Special thanks to Kurt Spindler. I joined Kurt in practice in 2002 and joined the ACL group. Actually, I think I'm the only person to have left the ACL group because my numbers of ACL were so low that Kurt made me leave, but we used his model and his mistakes to avoid to create the shoulder group. And these are some of the things that we have learned. Let's see if I can get this to advance. There we go. So I want to thank Kurt Spindler. As I mentioned, he set the bar for me and helped me build this group. And again, the most important thing that I've learned is that multicenter research is extremely hard, but it's also very rewarding. We've helped a lot of patients through the years. It's been rewarding academically for everybody in our group. And of course, we've made great friends and colleagues along the way, and I wouldn't be here without my co-authors and co-contributors to this effort. There we go. So this is our moon shoulder group. These are the sites that we used to start with. It's much larger now. We started in May of 2004 with 16 surgeons, 11 sites around the country, both academic and private practice, and we decided to get together to study shoulder disorders. We wanted to look at rotator cuff disease, and we wanted to do that because we thought there was more funding in cuff disease than instability. And we spent three years doing preliminary work. We had to define our classification systems, define the disease using the best evidence. We did a lot of agreement studies classifying rotator cuff disease on MRI, arthroscopically, and radiographically. And we wanted to align our practice patterns to reduce variables. So we did systematic reviews looking at the effect of injections, rehabilitation, post-operative treatment, and we tried to get the same indications for surgery. We couldn't have people in Iowa operating on a different cohort of patients than people in California. When we did this, however, what we found is we could not agree on the indications for surgery. Every surgeon had a different idea about who needed surgery and who didn't. We knew some people didn't need surgery, but we couldn't agree on who those people were, so we decided to do a pilot study that morphed into a very large study. Our pilot study was basically designed to send all atraumatic, symptomatic, full-thickness rotator cuff tears through the best therapy program we could find. We knew that some would do well, and we knew that some would fail, and if we could compare those two groups, we might develop indications for surgery. So what we did is we took 452 patients. We gave them the best physical therapy program we could find that was based on a systematic review of the literature. They worked with a therapist until they were ready for home therapy, and we assessed patients at 6, 12 weeks, 1, 2, 5, and now we have 10-year data, which I'll be presenting today, but we have not published it yet. When the patients did their physical therapy and came back to clinic, we asked them, are you cured? And if they were cured, we would just follow them along with data. If they were better, we'd let them continue therapy for another six weeks, and if they were no better and wanted to have surgery, they could have surgery anywhere along the course of their treatment. We looked at a lot of covariates, anything we could think of that might influence the decision to have rotator cuff tears. Age, handedness, work compensation status, comorbidities, pain level, duration of symptoms, and we looked at rotator cuff severity, which we measured a number of different ways. We looked at the amount of retraction, the number of tendons involved. We looked at muscle atrophy and superior migration of the humeral head. We looked at a number of different patient-reported outcome scores. Some are specific for rotator cuff disease. We looked at shoulder activity level and patient expectations, and the surgery question is quite simple. Did they have surgery, yes or no? So what are our results and what have we learned? This is the data that we got after patients did physical therapy for 12 weeks. We saw that they got both statistically and clinically significant improvement in all the patient-reported outcome measures, except the SF12, which was not very responsive, and their activity, of course, over 12 weeks did not change. This is really interesting. This is data that we didn't expect to see. This is the results of our patient-reported outcomes for 10 years follow-up. The bar on the left shows where they started, and then the subsequent bars are one year, two year, five years, seven years, and 10 years. And what we can see in that every patient-reported outcome we measured, the patient-reported outcome stayed stable for 10 years, despite non-operative treatment of their cuff tears. They did not get worse over time. The bottom right table is pain on the visual analog scale. That improved significantly as well and stayed stable through the course of 10 years. Their activity level changed slightly over the course of 10 years, not statistically significant, but this is actually expected. We know normative data shows that the activity level in patients does decline as they get older. This data shows our surgery rates, and what you can see is in the first six months, a lot of patients had surgery, and then after that, these patients didn't really have surgery at very high rates. Overall, only 30% of our cohort had surgery, and that was surprising to me because I was taught that most of these patients would need an operation. Over 10 years, 9% of our cohort died, and over 10 years, only one patient, 0.2% of the entire cohort, had a reverse arthroplasty. So we can divide that curve into two groups. There's a group that had surgery early, and there's a group that had surgery later, and they're different. If we look at the early surgery patients, these are predictors of people that had surgery in those first six months, and what we found is that patient expectations drove it. The farther to the right on this curve, the greater the effect on deciding to have surgery, and patient expectations was basically if a patient thought therapy would work, it would work. If a patient didn't think therapy would work, it wouldn't work, and so this really drove the decision to have surgery early. Activity level was also very important, and rotator cuff tear retraction came up as well as a statistically significant important predictor, but what was interesting is that it wasn't what you would expect. The smaller tears were more likely to have surgery than the larger tears, and we're still trying to figure out what that means. Later surgery was not affected by rotator cuff anatomy at all. The predictors of later surgery included workers' compensation, patient activity level, and again, patient expectations came into view, but again, the anatomy of the rotator cuff tear had nothing to do with the decision to have surgery later. We did some other data research looking at different questions that we had with this great data set that we had, and what we found is that the rotator cuff tear severity had no relation to the patient's level of pain. It had no relation to their duration of symptoms, which is counterintuitive. You would think people with larger tears would have had symptoms for a longer period of time. They did not, and it was not correlated with activity level at all, and the question then becomes prominent, is the anatomy important? What are we really treating here? So what have I learned? Well, our data leads to more questions whenever you do research. The first thing we note is that physical therapy is effective for more than 70% of our patients with atraumatic, symptomatic, full thickness rotator cuff tears, and the effect lasts for 10 years. What's interesting is that if you take away the patient expectation part, 88% of those patients would get better with physical therapy. We also learned that the patient reported outcome measures do not deteriorate over 10 years. Now, we know that rotator cuff tears progress. Ken Yamaguchi and Jay Keener at Wash U have done some really great natural history studies that show that rotator cuff tears do progress over time, but I can tell you in 10 years it does not seem to be affecting their patient reported outcome measures, and you wonder if progression develops to the point where it becomes clinically important. We know that early surgery is really driven by patient expectations, and it makes us wonder, are patient expectations modifiable? Could you educate a patient to let them know that physical therapy is effective, and it'll change his opinion about physical therapy and maybe respond to it? That's a really great research question that we have to do. Activity level seems to influence both early and late surgery, so for me, activity level becomes a better indication for surgery than pain. And cuff tear severity is not related to symptoms or activity level, and so when we look at the anatomy, it's really helpful for us to plan surgery, but it really probably shouldn't be used to make a decision about whether somebody needs surgery or not. Most important thing we've learned is that the team approach works. Beauchamp Beckler said, no man is more important than the team, but no coach is more important than the team. And as a Michigan grad and alumnus, we do have people in our group from Ohio State, so I just wanted to put this in their face. I would tell you to stay tuned. Carolyn Hetrick has used our group and expanded it to study instability, and the next few years you're going to see some amazing data coming out of this group. Thanks for your attention.

moon shoulder group

rotator cuff tears

non-operative treatment

patient expectations

tear severity