In October and November 2010, water and seafood samples were collected from 2 cholera-affected communities in Haiti and tested for V. cholerae. Eight freshwater and 6 marine water samples were collected from 13 sites in Artibonite and Ouest Departments (Figure). Freshwater samples were collected from rivers, including the Artibonite River, and irrigation canals. Dead-end ultrafiltration, a newly developed technique that has been used to recover diverse microbes from large-volume water samples (5), was used to collect water samples (8–30 L each) at the freshwater sites and 3 of the marine water sites (HWS-11, -13, and -18). At the Haiti National Public Health Laboratory (LNSP), bacteria were recovered from ultrafilters by back flushing with a surfactant solution, and the solution then was added to an equal volume of 2× strength alkaline peptone water (APW). Grab samples (1 L) were collected at 3 harbor sites (HWS-15, -16, and -17) and on arrival at LNSP, they were split into two 500-mL portions for separate testing by CDC at LNSP or for shipment in chilled coolers to FDA (Dauphin Island, AL, USA). At LNSP, all water and ultrafilter back flush samples were incubated in APW at 37°C for 6 h (6). After APW enrichment, the culture broth was streaked onto thiosulfate citrate bile salts (TCBS) agar (Remel, Lenexa, KS, USA) and incubated overnight at 37°C. For each sample, up to 10 colonies suspected of being positive were picked from TCBS agar and grown on nonselective media for multiplex PCR testing (7).

Nine seafood samples were collected along the coast: 5 between Saint-Marc port and Grand Saline (1 site) and 4 from Port-au-Prince port (3 sites) (Figure). Seafood samples were obtained as convenience samples from local fishermen on the water, placed in Ziploc (SC Johnson, Racine, WI, USA) bags, and sent in chilled coolers to FDA, along with 500-mL grab samples of marine water. After enrichment, APW cultures from seafood and water grab samples were tested at FDA by using a real-time PCR specific to the cholera toxin gene (ctxA) of V. cholerae (8). If the APW culture was positive for the ctxA gene, then isolates were obtained by streaking onto TCBS agar as described for water samples.

Identification and characterization of suspected V. cholerae isolates were performed at CDC. Colonies or sweeps of growth typical of V. cholerae were selected from TCBS plates and tested by multiplex PCR for ctxA, tcpA_{El Tor}, tcpA_Classical, ompW, and toxR genes (9–11). Colonies positive by PCR for ctxA or other V. cholerae markers were tested for agglutination in serogroup O1 antiserum and, if positive, also in Inaba and Ogawa serotype antisera. V. cholerae isolates were subtyped according to the PulseNet standardized pulsed-field gel electrophoresis (PFGE) protocol, using primary and secondary restriction enzymes SfiI and NotI, respectively (12). The cholera toxin gene, ctxAB, and tcpA gene regions were amplified by PCR and sequenced (13,14).

V. cholerae O1, serotype Ogawa, ctxA-positive strains were isolated from 2 irrigation canals north of Port-au-Prince in Ouest Department (Table 1). Both of these canals were used for drinking water by the local population, and communities near the canals were heavily affected by the outbreak. V. cholerae O1 Ogawa ctxA-positive strains were isolated from 1 mixed seafood sample (sample 7, containing multiple vertebrate fish and 1 crab) and 1 bivalve sample (sample 8, containing multiple species) that were obtained from fishermen at 2 different locations in the Port-au-Prince port (Table 2). All ctxA-positive V. cholerae isolates were indistinguishable from the outbreak strain by PFGE with both enzymes (pattern combination KZGS12.0088/KZGN11.0092) (15). Sequence analysis for the toxigenic V. cholerae isolates provided additional evidence that the isolates from these samples matched the isolates from humans infected with the outbreak strains. The tcpA sequence of the freshwater and human isolates from Haiti matched that of CIRS 101, an altered El Tor strain from Bangladesh, and the ctxAB sequences matched the sequences from strains isolated in 2007 during an outbreak in Orissa, India (15). The ctxAB and tcpA sequences differed by 1 nt polymorphism from prototypical classical and El Tor alleles, respectively. These isolates were recovered from 30-L freshwater samples having turbidities of 11 and 16 nephelometric turbidity units, which were among those with the lowest turbidity collected during this investigation. All V. cholerae non-O1 ctxA-negative strains possessed unique PFGE patterns distinct from the outbreak pattern. In addition to samples from which toxigenic V. cholerae was isolated, real-time PCR testing by FDA detected the ctxA gene in APW culture broths for 3 seawater samples and 3 other seafood samples.

Isolation of the outbreak strain in seafood samples from Port-au-Prince and detection of the ctxA gene in APW cultures of water and seafood samples from Port-au-Prince and Saint-Marc suggest that harbor waters were contaminated with toxigenic V. cholerae O1. This finding underscores the need for adherence to public health recommendations disseminated during the outbreak regarding making drinking water safe and cooking seafood thoroughly to prevent infection and conducting ship ballast water exchange to limit potential transfer of the organism to other harbors. We report recovery of V. cholerae O1 from large-volume water samples by use of ultrafiltration. Although V. cholerae O1 was not isolated from marine water samples, real-time PCR detection of ctxA in these samples provided additional evidence that harbor water samples were contaminated with toxigenic V. cholerae. Use of this real-time PCR method has provided analytical data that reflected the presence of viable V. cholerae in marine water samples (8). Further assessment by using high-volume filtration and seafood sampling may be useful for tracking the persistence of the strain in the Haitian environment in the future.