To The Editor

Letter to the Editor concerning “Development and Validation of Cervical Prediction Models for Patient-Reported Outcomes at 1 Year After Cervical Spine Surgery for Radiculopathy and Myelopathy” by Archer et al. Spine 45:1541-1552, 2020

Archer et al¹ provide a great deal of information which required a lot of work to gather and analyze from a large dataset. But the article raises some questions.

• 1. There is no comparison with conservative care. The study claims to provide patients and their surgeons with surgical outcomes for patients with particular demographic characteristics that they can use to make more informed decisions whether to proceed with an operation or not. I would expect the patients to then ask, “What will happen if I don’t have an operation?” This study is not designed to answer that question, which limits its usefulness.
• 2. The results are given for groups of patients, and cannot be used to “provide an individualized estimate” as the authors claim. The average outcome for a group of patients does not predict the outcome for an individual member of the group. Tossing a coin 100 times and getting heads 50 times does not predict the result of the next toss, only the average of the next 100 tosses.
• 3. Incomplete follow-up. As the authors point out, results were available for only 61% of the patients treated. This low figure limits the conclusions which can be drawn, particularly for some small demographic groups: those with Worker's Compensation who composed 3% of the total, ambulating with assistance 8%, posterior disc decompression 9%, and posterior fusion 9%. We are not told how many of the patients in these small groups were followed for 12 months.
• 4. What is “better”? The numerical outcomes are not given. Figures 1 and 2 give “odds ratios of having better 12-month PRO.” The authors do not tell us how they decided what was better, any numerical improvement, or a minimal clinically important difference.
• 5. Modifiable risk factors. The authors claim to identify modifiable risk factors, but the only one they report is smoking. They do not provide evidence that stopping smoking improves outcomes.
• 6. Generalizability. The Quality Outcome Database used seems to have been assembled from academic centers. The analysis may not be applicable to other practitioners.

This article contains a lot of information which may reasonably provide guidance to surgeons and patients, but not to the individualized precision which the authors suggest.