Orthopedics & Sports Medicine

Orthopedic Surgery Patients Don't Care About P Values: Why Researchers Need to Change Their Approach

Sep. 27, 2023 - Eden McCleskey

The foot bone's connected to the leg bone. The leg bone's connected to the knee bone. And the knee bone's connected to … well … every other bone and every other orthopedic surgery patient in the world thanks to the increasingly complex, constantly changing, ubiquitous web of interconnected data at the heart of health care today.

Gone are the days when follow-up was optional, outcomes were anecdotal and patient satisfaction was conveyed exclusively by word of mouth. Nearly 90 percent of office-based physicians use electronic medical records. Data from over 7,000 CMS-certified hospitals is compiled into public websites tracking over 150 quality measures. There are over 28,000 peer-reviewed medical journals in the world, and more than 1 million papers added to the PubMed database annually.

Although the enhanced focus on big data and clinical research has undoubtedly advanced public health, the threat of information overload and "missing the forest for the trees" has never been more real.

Perhaps that's why an Arthroscopy article published in 2017 called Research Pearls: The Significance of Statistics and the Perils of Pooling quickly became one of the most downloaded and cited papers in the journal's history.

In the paper, Dr. Joshua Harris, a Houston Methodist orthopedic surgeon and Arthroscopy's then associate editor, and his co-authors called for a revolution of sorts, urging orthopedic surgery researchers to focus less on statistical significance and objective clinician-measured data and more on clinically significant outcomes and meaningful patient-centered endpoints.

"Patients do not care about P-values," Dr. Harris said. "They don't care if their knee is technically 3.4 percent stronger if it still doesn't feel good to use it. They care if they can stand up and walk without pain; if they are able to do the things they want to do; and if their quality of life has noticeably improved."

In a recently published follow-up piece, Guidelines for Proper Reporting of Clinical Significance, Dr. Harris argues that measures of clinical significance — including minimal clinically important difference (MCID), patient acceptable symptomatic state (PASS) and substantial clinical benefit (SCB) — should go beyond statistical measures and emphasize patient-specific clinical relevance.

"Nobody goes into orthopedic surgery looking for the smallest detectable clinical benefit," Dr. Harris said. "They want a major, durable improvement. Anything short of that will probably not meet their expectations."

Dr. Harris suggests using a system called "maximal outcome improvement," which measures actual outcome improvement relative to potential improvement using anchor-based questions about patient satisfaction.

It's not that objective data isn't important. We shouldn't throw the baby out with the bathwater, Dr. Harris emphasizes, because the best insights come from studies that show alignment between statistical significance and clinical importance.

"The important thing is not to let the pendulum swing too far in any direction," said Dr. Harris. "I think our original study really resonated because publishers, journals and researchers relied heavily on statistical significance. Our paper went beyond the stats and simply explained what measures of clinical significance were and how to use them in writing, reading and interpreting orthopedic surgery results."

Dr. Harris sat down with Leading Medicine blog to discuss the significance of his latest article and offer up a fresh round of "research pearls" for anyone in the market for them.

Q: What is the basic thesis of this piece?

Dr. Harris: Since we published the original piece in 2017, we've come a long way in terms of realizing that patient-reported outcomes measures (PROMs) are the best measures of patient satisfaction. However, not all PROMs are created equal. They need to be responsive, reliable and validated for the specific condition/treatment being investigated, and they need to exhibit a dose-dependent response across a diverse patient population, unlimited by floor and ceiling effects.

A recent Arthroscopy meta-analysis by Levy et al. investigated 81 studies analyzing more than 9,300 hips. They showed that subjects met MCID 97 percent of the time for modified Harris Hip Score (measure of functional outcome and pain), 90 percent of the time for Hip Outcome Scores Activities of Daily Living subscale and 93 percent of the time for Sport-Specific subscales.

MCID is the lowest bar — the minimum amount of patient-detectable difference. Only 88 percent, 25 percent and 30 percent of the same patients met PASS, respectively. PASS is a numerical score threshold of acceptability or satisfaction, so a little better than MCID, but obviously a variable amount based on the person.

This discrepancy between MCID and PASS scores shows us that, although patients are able to detect their post-surgical change as an improvement, the absolute value of their level of satisfaction may not meet their own expectations.

As authors, readers and purveyors of research quality, we should focus on measures of clinical significance, not just statistical measures like P values in isolation. We propose utilizing metrics that more meaningfully assess outcomes improvements to help determine the appropriate treatment strategies and counsel our patients on comparative therapeutic options for their condition.

Q: Is that what maximal outcome improvement (MOI) scores do?

Dr. Harris: Basically, yes. Maximal outcome improvement is a simple calculation that anyone can do — no advanced degree required. It is calculated as: Outcome score at most recent follow-up – Preoperative outcome score × 100%/Highest possible outcome score – Preoperative outcome score.

In layman's terms, it is the ratio of how you did relative to how well you could have done. It is reported as a percentage, with higher values reflecting achievement of maximum amount of potential improvement. Using MOI with a satisfaction anchor question for established PROMs has very good levels of discrimination and high diagnostic accuracy values. It is a more accurate way to represent the full spectrum of patients, including those with higher baseline or preoperative PROMs.

Q: What advice do you have for today's orthopedic surgeons?

Dr. Harris: As health care increasingly relies on PROMs to gauge the effectiveness of interventions, it is essential to recognize the distinction between statistical significance and clinical relevance. While statistical significance is determined through mathematical calculations and may indicate a measurable difference, it may not necessarily reflect a change that patients perceive as meaningful.

In the past, when a patient came to an orthopedic surgeon with a complaint, they may have said, "I did the surgery, and your X-rays look fine" or "I have no idea why you're still hurting." That's just not good enough in today's world. We treat patients, not X-rays, not MRIs. Yes, you still need both objective and subjective data, but if you have to pick just one, the subjective matters more.

If my patient wants to run a marathon again, or be able to walk a mile, who am I to judge their desires? It's my job to help them either achieve what they want to achieve or at least give them a realistic picture of what orthopedic surgery is capable of before they go under the knife. That's how we practice here. That's how we build trust and demonstrate that we're 'Leading Medicine.'

Stay up-to-date
By signing up, you will receive information on our latest research, educational opportunities and surgical videos.
Please Enter Email
Please Enter Valid Email


Ortho Research Orthopedics & Sports Medicine