Interrater reliability of the Tinetti Balance Scores in novice and experienced physical therapy clinicians.

OBJECTIVE

To examine interrater agreement of scores by physical therapy novices and experienced clinicians on videotaped and live performances of the balance portion of Tinetti's Performance Oriented Mobility Assessment (BPOMA).

DESIGN

A reliability design was used to assess the interrater agreement and consistency of the BPOMA scores in an elderly population.

SETTING

General community hospital and skilled nursing facility.

PATIENTS

Twenty-six residents of a skilled nursing home, ranging in age from 66 to 99 yrs (mean = 80.4, SD = 6.8), participated in Phase 1. Twenty-four hospital inpatients and five residents of a skilled nursing home, ranging in age from 60 to 92 yrs (mean = 74.7, SD = 7.9), participated in Phase 2. RATERS: Three student physical therapists scored the patients in Phase 1. One student was designated the administrating rater (AR). The AR instructed, guarded, and scored the subjects. The other two students were the observing raters (ORs), whose role was to observe and score the subject's performances. Nine physical therapy clinicians, ranging from 0 to 6 years of experience, rated subjects in Phase 2.

MAIN OUTCOME MEASURES

Consistency and agreement of BPOMA scores were compared between clinicians with varying levels of experience. In Phase I, BPOMA was scored on-site by three student physical therapists. In Phase 2, videotaped performances were scored by five physical therapists, one physical therapist assistant, and three student physical therapists.

RESULTS

Phase 1 demonstrated fair to excellent kappa coefficients (.40-1.00) in all maneuvers across all raters. The ORs had higher agreement compared with the AR, ranging from good to excellent (.75-1.00). Phase 2 demonstrated fair to good kappa coefficients (.40-.75) in 5 of 8 maneuvers across all nine raters. When comparing proportion of observed agreement to evaluate the years of experience on rater agreement, there was no significant difference between clinician groups.

CONCLUSIONS

Fair to good reliability of BPOMA scores occurred across many rates of varied experience with a small amount of training.