Respondent Selection

During August–September 2016, we performed a serologic investigation of farmworkers and household members in Cundinamarca Department. We selected farms based on respondent availability from a list of farms provided by the local secretary of health. After obtaining informed consent from adults and permission from parents of children <18 years of age, we administered a questionnaire regarding demographic characteristics, exposures, travel history, and farming practices. We also collected serum samples from interviewees to correlate risk factors with serologic evidence of VACV exposure. We received a total of 134 responses and corresponding specimens from persons on 56 separate farms.

Orthopoxvirus Antibody Detection

We used IgG ELISA to evaluate the presence of orthopoxvirus-specific antibodies (i.e., anti-OPXV) as previously described (33). We coated Immulon II High Binding microtiter plates (ThermoFisher Scientific, https://www.thermofisher.com) with purified VACV DryVax strain at 0.1µg/mL in carbonate buffer, incubated overnight at 4°C, inactivated with 10% formalin, and washed 3 times with PBST (PBS with 0.05% Tween-20) by using a BioTek 405TS plate washer (Biotek, https://www.biotek.com). We then blocked plates at room temperature for 30–60 min with assay diluent containing 5% dried skim milk, 2% normal goat serum, and 2% bovine serum albumin in PBST. After blocking, we washed plates 3 times with PBST, added serum samples at 1:100 dilution in duplicate, and incubated for 1 h at 37°C. We washed plates again and added goat anti-human IgG horseradish peroxidase conjugate (KPL antibodies) (SeraCare, https://www.seracare.com) at 1:2,000 concentration, incubated for 1 h at 37°C, and washed. We then added SureBlue TMB 1-component microwell peroxidase substrate (KPL antibodies) (SeraCare) and developed for 4 min at room temperature before stopping the reaction with addition of equal volume of TMB Stop Solution (SeraCare). We read optical density (OD) on an Enspire plate reader (Perkin Elmer, https://www.perkinelmer.com) at 450 nm.

For the IgM ELISA, we coated microtiter plates (Immulon II) with goat anti-human IgM KPL antibodies at 1:800 dilution in PBS (pH 7.4) and incubated overnight at 4°C. We then washed plates 5 times with PBST by using a plate washer and blocked for 30 min to 1 h at room temperature with assay diluent buffer containing 0.5% gelatin, 2% BSA, 5% skim milk, and 2% normal goat serum in PBST. We added test serum samples at 1:50 dilution in duplicates in assay diluent and incubated for 1 h at 37°C. We washed plates, added antigen (purified VACV) at a concentration of 0.5 µg/mL, and incubated for 1 h at 37°C. We washed plates again and incubated with 1:250 dilution of anti-variola virus hyperimmune mouse polyclonal ascetic fluid for 1 h at 37°C, followed by washing and incubation with 1:6,000 dilution of goat anti-mouse IgG horseradish peroxidase conjugate (KPL antibodies) for 1 h at 37°C. We then washed the plates again and developed with SureBlue TMB 1-component microwell peroxidase substrate for 8 min at room temperature, after which we added equal volume of TMB Stop Solution to each well. We read on an Enspire plate reader at 450 nm.

We averaged OD values for known negative controls and determined a cutoff value by using the equation cutoff value: average negatives + 3 × SD of negatives. We subtracted the cutoff value from the OD values of test samples. If the resulting value was >0.05, we considered the serum sample positive for the presence of OPXV antibodies.

Data Analyses

To identify risk factors associated with OPXV exposure, we performed a nested case–control analysis on the basis of serologic test results. We classified as case-patients those persons with a positive test for OPXV IgM or IgG, which is not specific for VACV but is a reasonable approximation of exposure (either through natural infection or vaccination), given a lack of other known circulating OPXVs in this region. Conversely, we identified as controls those persons without serologic evidence of OPXV exposure. To determine odds ratios (ORs) and 95% CIs, we performed a complex sample analysis to account for clustering of responses and serologic outcomes by farm. Variables found to have a p value <0.1 in bivariate analysis were included in a multivariable model analysis.

We also evaluated the correlation of farm-level characteristics with seropositivity of any persons associated with the farm. We performed bivariate analysis on individual risk factors to determine ORs and 95% CIs. Given that the last smallpox vaccination campaign occurred in Colombia in 1972, we separated persons on the basis of age of eligibility to have received the smallpox vaccine (34).

We subsequently built 2 multivariate logistic regression models by using individual-level risk factor data and farm survey data, respectively. We used seropositivity as the outcome variable. Using simple logistic regression, we included all variables found to be statistically significant at an α level of 0.1 in a stepwise model selection procedure. For individual-level risk factor data, we incorporated the variable that was most significant after being solely added to the model (if any were significant at an α level of 0.1) into the model. If any of the tested variables were no longer significant after this addition at an α level of 0.1, we dropped it from the model. This process continued until no variable was found to be significant, after each was solely added to the model. We checked variables for collinearity by using Pearson correlation coefficients; values <0.4 were considered to not be collinear.

We conducted a similar process with the farm survey data. In that case, we also forced into the model the variable indicating whether any animals with a history of vaccinia-like lesions were on the farm. We did this to evaluate the influence of suspected animal vaccinia virus infections on human seropositivity. Afterward, we conducted the same step-wise procedure.

Ethics Statement

Review by the Colombian Instituto Nacional de Salud and a human subjects advisor at the US Centers for Disease Control and Prevention determined that the activities did not meet the definition of research under 45 CFR 46.102(d). All adult participants provided informed written consent before interview participation and collection of specimens. Participants <18 years of age provided age-appropriate assent, and parents or guardians provided consent on their behalf.

Demographics and Descriptive Analysis

Commensurate with the agricultural setting that characterizes Cundinamarca Department, participants in the investigation tended to live in rural environments and had frequent contact with animals (Table 1). Approximately equal numbers of men and women were enrolled; median age was 46 years. Based on the given age threshold (44 years of age), slightly more than half of the participants (57%) would have been eligible to have received a smallpox vaccine before the end of the vaccination campaign. However, only 34% of participants recalled a history of smallpox vaccination.

Nearly all participants (96%) reported contact with cows, and most of these persons participated in the milking process (86%). Thirty-seven percent of participants reported working concurrently on multiple farms, and ≈21% of participants had previously heard of VACV or other poxviruses.

Laboratory analysis demonstrated that 70 (52%) of the 134 participants were OPXV IgG positive, including 2 (1.5%) persons who were also OPXV IgM positive, suggesting a recent exposure (<6 months before). None of the participants was only positive for IgM. Excluding those born in 1972 or earlier, seropositivity for OPXV IgM or IgG was found among 18 (31%) of 58 people included in this category (Appendix Figure 1).

Eighteen seropositive persons also reported a history of a vaccinia-like lesion, primarily occurring on the hand (94.4%), but 3 persons reported eye involvement (Table 2). Of these 18 persons, 12 (67%) were <44 years of age, making them ineligible to have received a smallpox vaccine. The risk for symptomatic vaccinia-like lesions was not statistically different between those who self-reported smallpox vaccination and those who did not recall a history of vaccination (OR 0.35, 95% CI 0.1–1.3). However, when we compared age of smallpox vaccine eligibility, being <44 years of age was strongly correlated with having a symptomatic vaccinia-like infection (OR 15.3, 95% CI 4.2–56.1).

Symptomatic persons experienced a median of 1.5 lesions, and lesions resulted in scarring in 13 of the 18 patients. Approximately one half of these persons took time off work because of their lesions, for a median of 10 days (range 3–15 days). Eleven people sought care from a physician, and 2 persons were hospitalized.

The lesions were most frequently characterized by localized pain and swelling, pruritus, and increased warmth (Table 2). Two thirds of patients also reported discharge from the lesion and lymphangitis. Many of the patients cited the co-occurrence of other symptoms including fever, malaise, and headache.

In the analysis of farm-level characteristics, 22 (39%) of the 56 farms reported animals with vaccinia-like lesions. Cows were the only domesticated animals noted to have vaccinia-like lesions, with the exception of 1 farm that also recalled pigs having similar lesions. The lesions were located on the udders or teats in all cases; 2 farms also reported oral lesions, and 1 farm reported genital lesions. Twenty (91%) of the 22 farms continued milking their cows in spite of the lesions. Outcomes of the lesions resulted in decreased milk production at 5 farms and caused scarring of the affected cows at 3 farms.

Bivariate Analysis

In the bivariate analysis of individual-level risk factors that we assessed, 13 variables were significantly associated with anti-OPXV seropositivity, including age as a continuous variable or as a dichotomous outcome based on eligibility for smallpox vaccination (Appendix Table 1). Age (dichotomous) and consumption of pork were the variables most strongly associated with seropositivity (OR 4.81 for age, OR 4.86 for pork consumption). Among the other significant factors were municipality of residence, self-reported history of smallpox vaccination, presence of smallpox vaccination scar, cows living on the property of residence, time spent working on the current farm, previous work on other farms, and consumption of unpasteurized milk or cheese.

In the evaluation of farm-level risk factors, 7 variables were associated with seropositivity among farmworkers at the 0.1 level (Appendix Table 2). These variables included animals with vaccinia-like lesions, type of cattle feed, habitats surrounding the farm, and humans on the farm with vaccinia-like lesions.

Multivariate Analysis

For the multivariate analysis of individual-level risk factors, 5 variables were included in the final model: age (dichotomous), smallpox vaccination scar, in-country travel in the previous 12 months, duration of time spent working on current farm, and municipality of residence. All variables were significant at the p<0.05 level, and none were found to be collinear. Age >44 years, presence of a vaccination scar, and longer duration of time working on the current farm were predictive of anti-OPXV seropositivity, whereas in-country travel and residence outside of Medina were protective (Table 3).

Farm-level risk factors in the final model included animals with a history of vaccinia-like lesions, use of commercial feed, and feeding cattle after milking. Variables were significant at the p<0.1 level. Animals having vaccinia-like lesions was predictive of anti-OPXV seropositivity of farmworkers, but the other 2 variables were noted to be protective (Table 3).