Results from a pilot study published last week by the researchers at the University of Michigan Medical School demonstrated that among low-income urban African-Americans in Detroit, “text messaging is not only acceptable and feasible but is the preferred method of collecting real-time survey data” over other methods of survey including paper, phone, internet, and in-person.
The survey used in the study consisted of 14 hypothetical medical scenarios – 10 questions reflected 10 leading reasons for urgent outpatient medical visits, 4 questions reflected extreme scenarios. For example, as a follow to, “You slipped in the bathroom, injured your back, it hurts to lie down and when you bend over or twist”, participants were asked to choose between visit emergency department or ask primary care doctor or do nothing. Four weeks after the paper survey was completed, identical questions were asked by text messages. These included two text message sent per day at different times of day for 6 weeks. Each question was sent six times: twice during regular hours (8 am-4 pm), twice during off hours (5 pm-7 am) and twice during the weekend. The participant’s response was followed by a question seeking a free text explanation for the choice made.
The study consisted of 20 subjects (17 women) with a median age of 30.7 years and were compensated for their participation – receiving $1 for a multiple choice response, $1 for additional free text response, as well as $20 for attendance at a meeting (recruitment meeting, focus group, etc.) All the participants lived in Village at Parkside, a public housing complex on the eastside of Detroit.
The results from the study found that the average response rate by participant was 72% and the median response time was five minutes. Response rate during business hours was 72.0% with a median response time of six minutes compared to 72.4% response rate during evening hours and weekends with a median response time of four minutes. The responses were elicited for 72% of all multiple choice questions sent by text message and 76% of the questions requiring responses on Likert-like scales. The responses on the paper and text message surveys did not differ. Ninety-five percent of the participants provided free text responses such as – “Er why u slip in tub an u cud of broke anythin so yup er”; “Nothin cuz a asprin will do da trick”; “Er because it mite be broken”; “Nothing cuz over da corner medicine will do”.
Given the limitations of this study – small sample size, subjects from one urban, low-income community in Detroit, to make these results more generalizable, the future studies should have a larger sample size, and include greater diversity in ethnicity, socio-economic status, and geography.