Acanthamoeba keratitis (AK), a painful corneal infection that may lead to vision loss or enucleation, is caused by the ubiquitous free-living Acanthamoeba spp. ( 1 – 4 ). AK occurs primarily among users of soft contact lenses ( 5 ), with an estimated US annual incidence of 1–2 cases per million contact lens users ( 6 ). In May 2006, the Centers for Disease Control and Prevention (CDC) was contacted by the Illinois Department of Public Health regarding a possible increase in AK cases in the Chicago area during the preceding 2 years. Investigators at the University of Illinois at Chicago were conducting a case–control study to identify possible risk factors. In October 2006, CDC informally surveyed ophthalmologists across the country to ascertain whether cases of AK were increasing elsewhere; results were inconclusive. In January 2007, CDC initiated a retrospective survey of 22 ophthalmology centers nationwide. By early March 2007, results obtained from 10 centers in 9 states showed a rise in the number of culture-confirmed cases during 2004–2006 compared with 1999–2003. On March 16, 2007, we initiated a national outbreak investigation. On May 23, a preliminary analysis compared data from the first 46 interviews of patients with culture-confirmed AK, with data obtained from 126 healthy adult contact lens users ascertained in a 2006 national outbreak investigation of Fusarium keratitis ( 7 ). The analysis indicated that the odds of having ever used Advanced Medical Optics Complete MoisturePlus (AMOCMP) multipurpose contact lens solution were 20× greater for AK case-patients than for controls. These results were communicated to the Food and Drug Administration (FDA) and were rapidly disseminated ( 8 ). On May 26, 2007, the company voluntarily recalled AMOCMP from domestic and international markets. Although public health action was taken on the basis of the preliminary analysis, we report here the results of a matched case–control study designed to verify the findings of the preliminary analysis, to identify additional risk factors for AK, and to guide recommendations to prevent future cases. Methods Case Definition and Case Finding Case-patients had been given a diagnosis of AK by an ophthalmologist; had symptom onset on or after January 1, 2005; and had Acanthamoeba spp. identified from cultures of corneal specimens. Requests to report AK cases were disseminated through CDC’s Epidemic Information Exchange system and through ophthalmology and optometry electronic mailing lists; websites; and associations at the national, state, and local levels. We also queried several referral microbiology laboratories and ophthalmology centers to find cases. Cases included in a concurrent study by University of Illinois at Chicago investigators were excluded ( 9 ). Case-Patient Data Collection and Laboratory Investigation We used standardized questionnaires to interview case-patients by telephone to obtain demographic characteristics, information regarding illness, contact lens–related product use, and hygiene practices and behavior during the month before symptom onset. An Internet-based visual aid was available to assist with specific product recognition. Ophthalmologists who were treating case-patients provided information by telephone- or self-administered questionnaires regarding diagnostic methods, treatment, and clinical outcomes. Available clinical specimens (e.g., corneal scrapings or biopsy specimens, Acanthamoeba culture isolates) and environmental samples (e.g., opened and unopened contact lens solution bottles, lenses, lens cases) were sent to CDC laboratories. Specimens were processed for Acanthamoeba spp. by culture ( 4 ) and molecular analysis ( 10 ), including genotyping ( 11 ). Case–Control Study All interviewed case-patients were eligible. Control subjects had no history of AK and were >12 years of age. We attempted to match 3 controls to each case-patient by contact lens use (i.e., soft lenses, rigid lenses, or no contact lens use) and by geographic location using a reverse address directory to identify controls who resided near each case-patient. Because rigid lens use is uncommon, we did not attempt to obtain geographically matched controls for this group. Controls completed a standardized, telephone-administered questionnaire that asked about behavior and product use during the 1 month before their matched case-patient had symptom onset. Data Analysis Data were double-entered by using Visual FoxPro 8.0 (Microsoft Corp., Redmond, WA, USA) and analyzed by using SAS 9.1 (SAS Institute Inc., Cary, NC, USA). Conditional logistic regression was used to estimate odds ratios (ORs) and 95% confidence intervals (CIs) for univariate and multivariate analyses; significance was defined as p 50 22 (21) Contact lens wear Did not use contact lenses 12 (11) Used contact lenses 93 (89) Lens type Soft lenses 82 (88) Rigid lenses 10 (11) Hybrid lenses 1 (1) Contact lens solution use among contact users† Did not use cleaning or disinfecting contact lens
solution 6 (6) Used any type of cleaning or disinfecting contact
lens solution 87 (94) Type of solution used§ Multipurpose solution 78 (90) Hydrogen peroxide solution 6 (7) Daily cleaner 11 (13) Affected eye Right 53 (50) Left 44 (42) Both 8 (8) Symptoms when treatment sought‡§ Pain 78 (74) Redness 78 (74) Sensitivity to light 76 (72) Sensation of foreign body 71 (68) Increased tearing 59 (56) Blurred vision 57 (54) Discharge from eye 20 (19) Clinical status¶ Resolved with pharmacologic therapy 32 (38) Currently receiving pharmacologic therapy 29 (34) Corneal transplant performed 21 (25) Corneal transplant planned 3 (4) Most recent visual acuity with best correction in affected eye# 20/20 17 (24) 20/25–20/100 24 (34) 20/>200 29 (41) *Median 29 y, range 12–76 y.
†During 1 month before illness onset; n = 93.
‡Not mutually exclusive.
§Median time from symptom onset to anti-Acanthamoeba therapy (n = 80) was 49 d (range 4–197 d).
¶At the time of treating ophthalmologist interview; n = 85.
#At the time of treating ophthalmologist interview; n = 70. The most frequently reported symptoms among case-patients were pain, redness, sensitivity to light, and sensation of a foreign body (Table 1). The median time from onset of symptoms to initiation of anti-Acanthamoeba treatment was 49 days, range 4–197 days. At the time of their ophthalmologist interview, 24 (28%) of 85 had either undergone or were awaiting a corneal transplant, and 29 (41%) of 70 had a visual acuity of 20/200 or worse with best correction (i.e., legally blind) in the affected eye. Case–Control Study During June 14–July 10, >11,000 phone calls were made to obtain 184 controls matched to 91 case-patients; case-patients with no matched controls were excluded from subsequent analyses (Figure 3). Because of differences in possible exposures (primarily the use and type of contact lens solutions) between soft lens, rigid lens, and non–contact lens users, we further restricted the analysis to case-patients (n = 72) and controls (n = 140) who reported wearing soft contact lenses only. Separate analyses were performed among users of rigid contact lenses and non–contact lens users; however, sample sizes were small, and no associations were found. Users of soft contact lenses who were excluded for lack of a matched control were not significantly different from those included in the analysis with respect to age, sex, race, and ethnicity. Figure 3 Matching of case-patients with Acanthamoeba keratitis and controls, United States, 2005–2007. Matched univariate analysis of users of soft contact lenses (Appendix Table) indicated that any use of AMOCMP within the month before symptom onset was a substantial risk factor (OR 15.8, 95% CI 5.6–44.6). No other contact lens solutions were associated with disease. Variables in univariate analyses that were included in the multivariate modeling included the following: any use of AMOCMP, Hispanic ethnicity, age (12–17, 18–24, 25–34, 35–49, versus >50 years), male sex, history of ocular trauma, contact lens use 30 years (A. Stark, City of Chicago Department of Water Management, pers. comm.). A preliminary analysis (J. Verani, unpub. data) conducted during the early phase of this investigation found that only 12 (29%) of 41 case-patients for whom water treatment data were available received household water from chloraminated systems during the month before symptom onset, compared with an estimated 32% of the general US population ( 34 , 35 ). These findings suggested that water disinfection type was not an important risk factor in this outbreak. This study had several limitations. First, because AK culture is a highly specific, but insensitive, diagnostic tool ( 36 ), and because preferred diagnostic methods vary by medical center, inclusion of only patients with culture-confirmed cases may have introduced regional testing bias and underestimated the scope of the outbreak. Second, the response rate among persons approached for control interviews was low; therefore, demographic differences between cases and controls may have been due to selection bias among controls. Third, recall bias may have been introduced as we asked participants to report on contact lens product use and behavior during the previous 2 years. Fourth, misclassification bias may have been introduced because at least 2 case-patients appeared to not differentiate between use of saline and cleaning or disinfecting solutions. Fifth, because >40% of case-patients and all controls were interviewed after AMOCMP was recalled in May 2007, reporting bias may have been introduced. Despite these limitations, among users of soft contact lenses, case-patients had almost 17 times the odds of reporting any AMOCMP use compared with matched controls, validating the results of the preliminary analysis comparing AK cases to Fusarium keratitis investigation controls ( 8 ). The use of this existing Fusarium comparison data enabled rapid public health action months before the case–control study was completed. Recent associations of 2 distinct multipurpose solutions with outbreaks of rare corneal infections highlight the need for improved surveillance to promptly detect contact lens–related outbreaks and raise concerns about the effectiveness of multipurpose solutions. Continued monitoring of AK case trends to assess the impact of the AMOCMP recall and research on the anti-Acanthamoeba efficacy of AMOCMP and other solutions are under way. Our findings highlight the importance of promoting healthy habits among contact lens users, particularly discouraging the practice of topping off solutions and reinforcing safe hygienic practices among new users of contact lenses, as well as the need for standardized anti-Acanthamoeba testing of contact lens solutions. Supplementary Material Appendix Table Univariate analysis of demographic characteristics and exposures among 72 Acanthamoeba keratitis case-patients and 140 controls who used soft CLs, United States, 2005-2007*