Sunday, 26 August 2012

Is back fusion surgery just a placebo?


It is possible that spine fusion surgery for back pain achieves its effectiveness through the placebo effect. I would like to make the case that it is not only possible, but also probable.

Background
Spine fusion can be done for many reasons, but the most common reason is degenerative conditions in the lumbar spine (here). Yet there is very little evidence that spine fusion surgery for back pain is effective. It is very expensive (the implants alone are often in the tens of thousands per case), often leads to complications, often requires further surgery, is associated with increased mortality (here, are, some, good, links), and often does not even result in the spine being fused. That last one is not a big deal, because the results of the surgery are not well correlated with whether or not the spine fuses. Nor are the results related to how you fuse the spine, or whether or not you use implants (screws, rods and cages), except that “instrumented” fusions (using implants, which is now routine) are associated with more complications (Cochrane review).

The rates
The rate of spine fusion surgery is increasing and has been increasing for many years (here). The most recent data shows that in the USA, the rate has gone a long way past the rate of 1 spine fusion per 1,000 population per year. The rate has already overtaken hip replacement surgery and continues to rise. The rates of surgery vary widely across the USA, where back fusion is associated with the highest degree of practice variation (here). In my state, back fusion just beat prostatectomy as the procedure with the most regional variation (here). The US rates are far higher than most other countries, but the rates in many of those countries (like Australia) are increasing too.
  
The evidence
Three randomised clinical trials have been published comparing surgery to non-operative treatment for back pain. There have been no sham surgery trials, but the evidence from these three trials indicate that this surgery might achieve its results through the placebo effect.

Two of the studies put spine fusion surgery up against structured non-operative treatment alternatives: cognitive behavioural therapy in one study, and intensive rehabilitation in another. These studies found no significant differences in the outcomes between the operative and non-operative groups (except that the complication rate in the surgical group was higher). As usual (and from my other blogs about placebo surgery, you will guess), both groups showed improvement. The point is that about the same proportion of patients in each group improved, by roughly the same amount.

The third study concluded that the surgical group did better. Interestingly, the surgical group didn’t do any better than in the other studies; the difference was that the non-operative group didn’t get better at all. This is because the non-operative treatment was not dressed up as something that might work (i.e., it wasn't a good placebo). The authors state that the non-operative treatment "could vary within broad but commonly used limits reflecting the nonsurgical treatment policy in the society”. Patients were basically given more of the treatment that they had received before (the treatment that had previously failed and that led them into the study). Faced with more of the treatment that had already failed, this group was never going to get better – there is no placebo effect in doing nothing.

Interestingly, in the rare cases that physical therapies have been compared to sham treatments, there is no difference in the results (two examples here and here)

The bottom line
My conclusion from the three trials is that if you do something that looks like it might work, is structured and has plausibility, about 2/3 of the patients will get better. Much like many of the placebo treatments already discussed in my blog. In fact, my review of spine injections showed that many patients improve with injections, but when compared to placebo injections, the results are no better.

Surgery has the exact effect I would expect it to have if it was a placebo. If it was a placebo, it would also explain why it doesn’t matter what surgical approach you use, whether or not you put any implants in, and whether or not the spine fuses.

I once attended a lecture by an internationally renowned spine surgeon who explained that nearly all treatments for back pain result in improvements in about 2/3 of the patients. I asked him why he did spine fusions for back pain, if it was no more effective than the alternatives. He said, “Because it works in about 2/3 of the patients”.

Spine surgery is not just a sugar pill; it is a much more elaborate placebo than that, and it is much more dangerous. The onus is on doctors to prove that spine fusion surgery for back pain is better than placebo before subjecting so many people to the risks of such major surgery.


Additional points:

Spine fusion surgery for other indications is not that clear either. 
Earlier this year, a review of the indications for spine fusion surgery for the most common type of spine fracture (thoraco-lumbar burst fracture) showed that spine surgery led to similar clinical results compared to non-operative treatment, but was more expensive and associated with more complications.

Spine fusion surgery is a multi-billion dollar per year industry rife with conflicts of interest and industry influence. 
I can’t possibly cover that side of it in one blog post. Many others have written extensively on this aspect of the topic. Try Shannon Brownlee from her book Overtreated, or Nortin Hadler’s cleverly titled book Stabbed in the Back for starters.

Why do surgeons still perform this operation? 
The next blog post will cover this topic.

Addit 18 Jan 2016:
Long term follow up of one trial of surgery versus non-operative treatment for burst fractures showed that despite not much difference in the early results, the non-operative group did significantly better long term (here)

42 comments:

  1. When you speak of spinal fusion in this setting - are you only referring to it for "back pain" as opposed to radicular pain or spinal canal stenosis?

    (I did see the additional note re: thoracic fractures - however the link is broken)

    ReplyDelete
    Replies
    1. Yes, the trials that were done were usually for "back pain greater than leg pain". Back fusion has no direct role in radicular pain, which is treated by decompression - that is a whole different story, but fusions are often added to decompressions, with conflicting evidence I might add.
      Link fixed (worked on my computer, but I think it is because I have direct access to journal). Now goes to PubMed.
      Thanks.

      Delete
  2. Often in cases of severe spondylolisthesis with foraminal stenosis with facet hypertrophy it is impossible to decompress the nerve roots without destabilising the spine. The patent therefore needs a fusion, don't they?

    ReplyDelete
    Replies
    1. Thanks John,

      Fusion as an adjunct to decompression surgery is another question. As you know, it is sometimes done routinely, and sometimes only if the facet joints have been rendered unstable. The trials in this blog post were done for back pain, not in the presence of spondylolisthesis or canal stenosis etc.

      Routine fusion with spinal stenosis decompression is probably one of the biggest current debates, and it is important, as surgery for spinal stenosis is common. I will put it on the list for a future blog post.

      Delete
  3. Thank you for giving such a valuable information on spine surgery,This information is really helpful for my grandmother who is suffering from spine problems and will undergo spine surgery in india.
    Once again thank you.

    ReplyDelete
    Replies
    1. If she has back pain as her main complaint, I would have hoped that the valuable information helped her avoid surgery altogether.

      Delete
  4. Thank you for giving such a valuable information on spine surgery,This information is really helpful for my grandmother who is suffering from spine problems and will undergo spine surgery in india.
    Once again thank you.

    ReplyDelete
  5. Dr. Skeptic,

    I agree with your interpretation of the trials' results. You might want to add the following for your readers, especially those in the USA:

    Surgery has a very limited role in the treatment of back pain. Although it is probably offered to patients more often than is necessary (this is, after all, how surgeons make their living), there are a few good reasons to have surgery. A small number of surgeons think the only way to cure a medical condition is with cold, hard steel. There are also some surgeons whose only interest in the patient is in the fee they collect; they collect more for doing surgery than for talking patients out of surgery. Honest, ethical surgeons give the patient options and honest opinions. The problem may be in figuring out who is being honest and who isn't.

    Most insurance companies, especially workers compensation companies, would prefer not to pay for surgery, or long term rehabilitation instead of, or following, surgery. The less they pay out for procedures or physical therapy, the more profit they make. Stock holders like that; patients don't. And if insurance companies can find a way to no longer be the financially responsible party, all the better.

    Keeping the above two paragraphs in mind, some times surgery is a necessity. Statistically, necessary surgery probably falls in the range of 5% of cases. In another 5% of cases surgery may be appropriate because it saves the patient healing time and time off work, even if it does not affect the eventual overall outcome. This means that in 90+% of back pain cases, surgery is not needed! All surgery has the potential for complications, from pain to numbness to paralysis to death. Think hard about your choices.

    In general, a patient needs surgery when surgery can save his life, repair an injury the patient’s body cannot, or shorten significantly the recovery from an injury or disease. When it comes to the lower back, there are five absolute indications for surgery, but several elective reasons. The absolute indications for surgery are the following:
    1. Cauda equina or conus medullaris syndrome. These situations happen when there is a very large central herniated disc that compresses severely the nerves in the lower spinal cord. Without surgery, the compression would lead to eventual loss of function of those nerves, paralysis of muscles, and/or loss of sensation.
    2. Intractable pain, i.e. unremitting, severe pain.
    3. Progressive neurological deficit, loss of sensation, proprioception, muscle control, etc.
    4. New incontinence or retention, bowel or bladder.
    5. Hemorrhage into the spinal cord.

    The relative indications generally center on the relief of discomfort (less than intractable) and the shortening of recovery time from various problems: herniated disk, compression fractures, severe stenosis, severe spondylolysis, and severe spondylithesis, etc.

    There are also several types of surgery: open, micro, and minimally invasive (and endoscopic). Open takes less time, usually. The surgeon has a better view of the surgical field, but more structures are damaged and recovery is longer. Micro surgery takes longer; the field of view is smaller, but fewer structures are damaged and recovery is shorter. Minimally invasive surgery, in general, takes the longest; is the most difficult; has the narrowest field of view, but the shortest recovery time. There are trade-offs. Get the surgeon to explain them all to you. If he can't or won't, find another surgeon.

    If someone uses the words, laser surgery, he is trying to impress you. Nothing more. Lasers have their place in surgery -- usually cauterizing blood vessels. Very little surgery is done with a laser. A laser produces too much heat. Laser is a buzz word only. And the person using it is a salesman first -- surgeon second. He wants your money; he's not concerned with your best interests.

    Bill Yancey, MD
    Whatyourdoctor dot b l o g s p o t dot c o m

    ReplyDelete
    Replies
    1. Thanks for contributing Bill. Sounds like you are coming from a position of experience, so I am sure the readers will be interested in your opinions

      One point of disagreement, however. I do not believe that pain is an indication for surgery, unless the source of pain is clear and correctible. There are many patients out there with severe, chronic, unremitting, unexplained pain who should never be operated on.

      Delete
  6. Dr. Skeptic,

    You are correct. The pain source needs to be identified and any abnormalities on imaging must match the patient's complaints. The one caveat, in my opinion, would be a limited exploratory surgery for a pain source not visible on imaging. (Personal experience with terrible pain and an extruded disc fragment not visible with MRI or CT myelogram, but micro-surgery provided immediate pain relief.)

    Bill Yancey, MD
    Whatyourdoctor dot b l o g s p o t dot c o m

    ReplyDelete
  7. I had spinal fusion surgery with instrumentation from S1-L4 about two months ago, due to L5-S1 Spondylolesthesis and lumbar DDD. I've been given extremely limited information on post surgery activity and maybe I've overdone it, but my back feels like it wants to buckle just above the fusion. I requested the surgical notes and found that L3 had been relieved of its spinous processes, lamina, facet joints and transverse process along with S1-L5.. but no instrumentation was used at that level and no mention of autografting at that level. Should I be concerned about vulnerability there (it presents as a deep fissure in my healed back). And, more importantly, is anything protecting the L3/L4 disk from excessive stress? The surgeon denies having done anythng to L3, but the x-rays confirm the report (maybe he's thinking of someone else?). Finally my question: What should I do?
    I'm a very active person by nature and the thought of being crippled by instability at a new level is incredibly depressing.

    ReplyDelete
    Replies
    1. It is not really the role of this blog to give individual clinical advice, particularly on a patient that I have not seen. I find it odd the L3 had been completely stripped of its lamina, facet joints and transverse processes - that would not even happen if you WERE doing a fusion. I also don't trust all radiology reports. I assume you have a surgeon who is experienced in this area, but if you question their advice, it is reasonable to get a second opinion (but not from a radiologist).

      Delete
    2. Thank you for the prompt reply. I'll approach the surgeon again, with his own notes and x-rays in hand, and ask him what is supporting L3. If he can't remember, maybe the board of healing arts can encourage him. Incomplete and incorrect radiology readings have drawn this whole process out over years, so heed the advice given here and get second opinions or third or fourth, as seems to be the direction I'm heading. Also, have faith in your body to heal itself. It may take time, but as is suggested here - surgery may well not be your best option.

      Delete
  8. Well, this is really an eye opener. Many, including myself have been led to believe that spine fusion surgery is effective even though they should only be the last resort. Thousands of dollar for placebo is akin to day light robbery don't you think. :)

    ReplyDelete
    Replies
    1. Thanks,

      Well it certainly is expensive. The problem is that the practitioners do not think it is a placebo, so it is not robbery, it is simply biased (wishful) thinking on behalf of many practitioners.

      Delete
    2. Yes, but the question should be: why in the face of evidence (lack of effectiveness) the 'belief' remains, for a very invasive procedure? Why is dogma in medicine so difficult to challenge?! Why science-based medicine fails to live up to its promise of being science-based?

      Delete
    3. Thanks Mia,
      Beliefs are hard to dispel. We are hard-wired by evolution to perceive associations that may not exist and to jump to conclusions. Being objective and recognising the biases that distort our understanding are hard work, because they are not natural to us.
      I believe that science gradually guides practice in the right direction. There are many procedures that are no longer performed due to lack of effectiveness, that were thought to be effective at the time. Usually, however, that process takes a very long time.

      Delete
  9. Nice Post! I have been looking for such interesting blogs. I am a junior radiology consultant and i could learn tons of information from your blog! I am gonna bookmark this page. Thanks and keep posting more.

    ReplyDelete
  10. Excellent piece and discussion. How do we get this information into the wider culture?

    ANdy

    ReplyDelete
    Replies
    1. For my own personal benefit, send the link to your friends.

      For the benefit of future patients my advice is the same as with other areas of medicine: maintain a skeptical eye. Do not assume that something is helpful, just because it is new, sounds scientific, or has been recommended by a professional. Understand that the benefits-to-risk ratio of many medical interventions is overestimated by doctors themselves, and that this overestimation is magnified down the line so that by the time we get to popular media, there is little correlation with reality.

      Delete
  11. Hmm, very interesting. I am 43 and just underwent L5-S1 spinal fusion due to a Grade III Spondylolisthesis and the MRI showed moderate to severe foraminal stenosis and a moderate to severe degraded disc between L5-S1. I had been born with spina bifida and the facet joints on the L5 were malformed and broken. I had no issues and was even unaware of it until I began experiencing pain about 9 months ago. I tried physical therapy for a couple of months since I am not a big fan of surgery, but nothing stopped the lower-back pain I experienced when sitting down, which negatively impacted my ability to drive even short distances.

    So after speaking with an orthopedic/neurosurgeon and doing lots of research I determined that there was nothing to stop the L5 from continuing to slip forward or stop the disc below the L5 from continuing to degrade. Therefore I would continue to experience the current pain with the possibility of some future pain as well due to the continued degradation of the disc and slippage of the L5. So I opted for surgery sooner rather than later since I could choose my doctor, hospital and days off from work. I could have waited, but there was the possibility that some major damage could occur (like Cauda equina or conus medullaris syndrome). But why wait until that happens, which may cause more permanent damage to the nerves?

    It has been 5 weeks since surgery and I feel almost normal. I am back to my regular exercise routines (stationary bike, elliptical rider, planks, push-ups, crunches, and long walks) but without bending, twisting, or lifting heavy weights until the fusion is well underway.

    But after reading your post I get the impression that you think it is "placebo". Of course I don't think so, but how do you prove to someone who takes the position that a cure is all in the patients mind?

    ReplyDelete
    Replies
    1. Thanks Ken,

      Your condition, spondylolisthesis, is not included in any of the reviews or studies that I have mentioned. It is generally accepted as an indication for fusion surgery, although there is certainly some fuzziness around the edges when it comes to definitions of mild cases of spondylolisthesis. For example, common degenerative changes in the back are also associated with some degree of spondylolisthesis and this is often used as an excuse for surgery.

      There are cases out there where spine fusion is indicated. However, it is not indicated for back pain associated with degenerative changes, and it is not indicated for most spine fractures. Yet these are the indications used for most of the spine fusion surgery being performed today. Overall, it is a risky and expensive procedure that is being massively oversold.

      Delete
  12. Agreed re spinal surgery. Most chronic back pain has psychosocial cause. The sad thing, is when this IS recognized the sufferer has virtually no place to turn for help, as emotional cause for pain and CNS upset in other organs is not generally accepted by the modern medical establishment.

    ReplyDelete
  13. Doctor Sceptic what is your email? I am trying to use your sources for my paper on placebo surgeries and the ethics behind it yet I cannot find you as a credible source not knowing your degrees and such! Thank you!

    ReplyDelete
    Replies
    1. Ah, the "credible source"! So glad you brought it up. I have been asked about the reasons for my anonymity several times and taken some criticism for it. The reasons are manifold and include separation of my clinical practice, being able to unify my message with my name, and avoiding crazy people. But one of the biggest reasons is because I feel the evidence should speak for itself, and not rely on the eminence of the provider. I have many letters after my name, but there are people out there who know more, and understand the science better than me, but are without any letters after their name.

      Basically, I didn't want people to take my comments as true, based on who I am.

      Having said that, I have several people out there is cyberspace with whom I converse privately. If you send me a comment that includes your email, I will respond directly to you without publishing it and we can discuss this further.

      Delete
  14. I have a question on spine surgery outcome measurement

    Oswestry Low Back Pain Disability Questionnaire
    Beck Depression Inventory (BDI)
    Medical Outcomes Survey 36-Item Short-Form Health Survery (SF-36)
    McGill Pain Questionnaire
    Roland and Morris Disability Questionnaire

    These forms provide ordinal data (qualitative)
    In statistics, ordinal data is a statistical data type consisting of numerical scores that exist on an ordinal scale, i.e. an arbitrary numerical scale where the exact numerical quantity of a particular value has no significance beyond its ability to establish a ranking over a set of data points.

    How can the stern statistically significant improvement be used?

    ReplyDelete
    Replies
    1. Not sure what you mean by "stern statistical improvement". Also, ordinal data is not qualitative, it is quantitative. Even dichotomous outcomes (2 ordinal values) is quantitative.
      With scores on large scales, they can be treated as contintuous for statistical purposes, and tested for normality. If you don't like the idea of doing this, for several reasons, then like many, you can group them into excellent, good, fair and poor, or even satisfactory and unsatisfactory. But the bottom line is that these scores have been validated as being sensitive to change and reflective of the underlying phenomena that they aim to quantify.

      Delete
    2. I meant the "term statistical significance." typo. Sorry. Some of the research you quoted above calculated a mean. Means should not be calculated for ordinal data because the differences between a 4 and a 5 are not the same as the differences between a 10 and 11. From my reading a non parametric test should be used. Parametric testing may overstate outcomes. Do you agree?

      Delete
    3. Owestry Questionnaire- 10 items: pain intensity, personal care, lifting, walking, sitting, standing, sleeping, sex (if applicable), social, and travel. Each item consists of six statements correlating to scores of 0 through 5, with the patient choosing the statement that matches his or her ability. The statement correlating with a score of 0 indicates the least disability, and the statement correlating to 5 represents the greatest disability.

      Scores are calculated as follows: [total score/(5 x number of questions answered)] x 100%, falling within a range of 0 through 50

      Pain, sex, and disability are not additive. It is an ordinal scale and comparing means between non surgical and surgical groups should not be done.

      Delete
    4. Thank you for reponding to my question. "With scores on large scales, they can be treated as contintuous (interval) for statistical purposes, and tested for normality." Do you have evidence for this claim? I realize many ordinal scales are out there and have been demonstrated to correlate with each other and have fair reproducibility.

      Ordinal variables are obviously not normal. There's no point in testing what you know for certain a priori -- your data are not normal.

      the intervals between adjacent categories in an ordinal variable are arbitrary. It's not really meaningful to test for normality, since it depends on imposing an arbitrary scale choice.

      Delete
    5. Yet the analysis depends on a normal distribution, not on whether or not you think it is normal. You could argue that nothing is "normal" - that the difference between an age of 1 year and 2 is not the same as the difference between 91 and 92. And from my understanding, with enough numbers, parametric tests will be reliable.

      However, you are correct in that non-parametric tests should be used when you are in doubt and that they will inherently provide a more conservative estimate of statistical significance.

      But this is all predicated on the assumption that statistical significance (the p value) is important, which is arguable. If the study is of good quality (low risk of bias), it is the result of the study (the effect estimate, whether it be a mean, median, or proportion of patients getting x% better) that counts. If it is big enough for you to sit up and take (clinical) notice and is very unlikely to have occurred by chance (if the p value is less than 0.001, then knowing that it is 0.002 on a non-parametric test is not important) and if it is repeated in a separate experiment, then it is likely to be real.

      You might be less enamoured with p values and statistical testing of research findings if you view the Dance of the P Values: https://www.google.com.au/webhp?sourceid=chrome-instant&ion=1&espv=2&ie=UTF-8#q=dance%20of%20the%20p%20values

      Delete
  15. What is the MCID (minimum clinically important difference) for Owestry and is it valid? I have heard 15.

    ReplyDelete
    Replies
    1. Depends what you want to use it for. There are many studies on this. One reference is here: http://www.rehabmeasures.org/Lists/RehabMeasures/DispForm.aspx?ID=1114

      Delete
    2. Great article. Sounds like 10-15 for Owestry. 15 would make many spine intervention outcomes not clinically significant.

      Delete
  16. Thanks for the excellent posts Dr Skeptic. These posts are more valuable and informative than much of what I learned in medical school. Correct reasoning and observation are critical to knowing what truth is. I too believe spine surgery is largely a placebo. Reasons- Many spine findings are asymptomatic, disk osteophyte complex, neuroforamenal narrowing, spinal stenosis, degenerative disk disease. Biologic plausibility reigns supreme in spine care which is faulty logic- surgery corrects a deformity and alleviates pain (sounds reasonable but isn't true), many patients do not get better with surgery. There is no convincing evidence that surgery works- where sham surgeries and RCTs have been done no significant benefit is seen. Much lower quality evidence rests on ordinal data from patient surveys. The surgical group has a lower Owestry score or pain score. This means little if anything. Many patients in both surgery and non surgery groups still have pain.
    Another issue is risks versus benefits With time the harmful effects of surgery come to pass negating any so called benefit of surgery- I see many complications from implants including irritable fixation, infection, accelerated disease above and below the operated segment. The so called benefits if they are real decrease with time and pain scores tend to equalize between surgery and "conservatively treated" groups after a few years.

    ReplyDelete
    Replies
    1. If you like the blog, you will love the book: https://www.newsouthbooks.com.au/books/surgery-ultimate-placebo/#buy (or try Amazon)

      Delete
  17. It would be preaching to the choir! Perhaps we could get our government leaders to read it. I would say the public but marshalling the public is a fool's errand. It may well be that around a trillion dollars or more are lost every year in the US alone. This is a truly staggering cost.

    ReplyDelete
  18. Trillion dollars lost in total healthcare waste, not just spine- that would just be in the billions.

    ReplyDelete
  19. Patients understand biologic plausibility- there is a narrowing, the nerve is pinched- not complex statistical outcomes- randomized controlled trials with ordinal data Oswestry Low Back Pain Disability Questionnaire and pain scale outcomes. Statistical vs clinically significant effects, numbers needed to treat vs numbers needed to harm, p values, etc. I know of no surgeon who obtains informed consent using anything other than biologic plausibility while mentioning possible complications to include bleeding, infection, need for more surgery, nerve damage, etc. (the probability of these complications is not mentioned) Is your experience different? Informed consent is not really informed. At the end the patient often says doctor if the surgery will help lets do it, or do what you think is best, or if I was your mother would you do it (this last one is slightly more sophisticated.)

    ReplyDelete
    Replies
    1. I certainly use percentages in my discussions with patients, when they are known. However, there is a need for patients to be better informed of risks and benefits, which is why many groups are working on more detailed "decision tools" for certain procedures. A Cochrane review has shown that for some conditions, patients are less likely to choose surgery when better informed.

      Delete