Tuesday, 28 April 2015

Do shoulder fractures need surgery?

Fractures that occur at the upper end of the humerus near the shoulder (called humeral neck fractures) are common. They are often treated with surgery despite a lack of supporting evidence for this, particularly in older, osteoporotic patients. Now, thanks to a recent study from the UK, it is possible that most of these fractures don’t need surgery, even in young patients. This is big news, but will this research jump the gap from research into practice and influence the decision making of the end users – the patient and their surgeon?

Humeral neck fractures are one of the most common fractures in the elderly but they can occur from falls or accidents at any age. They come in many shapes and sizes. The ones that are not very displaced do not need surgery. The ones that are associated with dislocations of the shoulder joint are treated surgically. However, for all the ones in between – the displaced fractures – the decision making is not so clear.

Current decision making
Like a lot of other fractures, we know that we can make the X-ray look better if we operate to put the broken pieces back in alignment, or if we replace the broken bits with an artificial shoulder joint. This is what is commonly done for displaced fractures, but are we just treating the X-ray and not the patient?

Fractures in older patients
There has always been some debate about the advisability of surgery in this group of patients. The advantages of making the X-ray look better are balanced against the risk of failure due to the screws coming loose in the soft bone, infection, further fracturing, dislocation of the joint, and other general surgical complications, all of which are a little more common in frail patients.

The evidence up to now
There have been several randomised trials comparing surgery to non-operative treatment for displaced fractures, starting from the 80’s. In a review of these studies (here), it was shown that there is very little evidence that our attempts to fix things up with surgery make any difference to the patients in terms of their pain or shoulder function. In fact, the only difference is that surgery tends to lead to more complications and more secondary surgery than non-operative treatment.

The practice up until now
There has always been considerable practice variation, with some surgeons treating many of these with surgery, and other surgeons treating most of them without surgery. And the decision making for each type of surgeon was not based on solid science – more of a feel. Over the years I have seen young surgeons who were very keen to fix these, thinking that they will make the patient better by making the X-ray better, and then seen these surgeons reverse their practice after a few unfavourable cases and treat most of them non-operatively.

Despite evidence showing a lack of benefit with surgery, practice still varies between surgeons and as long as newer implants are being introduced, there is always someone that believes that the new implant will be the answer to this problem. The introduction of “fixed-angle” or “locking” screws in plates was an example of this. It was widely taken up but has failed to deliver benefits to patients. I must admit that the studies up until now (that show surgery to be of no benefit) have not been without flaws, but that shouldn’t allow us to assume that the results from these studies are all wrong. There are still no good studies supporting surgery.

The evidence now
A study released in 2015 in JAMA (here) has provided good evidence that surgery is not necessary for most of these displaced fractures. There are two things about this study that I would like to highlight. Firstly, it is a good study that is likely to be generalisable to practice in most developed countries. This is because it is a “pragmatic” study (my favourite type) because it took all comers and left it to the surgeons as to what operation they thought would be best (in the group that was randomised to surgery) therefore reflecting real life. Secondly, unlike the previous studies, it recruited all patients aged 16 and over, not just older patients. I thought this was a big call because the assumption has always been (admittedly in the absence of evidence) that surgery was beneficial for younger patients, in whom there was a lower likelihood of the implants coming loose.

They recruited 250 patients from 32 centres across the NHS with fractures that the surgeons thought would benefit from surgery, and then randomised them to having surgery or non-operative treatment. They had good follow up and measured the results up to two years. Other aspects of the study were very good and they only excluded very bad fractures (dislocations of the joint or cases where the bone was pushing under the skin). There was no meaningful differences between the groups in the outcomes measured (shoulder function, quality of life etc.), including in the younger patients where, if anything (and surprisingly), the non-operative group did slightly better.

The bottom line
The lure of newer implants, of making bones line up and of actively doing something has led doctors to treat many of these common fractures with surgery, despite a lack of evidence of any benefit, and some evidence of a lack of benefit. Now there is high quality evidence that surgery is of no benefit for most of these fractures. I am interested to see if this will change the way people treat these fractures. I think it will, but these things take time.


  1. I received a good comment just now that I accidentally rejected, but I have salvaged it and pasted it below. It was from "Anonymous".

    Although it seems ironic to ask someone who refers to themselves as “Dr Skeptic” to demonstrate a little more skepticism about what they read- I can't help but think your above summary of the PROFHER trial lacks critical appraisal.Your statement that it is “a good study…because it is a pragmatic study (my favourite type)” belies this lack of critical rigour.
    Pragmatic studies are neither better nor worse than explanatory studies. They are suitable for the investigation of a treatment that is easily and reliably reproducible. The assumption inherent in thisl is that the real world has well established treatments for this condition that do not have a high skill/technique demand or variability. A classic example of where they are appropriate is a drug with a broad therapeutic range and few adverse effects. Contrast this to an investigation of a drug with a narrow dose range and potential toxicity investigated in a study that allows participants to be given any dose. Even that may test the “real world” use of the drug- it does little to indicate if the drug is effective or not.
    ORIF/hemiarthroplasty for proximal humerus fractures is a particularly technically demanding procedure. Most centres would advocate it only being done by high frequency surgeons (>15 per year). In this series 66 surgeons at 30 centres performed 109 operations in 4 years. That equates to a remarkably low 3 operative procedures per hospital in 4 years or 1.8 ops per surgeon in 4 years! Thus this series does not necessarily represent experienced and skilled surgeons.
    The PROFHER trial takes pragmatic trials to a ludicrous extreme- the surgical treatment arm includes 10 hemiarthroplasties, 4 nails, 5 “other” in addition to 90 plate ORIFs.Hemiarthroplasties in particular are clearly associated with inferior outcome and if technical aspects not tightly controlled (i.e. tuberosity fixation and reduction) invariably poor results are achieved in this setting. The potential skew effect of this group must be acknowledged.
    In this series non operative group reports results that have not been achievable in any literature preceding this paper. They report remarkable results- only 5/125 with stiffness! The AVN rate in the non operative group is 1/125- far less than previous series investigating this occurrence. They also report only 5 “symptomatic non unions” in the nonoperative group.
    You believe that a strength of the paper is the wide age range. This statement is counter to an understanding of these injuries. The elderly low velocity fracture is a very different entity to the young high velocity injury. Indeed, in terms of management they are separate conditions. Concomitant soft tissue injury bone quality and ease of fixation, patient expectations and demands all are relevant considerations. The paper presents age separation as 65yo for their data presentation. This is an arbitrary figure. They do not present subgroup analysis of a clinically relevant group of 45 yo or less (I assume because they numbers are too low).
    The real conclusions of this paper are as follows: In the participating hospitals in the UK they can achieve 'world's beating' non operative treatment of SNOHs. In the UK, if relatively inexperienced surgeons perform a variety of uncontrolled and undefined procedures on all SNOH# without clearly defined surgical goals to prevent failure then the results on a gross shoulder score test will be the same.
    The great disappointment of this paper is what it could have been. If clinical research is better designed, by clinicians with a clear understanding of previous literature and the clinical context then hopefully a more rigorous study that answers the question of who requires fixation and how best to do it. Pragmatic trials done, ‘because they are easier’ in a web of preexisting evidence, should be recognised as such.

    1. Sorry for the accidental (possible subconscious?) initial rejection of this comment. Thanks for raising the pragmatic versus explanatory trial argument. You raise some good points, but that particular argument is complex and readers should google 'pragmatic versus explanatory trials' to get the background.

      Firstly, about surgeon experience. These were sites in the UK that commonly treat these fractures. Where I work, most of us only treat a handful of these per year, and I work at 2 major trauma centres. I have heard the experience argument before and I have two answers. Firstly, this study reflects what happens now - it is real life - these are the surgeons who treat these fractures when they occur. If experienced surgeons are better, then we need to change the system. Secondly, I don't think that being experienced means that your results are necessarily better than non-op treatment. That is an assumption that "sounds good" and we are quick to jump to it, but the experienced surgeons need to prove it by doing their own RCT. I have heard the same argument for vertebroplasty: "the doctors in the sham RCTs weren't as good as I am". It is one of the oldest argument in surgery.

      There is no angle you can take on this paper where you come out concluding that surgery is better. At worst, you could say that the study doesn't answer the question as well as it could, but that doesn't allow you to assume that the opposite is true.

      Regarding the different ages and operations. Yes, they attempted to look at that and found no differences. If the numbers in some groups were small, then that weakens their argument, but it doesn't allow us to assume the opposite.

      To me, this study tells me that if you have one of these fractures in one of these many centres in the UK today, you are unlikely to significantly benefit from surgery. You might get better results if you are treated somewhere else, but I am still waiting for that evidence. RCTs since the 1980s have been telling us that surgery does not provide a benefit for these fractures, but I look forward to the explanatory trial from the high volume surgeons.