Goal attainment scaling: a direct comparison of alternative rating methods.

Purpose: To compare two alternative rating systems for goal attainment scaling (GAS) intended to improve capture of goal underachievement.
Setting: A tertiary inpatient neurorehabilitation unit.
Design: Secondary comparative analysis of prospective cohort data.
Participants: A total of 243 patients were included, with 977 goals between them; M:F ratio 146:97, mean age 44.8 (SD 14.2) years.
Diagnosis: 200 (82%) acquired brain injury, 21 (9%) spinal cord injury, 22 (9%) other neurological conditions.
Methods: Goal attainment scaling-rated achievement of 1-6 patient-selected goals was collected using the standard 5-point scale (-2 to +2) and formula to derive aggregated T-scores, as recommended by its originators. Two alternative 6-point rating systems were applied retrospectively. Version 1 set all baseline scores at '-2' and added '-3' to denote 'worsening'. Version 2 added a '-0.5' score to denote 'partial achievement' for goals starting at '-1', without change to range or baseline scores.
Results: Baseline scores were identical for standard rating and version 2 (median 34.9, interquartile range (IQR) 31.9-35.5), but lower for version 1 (median 21.0, IQR 19.8-22.6), which was reflected in the change scores. While median achieved T-scores were 50.0 for all three methods, version 2 marginally overestimated goal attainment (Wilcoxon z -6.8, P < 0.001), while version 1 underestimated it (Wilcoxon z -7.2, P < 0.001), in comparison with standard goal rating.
Conclusion: Different goal rating methods may have significant impact on goal attainment scaling results. Version 2 provided the closest match to standard rating, supporting documentation of partial goal achievement without affecting score range. For analysis and reporting, '0.5' scores should be converted to '-1', to maintain parity with standard rating systems.