Use of Interactive Voice Response technology in collecting community data: Lessons from LiveWell Colorado

BACKGROUND: Limited information exists on the use of interactive voice response (IVR) technology for population-based health promotion surveillance and research. LiveWell Colorado, a statewide collaboration to ensure all Coloradoans enjoy a lifestyle that includes healthy eating and active living, funded 17 communities to develop and implement multi-sectoral initiatives. We designed an IVR survey to evaluate community-level change resulting from these initiatives. PURPOSE: To describe IVR survey respondents and to discuss the advantages and limitations of using IVR technology for collecting community-level health data. METHODS: We administered a 17-30 item IVR survey in 17 communities from 2006-2008. Seventeen validated core survey questions regarding physical activity, fruit and vegetable consumption, health profiles, and demographics were asked in all communities and additional strategy specific questions were added in some communities. IVR technology contacted a randomly selected sample of landline telephone numbers in each community. Consenting participants were entered into a gift card drawing as incentive to participate. Unweighted descriptive statistics were used to describe the demographic characteristics of IVR survey participants, and response rates were calculated for each survey. RESULTS: 11,017 Coloradoans participated. 81% of respondents were female, 11% were Hispanic, and 63% were White. Response rates in each community were low, ranging from 8.23% to 21.07% in 2008. IVR technology was an efficient method to gather a large number of survey responses in a short period of time, and with fewer financial and personnel resources as compared to human-administered phone surveys. Limitations included non-response bias and the survey was not tailored to meet the sociodemographic diversity of all Coloradoans. CONCLUSION: IVR is a novel and cost-effective approach to collecting community-level data on healthy eating and active living behaviors. However, analysis of respondent characteristics indicates a need to address sampling strategies, survey design and recruitment to reduce non-response bias.