Jordy van Sambeeck

Chapter 8 130 Discussion The most important finding of this study was that the NPI score and the BPII demonstrated good reliabyutility in patients after surgery for patellar instability. This is the first study to translate the Norwich Patellar Instability score and Banff Patellar Instability Instrument to Dutch and validate it for use in patients treated for patellar instability. Cronbach’s alpha was calculated to evaluate internal consistency, it was excellent for both the NPI and BPII. This is in concordance with previous validation studies of these measurements 8, 12, 13, 18 . Cronbach’s alpha depends on the number of items in a questionnaire and on the variation in the population, since BPII consists of more items, one could expect Cronbach’s alpha is higher in the BPII than in the NPI, but our data does not support this. The high values indicate that the scores measure an unified construct, but also that there is a redundancy in the measurement and both scores could perhaps be shortened. Various questions in both scores were left unanswered. This may indicate that the wording of the questions was unclear, but also that patients experienced that the questionnaire is too long or seems repetitive. Floor and ceiling effects are measures of content validity of a questionnaire, low effects are desired in instruments with good content validity 22 . NRS pain at rest had a large floor effect. There was a large ceiling effects for SF36 RP, RE and SF and a small ceiling effect for KOOS ADL. The ceiling effects of these scores make them clinically less meaningful for evaluation and follow up since patient’s clinical improvement cannot be quantified by the scores. This is worrisome, especially in the light of improving clinical patient outcomes. The floor effect of the NPI is in concordance with the results of Smith et al. 13 , and demands future study assessing appropriateness of the questions in the NPI which demonstrated this floor effect. Both the Dutch BPII and the NPI scores had fairly good correlation with almost all other scores. The correlation was strongest with the KKS, as one would expect because all three scores are specifically designed for patients with patellofemoral complaints. Therewas amoderate correlationwith other the knee specific outcome scores, the KOOS, on all subscales. For the non-specific scores (SF36, NRS), the correlation was acceptable. Only the SF36 sub-scale limitations due to physical problems showed poor correlation with the Dutch BPII and NPI (convergent validity, ρ > 0.5), and the SF 36 subscale social functioning correlated poorly with the BPII (divergent validity, ρ < 0.3). Conclusion The results of this study indicate that the Dutch version of the BPII and the NPI can be used for patients with patellar instability. The BPII is rather long, this might limit its routine use. The NPI score has a floor effect, resulting in limitations in