Internet surveys yielded efficient but not complete data collection on norovirus outbreak

Oh JY. Emerg Infect Dis. 2010;doi:10.3201/eid1611.100561.

More than 75% of telephone interviewees completed a survey regarding a norovirus outbreak in Oregon vs. less than 65% of Internet interviewees, according to study results.

The study was conducted in the wake of an outbreak of gastroenteritis that was initially self-reported by six participants in a 475-mile bicycle ride from Sept. 13 to 19, 2009. Five of the riders who reported illness to organizers of the event provided stool samples that tested positive for norovirus infection.

Officials from CDC and the Oregon Public Health Division administered 10- to 15-minute telephone and Internet surveys to determine which method was more effective in measuring response rates, attack rates and risk factors for illness. A survey defined as complete was defined as one that yielded an answer of “yes” to the final question, which was: “Did the participant become ill?”

Organizers provided contact information for all 2,273 registered riders, of whom 1,288 were Oregon residents.

Response rates

There were 204 participants randomly assigned to the Internet survey and 93 assigned to the telephone survey. The surveys contained 95 questions — 46 about food items eaten — and were identical in both groups.

Cases were defined as vomiting or more than three loose stools within 24 hours in a rider, and with an onset between Sept. 11 and 22, 2009.

Internet surveys were initiated by 76% of 201 participants selected for that approach; telephone surveys were initiated by 84% of 91 participants in that group. Among those who initiated, the survey was completed by 64% of Internet users and 79% of telephone users (P=.01 for difference in overall completion rate).

Within the Internet group, riders older than 50 years were more likely than riders younger than 50 years to complete the survey, 70% vs. 56%, (P<.05).

About 2 days were required to complete the survey by both groups.

More than 90% of the food item questions were answered by 57% of the Internet interviewees and 94% of the telephone interviewees.

Attack rates

A 23% attack rate was observed in a cohort of Internet survey participants who responded within 2 days of the survey release; a 6% attack rate was observed in Internet interviewees who responded later (P=.02). Among telephone interviewees, early responders had an attack rate of 17% vs. 20% among later interviewees.

Camping in the organizer’s tents during the event was significantly associated with illness among Internet interviewees (RR=3.0; 95% CI, 1.4-6.0) and in the combined data set (Mantel-Haenszel summary RR=2.3; 95% CI, 1.3-4.0), but not among telephone interviewees (RR=1.3; 95% CI, 0.5-3.8).

The researchers suggested that more frequent e-mail reminders among Internet interviewees may increase overall response, but that the absence of a prompter to encourage survey completion may have hindered the Internet response rate.

“Internet surveys will likely be increasingly used to investigate outbreaks,” the researchers wrote.