Rapid assessment of psychosocial well-being: are computers the way forward in a clinical setting?

OBJECTIVE: To compare the agreement between anxiety, depression and supportive care needs data obtained using a touchscreen computer survey with traditional pen-and-paper surveys.
METHODS: The sample consisted of 1,304 consecutive patients attending the medical or radiation oncology outpatient department in two public cancer treatment centres. Of the 357 eligible patients, 355 (99%) consented to participate and 350 (98%) completed both touchscreen computer and pen-and-paper versions of the Hospital Anxiety and Depression Scale (HADS) and short-form Supportive Care Needs Survey (SCNS-short).
RESULTS: Simple kappa values indicated moderate agreement (X = 0.57) between the two modes of survey administration for most HADS items. Simple kappa coefficients indicated only fair agreement (X = 0.28) when a simplified response option format was used in the computerised SCNS-short. When the paper and computerised survey used the same response format, simple kappa coefficients increased and indicated moderate agreement (X = 0.44) for most SCNS-short items. Kappa coefficients indicated at least moderate agreement (kappa > 0.41) in identifying patients with elevated levels of anxiety and depression; there were no significant differences in the proportion of patients identified with elevated levels of anxiety and depression according to computer survey compared to paper survey. Prevalence-adjusted bias-adjusted kappa (PABAK) coefficients indicated at least moderate agreement (X = 0.79) in identifying participants with moderate/high levels of unmet needs. However, participants tended to report lower levels of unmet needs with a simplified response format in the computerised SCNS-short compared to the paper survey. This was not observed when the response format of the computerised SCNS-short replicated the paper survey.
CONCLUSIONS: Despite the advantages that computerised surveys offer for simplifying survey presentation, current results suggest the need to exactly replicate the question and response option format of the original paper survey to ensure the data collected are equivalent. This finding is particularly important given the potential application of computerised surveys in the clinical setting to quickly assess and identify patients' concerns requiring intervention by health care providers.