Computer-assisted Telephone Interview Techniques

To the Editor: Fox et al. used computer-assisted telephone interview (CATI) techniques in an outbreak of cryptosporidiosis (1). Australian health agencies have used CATI for several years. A case-control study during an outbreak of Salmonella Mbandaka in 1996 employed CATI to interview 15 casepatients and 45 controls; contaminated peanut butter was implicated (2). Foodborne disease outbreaks are often geographically widespread and suited to using CATI. Australian health authorities investigate ≈100 outbreaks of foodborne disease each year, with 3–4 using CATI-based case-control studies. Some jurisdictions investigate outbreaks by using CATI interviews of controls sampled from a bank of potential study participants (3). Potential study participants are recruited at the conclusion of rolling risk factor survey interviews, similar to the Behavioral Risk Factor Surveillance System. A “control bank” allows investigators to rapidly obtain contact details for appropriately matched controls because age and sex of all household members are recorded in a database. Using control banks with CATI allows completion of studies quicker than CATI or traditional methods alone (4). South Australia has used CATI during 11 case-control studies of salmonellosis, legionellosis, Q fever, campylobacteriosis, Shiga toxin–producing Escherichia coli, and cryptosporidiosis (http://www.dh.sa.gov.au/pehs/ notifiable-diseases-summary/currentoutbreak-table.htm). During an Australian CATI survey about gastroenteritis, 5,123 (84%) of 6,087 households agreed to be in a control bank (5). This bank of 14,024 potential controls was used in 4 casecontrol studies of sporadic salmonellosis and campylobacteriosis. This system avoided randomly dialing thousands of households to enroll controls in young age groups. The control bank was used for 3 years after initial collection, although many jurisdictions update banks annually. Investigators may find CATI useful, although it can be costly and introduce biases (4). Programming questionnaires can delay investigations, which makes paper-based collection better in small outbreaks (4). CATI cannot be used in areas where a small proportion of the population has telephones. Despite limitations, CATI, when combined with control banks, may improve outbreak investigations.