Two surface sampling techniques were used. Rodac plates (65-mm tryptic soy agar [TSA] plates; Becton-Dickinson, Franklin Lakes, NJ) were pressed onto the surface being sampled. Immediately adjacent to the Rodac sampling site, swab sampling was performed with sterile rayon-tipped swabs moistened with a 0.5-mL solution of phosphate-buffered saline + 0.05% Tween 20 (PBS Tween). An approximately 100-cm² area was swabbed with sequential vertical, horizontal, and diagonal strokes. The swabs were individually placed in sterile, dry 15-mL conical tubes. Sampling focused on areas in the machine (electrical components, beneath belts, etc.) that were unlikely to have been cleaned with the topical bleach application.

The ventilation system in the mail facility was turned off when the facility closed, and the system remained off during testing. We were unable to simulate the “blow-down” procedure (used to clean the sorter) during testing because the air compressors for the air hoses had lost power. Machine operators typically use high-pressure hoses several times a day to clean accumulated dust and debris between mail-sorting runs. Two banks of 10-slit samplers were placed on two postal trolleys (approximately 5 feet above the floor) and connected to a vacuum pump. The slit sampler intake ports were approximately 10 inches above the trolley. Each of the samplers was loaded with 150-mm TSA plates. Slit sampler set “A” (SSSA) was placed next to the operator’s station (at a location and height where workers would spend most of their time), and slit sampler set “B” (SSSB) was placed at the opposite end of the sorter (Figure 1). To measure the temporal patterns of re-aerosolization, the samplers operated sequentially; the intake port of slit sampler no.1 was opened, allowed to run for either 1 min (SSSA) or 2 min (SSSB), and then closed. Then slit sampler no.2 in the set was activated, and so on, until all 10 slit samplers in the set had been sequentially activated. The rate of air flow through each of the slit samplers was 33 L/min.

SSSB was activated and ran for 20 min while the sorter was turned off. SSSA was then activated and ran for 10 min. The plates were removed from all slit samplers and new plates were loaded. The SSSA and SSSB design characteristics determined the duration of sampling.

Both sets of slit samplers were activated simultaneously while the sorter was inactive. Approximately 1 min later, the sorter was started, and clean dummy mail was processed. After several false starts, continuous operation was achieved in approximately 2 min. The operation of the sorter was interrupted several times by jammed envelopes and quickly restarted each time, until the machine was turned off 1 min before the end of the 20-min sampling period. Postal officials reported that false starts, jamming, and restarting are common during routine operation of the machine.

As with the previous sampling, SSSA ran for 10 min and SSSB ran for 20 min. The plates were removed from the slit samplers and sealed in plastic bags. The bagged plates were taken out of the facility and the exteriors of the bags were decontaminated with 0.5% hypochlorite solution.

The sampling team was outfitted with Canadian military C4 respirators with C7 canisters. The mask was equipped with 37-mm glass fiber collection filters mounted on the inlet port of the C7 canister, so that the entire inspirational volume of the investigators was sampled. The masked team members were located near the sorter to provide additional point sampling of respirable aerosol during the experiment. Team members were stationed at different work sites along the sorter and elsewhere in the facility to serve as point detectors. All mask filters were worn for at least 2 h.

Environmental swabs and TSA plates from the Rodac plates and slit samplers were stored at 4°C until shipped and processed. All specimens were shipped at room temperature overnight to the Centers for Disease Control and Prevention. Swabs were placed in 1.5 mL PBS-Tween and vortexed for approximately 1 min. The solution was heat-shocked at 65°C for 30 min, and 100 µL was plated onto a sheep blood agar (SBA) plate. Rodac and slit sampler plates were incubated for approximately 12 h, and CFU were counted by visual inspection. All colonies suspected to be Bacillus spp. were subcultured on SBA plates. Identification and confirmatory testing of B. anthracis were done according to standard microbiologic procedures (5). The mask filters were removed in the facility and placed in sterile glass tubes. After transport, they were suspended in 3 mL of heart infusion broth and incubated at 35°C for 36 h, after which 10 µL of broth was plated onto SBA plates. The filters and remaining broth were heat-shocked at 65°C for 30 min, and 10 µL was plated onto SBA plates.

Numbers of colonies detected in air sampling before and after the machine was activated were assessed with a one-sided one-sample test for difference in rates from a binomial distribution by using StatExact 4 v. 4.0.1. (Cytel Software Corp., Cambridge, MA).

Surface sampling was done by two methods, Rodac plates and premoistened swabs, to establish that the machine was still contaminated with viable B. anthracis spores. Ten Rodac plate samples and 10 swab samples were taken on the sorter surfaces. Both the Rodac plates and swabs yielded growth of CFUs that were too numerous to count at 7 of the 10 sites. Two additional Rodac plates were positive with low levels of contamination (1 and 3 CFUs); these locations were negative by the swab method (Figure 1).

A single colony of B. anthracis was identified on one of the SSSA plates (Figure 2) during 10 min of sampling before the sorter was activated. No B. anthracis was identified on any of the SSSB plates during the 20 min of sampling.

After the sorter was activated, SSSA and SSSB ran for 10 min and 20 min, respectively. A single colony of B. anthracis was identified during minute 10 of sampling by SSSA (Figure 2; Table). SSSB identified a single colony of B. anthracis at each of minutes 1, 5, 7, 8, and 19.

Five investigators wore mask filters while the sorting machine was inactive; four investigators changed to new mask filters while the sorting machine was active. All mask filters were negative by culture.