Effect of different questionnaire formats on item nonresponse in older patients with cancer.

Background: Self-administered questionnaire surveys are instrumental in revealing patient experiences. However, concerns about the quality of data and validity of results arise from nonresponses, either to the entire survey or to specific items therein. The format of the survey can affect the ease of answering and subsequent responses. Older patients who are more likely to have some cognitive decline may be especially sensitive to questionnaire comprehensibility and response methods.

Methods: We surveyed the experiences of patients aged 65 years and older with cancer in four hospitals. Randomly selected patients from the hospital-based cancer registries received questionnaires with either the grid or text-choice response formats. We compared item nonresponse rates for each question between the two types of questionnaires.

Results: Out of invited 1500 patients, 51.3% responded. We included data from 675 patients (331 and 326 in the grid and text-choice formats, respectively) in the final analysis. The non-response rate of items considering participant demographics, including age, gender, and cancer sites, did not differ between groups (mean age: 76.3 and 75.8 years in the grid and text-choice formats, respectively). Most items with the grid format had higher proportions of item nonresponse than those with the text-choice format. The difference was more pronounced in patients aged 75 years and above, compared to those aged 65-74 years.

Conclusions: In older patients, text-choice format survey questions elicited fewer item nonresponses compared to grid format questions. The text-choice format should preferably be used consistently throughout surveys targeting older persons.