A variety of professional staff screened calls coming into a central telephone bank. A prerecorded message instructed callers to contact their state or local health department if they had not done so. Calls that could be answered with “Frequently Asked Questions” documents or guidelines published in the Morbidity and Mortality Weekly Report (MMWR) were handled directly by public health advisors, epidemiologists, and junior staff (5–8). Callers were referred to the SLT for follow-up if they reported symptoms consistent with anthrax or other bioterrorism agents, noted exposure to a suspicious package or substance, or required detailed medical expertise. SLT staff included a team of public health advisors to obtain initial case information, and at least two physicians, epidemiologists, or veterinarians. The SLT had an average of nine staff members (range 2–15) that was reduced in evenings and on weekends depending on the volume of calls being received. Whenever appropriate, calls were also referred from the SLT to state epidemiologists for more detailed follow-up.

Two Access databases (Microsoft Corp., Redmond, WA) were created to assist in documenting and tracking all incoming calls. A general database was intended to document every incoming call to the EOC telephone bank. For all incoming calls—call volume permitting—central telephone bank staff were instructed to record information on the date, topic, and type of caller on call response forms. SLT staff regularly collected these forms for manual data entry. Reports of call volume, call type, and call topic by day were then shared with EOC management and communications personnel to assist them with staffing decisions, publication of MMWR reports, and determination of educational needs.

Calls referred from the central telephone bank to the SLT were manually entered into a second, more detailed SLT tracking database. Information collected in this tracking database included demographic background of the patient, reporter information, and any reported symptoms or exposures. SLT staff were also asked to assign each referred call to a risk category to prioritize follow-up within the large volume of calls.

Telephone call data were exported from Access databases into Statistical Analysis Software (SAS Institute, Inc., Cary, NC). Distributions of call volume by date of call, type of caller, and topic of call were produced from the central EOC telephone bank data. Descriptive analyses of SLT tracking data were undertaken by type of caller, state of reported occurrence, triage classification (level of urgency), reported signs and symptoms, and nature of reported exposure. Data were analyzed during the peak period of call volume during the anthrax investigations (October 8 to November 11, 2001).

From October 8 to November 11, 2001, the EOC received 11,063 telephone calls pertaining to bioterrorism and referred 882 of these calls to the SLT for diagnostic evaluation, consultation, and coordination of follow-up activities. The volume of calls received during this time period demonstrated a considerable public need for guidance during this emergency.

Highly publicized incidents such as the opening of the letter in Sen. Daschle’s office were likely catalysts for the observed increases in call volume. However, day-to-day patterns in the call volume to the EOC telephone bank suggest that at predictable times during the week emergency staff resources can be relaxed. During the data collection period, the mean call volume to the EOC was 80% lower on weekends than on weekdays. Within the working week, mean call volumes were 23% lower on Mondays and Fridays than during the rest of the work week. As many staff worked 12–20 hour days during the height of this emergency, allowing staff to remain “on-call” at off-site locations on days of predictably lower call volume may help maintain staff morale and stamina through long periods of emergency center operations.

The EOC implemented a tiered telephone call triage system designed to allow highly suspicious cases and exposures to be tracked more closely by field epidemiology and specialty teams (Figure 1). Using scripted responses to frequently asked questions, this system effectively screened out many calls involving general queries about anthrax vaccines, requests for bioterrorism information, and the use of personal protective equipment. This approach allowed the SLT at the second tier of the triage system to spend more time interpreting clinically confirmed symptoms and laboratory results, and monitoring possible exposures for further referral to appropriate specialized teams.

These findings, however, also indicate that many calls received by the SLT did not pertain to known high-risk situations. During the data collection period, nearly 75% of calls referred to the SLT did not include a report of any clinically confirmed signs or symptoms or any direct contact with an environment known to be contaminated with B. anthracis. Of these calls, nearly 60% mentioned contact with a suspicious powder or package, but included no report of illness. As a result, to maintain specificity in tracking high-risk cases, scripted responses were developed to questions regarding 1) contact with unknown substances, 2) the receipt of mail through a facility that had been contaminated with B. anthracis, and 3) the report of clinically unconfirmed signs or symptoms (5,7). We recommend further refinement of these response protocols for inclusion in the first tier of the triage system, along with additional training of telephone bank staff in the overall objectives and methods of triage during bioterrorism emergencies. These measures would substantially reduce the call volume burden on second-tier staff and decrease the chance that a high-risk situation would be overlooked during a similar bioterrorism event.

State health departments typically expect that CDC will direct local calls back to them unless they have previously been referred to CDC (9). As more than half the calls to the EOC were from private citizens, a larger number of calls should also have been redirected from the EOC to appropriate contact persons at the state level (with minimal data entry and analysis by CDC). Such referrals would have allowed the EOC staff more time to respond to questions from physicians or health departments. The extent to which state and local health departments were satisfied with the assistance received from the EOC also remains unknown. A survey of state and local personnel who contacted the EOC system would assist CDC staff with quality improvement of the triage system and provide additional insight into the state perspective of appropriate respective roles during periods of emergency response.

These data have several limitations. An unknown number of calls to the EOC telephone bank were undocumented as first-tier staff were unable to complete all telephone call response forms during peak periods of call intensity. These high call volumes periodically resulted in delays in information transfer between tiers of the telephone call triaging system. In addition, the manual completion of telephone response forms resulted in a substantial amount of missing data, as first and second-tier EOC staff often overlooked key data elements in their efforts to provide timely responses to public demands. Several coding classification schemes on the telephone response forms also require revision. For example, we were unable retrospectively to determine the number of law enforcement or emergency medical service personnel who called the central phone bank or whether callers from state health departments were medical or public relations personnel.

Telephone-based hotlines underestimate the true number of cases of a disease and are dependent on media reports and general public interest (10). However, a telephone bank at CDC during an outbreak of hantavirus pulmonary syndrome identified 38% of confirmed cases (10). Computerization of the EOC triage system, including required fields for date and topic of call and type of caller would allow for timely transfer and analysis of complete and accurate telephone call data and perhaps provide a similar layer of passive surveillance for emerging bioterrorism events. However, the maintenance of such a system would require additional technical expertise in database development, management, and analysis (11). Medical expertise in first-tier telephone bank staff will continue to be needed to assure the accurate entry of data into any automated system.

Our findings suggest that available on-site staff resources can be adjusted to predictable daily patterns of call volume to increase long-term effectiveness and stamina during emergency periods. While the first tier of the EOC telephone call triage system effectively addressed a substantial portion of all incoming public inquiries during this emergency, standardized health communication protocols that address contact with suspicious substances, handling of suspicious mail, and the clinical evaluation of suspected cases in the absence and presence of confirmed exposure should also be added to first-tier response activities in a computerized triage system. This standardization would allow for a more effective triage system for inquiries and more efficient focus for follow-up investigations by specialized epidemiologic teams.