. Am J Prev Med. 2001;20(3):219–24.

  21. 21.

    Francis TJ. On the doctrine of original antigenic sin. P Am Philos Soc. 1960;104(6):572–8.

  22. 22.

    Stone BL, Russart NM, Gaultney RA, Floden AM, Vaughan JA, Brissette CA. The western progression of Lyme disease: infectious and nonclonal Borrelia burgdorferi sensu lato populations in Grand Forks County, North Dakota. Appl Environ Microbiol. 2015;81(1):48–58.

  23. 23.

    Morshed M, Scott J, Fernando K, Geddes G, McNabb A, Mak S, et al. Distribution and characterization of Borrelia burgdorferi isolates from Ixodes scapularis and presence in mammalian hosts in Ontario, Canada. J Med Entomol. 2006;43(4):762–73.

  24. 24.

    Hanincova K, Mukherjee P, Ogden NH, Margos G, Wormser GP, Reed KD, et al. Multilocus sequence typing of Borrelia burgdorferi suggests existence of lineages with differential pathogenic properties in humans. PLoS ONE. 2013;8(9), e73066.

  25. 25.

    Strandh M, Råberg L. Within-host competition between Borrelia afzelii ospC strains in wild hosts as revealed by massively parallel amplicon sequencing. Philos T Roy Soc B. 2015;370(1675):20140293.

  26. 26.

    Nadelman RB, Wormser GP. Reinfection in patients with Lyme disease. Clin Infect Dis. 2007;45(8):1032–8.

  27. 27.

    Probert WS, Crawford M, Cadiz RB, LeFebvre RB. Immunization with outer surface protein (Osp) A, but not OspC, provides cross-protection of mice challenged with North American isolates of Borrelia burgdorferi. J Infect Dis. 1997;175(2):400–5.

  28. 28.

    Bockenstedt LK, Hodzic E, Feng S, Bourrel KW, de Silva A, Montgomery RR, et al. Borrelia burgdorferi strain-specific Osp C-mediated immunity in mice. Infect Immun. 1997;65(11):4661–7.

  29. 29.

    Humphrey PT, Caporale DA, Brisson D. Uncoordinated phylogeography of Borrelia burgdorferi and its tick vector. Ixodes scapularis Evolution. 2010;64(9):2653–63.

Download references


This work was supported in part by grants from the NIH (AR41511, AI45801, AI076342, and AI097137) and the Burroughs Wellcome Fund. The funding agencies had no role in the study design; collection, analysis, and interpretation of the data; writing the report; nor the decision to submit the report for publication.

Author information

Correspondence to Camilo E. Khatchikian or Gary P. Wormser.

Additional information

Competing interest

Robert B. Nadelman has received consulting fees from Guidepoint Global and Decision Resources and has served as an expert witness in medical malpractice cases regarding Lyme disease. Disclosures: Dr. Wormser reports receiving research grants from Immunetics, Inc., Institute for Systems Biology, Rarecyte, Inc., and bioMérieux SA. He owns equity in Abbott; has been an expert witness in malpractice cases involving Lyme disease; is an unpaid board member of the American Lyme Disease Foundation; and was a consultant to Baxter for Lyme disease vaccine development.

Authors’ contributions

GW conceived the study. RN, JN, IS, and GW participated in study design, coordination and data collection. CK, ML, DB, GW conducted statistical analysis. CK, DB, GW drafted the manuscript. All authors edited the manuscript and approved the final version.

Additional files

Additional file 1: Figure S1.

The proportion of cases occurring in patients with a prior infection and the proportion of averted cases increases exponentially with higher incidence rates in both the equilibrium dynamic (left column, panels A, C) and individual stochastic models (right column, panels B, D). The proportion of reinfections that are averted due to strain-specific immunity (bottom row, panels E, F) is constant across incidence rates in both models. The dashed lines describe the data output when strain-specific immunity is assumed to last 5 years; the black lines describe the data output when strain-specific immunity is assumed to last 30 years. (DOC 311 kb)

Additional file 2: Table S1.

Frequency of the different OspC types that were cultured from the skin of 200 patients with erythema migrans (17, 18) and expected percentage increase in total reinfections due to particular OspC types, if there were no strain specific immunity based on the deterministic probability model. Bold font indicates invasive OspC types, comprising 76.5 % of the total cases. (DOC 35 kb)

Rights and permissions

Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (, which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver ( applies to the data made available in this article, unless otherwise stated.

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Khatchikian, C.E., Nadelman, R.B., Nowakowski, J. et al. Public health impact of strain specific immunity to Borrelia burgdorferi. BMC Infect Dis 15, 472 (2015).

Download citation


  • Immunity
  • Epidemiological impact
  • Lyme disease
  • Borrelia burgdorferi