SAN DIEGO – The Neck Disability Index–10 did not perform as well as the Neck Disability Index–5 in assessing patient-reported outcomes in cervical spine patients – and neither was as good as the PROMIS physical function domain delivered by computerized adaptive testing.
Those are the key findings from an analysis of data from more than 500 cervical spine patients treated at University of Utah Health Care in Salt Lake City.
Dr. Darrel S. Brodke
“Previous studies by us and others have shown problems with the NDI [Neck Disability Index] as it is commonly administered” in 10 questions, lead study author Dr. Darrel S. Brodke said in an interview in advance of the annual meeting of the Cervical Spine Research Society. “It has a very poor floor effect, meaning that it does not differentiate between minimally disabled patients, and the scores cannot be appropriately handled with the kinds of statistics that we normally use – though because few of us know this, we still use it as a standard parametric measure.”
In what he said is the first study of its kind, Dr. Brodke, professor of orthopedics at the University of Utah, and his associates set out to compare the psychometric performance of the National Institutes of Health–funded PROMIS (Patient Reported Outcomes Measurement Information System) physical function (PF) domain, administered by computerized adaptive testing, with the standard NDI-10, the NDI-5, and the 36-Item Short Form physical function domain (SF-36 PFD).
In all, 566 patients completed the NDI and PROMIS PF computerized adaptive testing assessments, while 490 also completed the SF-36 PFD.
On average, the NDI-10 took the longest to complete (10 questions in a mean of 183 seconds), followed by the SF-36 PFD (5 questions in a mean of 123 seconds), the NDI-5 (5 questions in a mean of 99 seconds), and the PROMIS PF computerized adaptive testing (between 4 and 12 questions in a mean of 62 seconds).
The psychometric properties of the PROMIS PF computerized adaptive testing were superior to the other outcome measurement tools studied, Dr. Brodke reported. Specifically, the ceiling and floor effects were “excellent” for the PROMIS PF computerized adaptive testing (1.94% and 4.06%, respectively), while the ceiling effects were “fine” for the NDI-10 (4.77%), NDI-5 (7.60%), and SF-36 PFD (11.84%), he said.
However, the floor effects of these three instruments were poor (45.58%, 48.59% and 21.55%, respectively). “The NDI-10 also has the additional challenge of extremely poor raw score to measure correlation,” the researchers noted in their abstract.
“The legacy scale scores significantly predicted the PROMIS PF CAT scores (P less than .0001), with fair correlation for the PF CAT and NDI-10 (0.53) and good correlation of PF CAT and SF-36 PFD (0.62), allowing use of conversion equations to predict scores, which were generated,” the investigators explained.
PROMIS PF computerized adaptive testing “does much better than the NDI or the SF-36 physical function domain at characterizing patients’ physical function, with much better coverage,” Dr. Brodke said. “Not only this, but it is also much faster to fill out, so less burdensome to the patient and the clinic.”
One limitation of the study is that the researchers did not measure the responsiveness aspect of PROMIS performance. “We did not have enough pre- and posttreatment scores to do this measurement yet,” Dr. Brodke said. “The other thing is that minimum clinically important difference [MCID] is not yet worked out for PROMIS in this patient population, though we can infer an MCID as one-half of a standard deviation. More to come in future studies.”
Dr. Brodke reported having no financial disclosures.