27 Sham vs. Vertebral Augmentation
Summary
This chapter discusses the role of medical management, placebo, sham, or active control in vertebral augmentation trials and the role of real-world evidence with new perspectives and conceptualization. Randomized controlled trials (RCTs) continue to be in the forefront of decision-making process for therapeutic interventions. At the same time, negative RCTs have appeared questioning the effectiveness of vertebral augmentation. The vast majority of the extensive literature on vertebral augmentation procedures including RCTs, systematic reviews, guidelines, and development of appropriateness criteria have yielded support for the procedure but some negative RCTs have produced some discordant conclusions.
Among all the control modalities, various effects of sham controls have been described often with negative connotations. Two trials published in the New England Journal of Medicine showing lack of efficacy of augmentation procedures provided negative results, with resultant lack of access to augmentation procedures for many patients who meet the criteria, indications, and medical necessity. Only one true placebo control (VAPOUR) study has yielded appropriate results with significantly better results in treatment group with vertebroplasty. In addition, multiple RCTs and systematic reviews have provided basis for effectiveness of vertebral augmentation in managing acute or subacute osteoporotic fractures.
Thus, the emerging literature on placebo, nocebo, sham intervention, and active intervention with multiple conceptual factors including confluence of interest and intellectual bias provide overwhelming evidence of irregularities in interpretation of review of scientific literature. At the same time, this chapter also provides a perspective on utilization of therapeutic placebo in vertebral compression fractures along with robust development of real-world evidence.
27.1 Introduction
“It is simply no longer possible to believe much of the clinical research that is published, or to rely on the judgment of trusted physicians or authoritative medical guidelines. I take no pleasure in this conclusion, which I reached slowly and reluctantly over my two decades as an editor of The New England Journal of Medicine.”
— Marcia Angell, Former editor-in-chief of the New England Journal of Medicine
Numerous changes in medical, political, and economic spheres in health care, clinical evaluation of therapies, and the importance of randomized controlled trials (RCTs) in evidence synthesis continue to be in forefront of decision-making process for therapeutic interventions. However, over the last 30 years, negative RCTs have become the rule rather than exception, specifically in managing spinal pain with interventional techniques including vertebral augmentation procedures. 1 , 2 Several attempts are made to overcome the present problem in evidence synthesis, based on various factors including understanding of the conduct of RCTs based on controlled design (active control vs. placebo control), outcomes assessments and implications of placebo and nocebo effects. 1 Beyond these factors, evidence synthesis has been done with discordant opinions due to intellectual bias, confluence of interest, and peer review bias, which extends beyond honest differences of professional opinions. 1 – 5 At the center of the controversy is a multitude of interventional techniques and surgical procedures including vertebral augmentation. Extensive literature on vertebral augmentation procedures in managing vertebral compression fractures (VCFs) has yielded not only discordant opinions, 6 – 19 but often emotional debate primarily initiated by two RCTs published in New England Journal of Medicine in 2009. 18 , 19 Despite numerous publications exceeding 3,000 manuscripts on vertebral augmentation, with multiple controlled trials 6 – 33 debate continues on whether it is effective and superior to sham intervention. 17 , 34 , 35 All of this has produced a significant decline in the utilization of vertebral augmentation procedures. 36 The present evidence indicates increased morbidity in patients treated with nonsurgical management (NSM) compared to those treated with vertebral augmentation. 8 – 37 , 38 In fact, a study by Ong et al 8 including more than 2 million patients showed 24% higher mortality risk for NSM compared to balloon kyphoplasty at 10 years and a 55% higher mortality rate at 1 year. In another study of over one million patients, with 4 years of follow-up, 9 the nonoperated patients also had a 55% higher propensity-adjusted mortality risk than balloon kyphoplasty patients and a 25% higher mortality risk than vertebroplasty patients. Similar results were also shown by McCullough et al, although McCullough attributed the longer term perceived benefits to artifact rather than actual benefit despite three of the four time points showing a significant mortality reduction. 10 , 37 , 38
Even then, patients in need of vertebral augmentation procedures are facing reduced access based on RCTs and sham response. Consequently, in this chapter, we will discuss the role of placebo control or sham in vertebral augmentation trials and the role of real-world evidence with new perspectives and conceptualization.
27.2 Current Concepts of Controls in Trials
In the study of interventional or surgical trials, various types of controls are utilized, ranging from no treatment group, placebo control, to sham interventions. In RCTs for surgical or other interventions, any of the above controls may be utilized, but in most RCTs, treatments are typically tested by comparing the efficacy in an active treatment arm versus the efficacy in a placebo arm. The failure to detect significant differences between active treatments and placebos is one of the main sources of uncertainty in the RCTs, especially as “negative” RCTs continue to explode across spinal therapeutics. In fact, the failure rates of the active treatments were one of the highest in musculoskeletal diseases. 2 Placebo response is most commonly provided explanation leading to a smaller difference between the effect of the intervention and placebo. The placebo-nocebo phenomenon is the subject of increasing debate, with extensive research often stoking the controversy. 39 A multitude of contextual factors of placebo and nocebo responses and definitions continue to evolve. 1
Multiple related issues exist in control trials including placebo and nocebo response, masking, blinding, statistical analysis, and finally interpretation of the results and their application to clinical settings. Placebo control trials of pharmacological treatments are typically conducted in double-blind trials. In these studies, the process of masking the treatment assignment is considered ethically acceptable, provided that shared decision-making was made and the consent process established the nature of the study. However, in circumstances where a surgical or interventional procedure itself constitutes the treatment, a randomized, placebo-control trial raises different issues. 40 In these settings, only the patient is blinded, whereas the clinician can distinguish active from inactive treatment. Thus, the gold standard of clinical research, namely the double-blind randomized placebo-control trial, is challenging to make applicable in interventional settings. Consequently, in surgical and interventional techniques including vertebral augmentation procedures, instead of a placebo, a “sham” procedure is utilized. However, sham procedure itself is not same as placebo and creates a multitude of other issues.
Historically, it was long considered that there is no place for placebo in surgery and often any type of sham intervention was considered unethical. 40 , 41 However, in 1939 an Italian surgeon, Davide Fieschi tried a new technique for the treatment of patients with angina pectoris. The treatment increased blood flow to the heart, thereby alleviating the symptoms of angina, and this treatment was done by diverting chest arterial supply to the heart by ligating the internal mammary arteries. Approximately 75% of the patients showed improvement and 25% were considered as cured. Twenty years later, in 1959, Cobb et al tested this procedure in the way that one group of patients was treated using the Fieschi technique and the other group of patients just had incisions imitating real procedure. The results were shocking with no difference between the two groups. 42
Similarly, Moseley et al 43 conducted a study to assess effectiveness of arthroscopy. In this trial, authors divided patients into three groups. In the first group, they peeled off the damaged cartilage, in the second group they washed the knee and that way removed possible causes of inflammation, and in the third group they made a false operation with only incisions. Results showed that there was no difference in the therapeutic success of the groups. A similar study by Kirkley et al 44 was also published in the New England Journal of Medicine by another group of investigators showing lack of additional benefit to arthroscopic debridement over NSM in patients with moderate-to-severe knee osteoarthritis. The results of these two publications in New England Journal of Medicine have resulted in extensive reductions in the utilization of arthroscopic procedures. 45 , 46
27.3 Sham vs. Vertebroplasty
27.3.1 Similarities and Differences of Controls
In drug trials, a placebo is administered in a double-blind manner along with the concealment of allocation. Consequently, a placebo response has been defined as “the reduction in symptoms as a result of factors related to patient’s perception of the therapeutic intervention.” 1 However, with development of multiple modes of placebo intervention, the placebo has also been defined as a “psychobiology phenomenon occurring in the patient’s brain after the administration of an inert substance, or of a sham physical treatment such as sham surgery, along with verbal suggestions (or any other cue of clinical benefit).” 1 , 47 , 48 Apart from placebo, there is also nocebo activity, which has been discussed extensively and is somewhat controversial. 1 , 49 The term was coined to denote negative repercussions from a treatment or a placebo, with description of worsening of the symptoms or reduction of the beneficial effect by the administration of an inactive or active treatment. 49 A patient’s negative expectations regarding the treatment or an untoward effect from the sham treatment is thought to produce the nocebo response.
Placebo has been administered in interventional trials without appropriate forethought by administering an assumed inert substance, which may not be an inert substance into an active structure, or by administration of active substances into so-called inert structures that may not be inert. In contrast, conservative management groups receive no blinding, sham, or active treatment, thus avoiding the placebo effect that occurs by administration of a substance. On the other hand, placebo effect in these patients may occur due to education or other clinical cues. In sham surgery, typically an incision is made and procedure is carried out until the final intervention. The true placebo nature of such interventions is questionable due to induction of multiple psychological phenomenon. 47 , 48 , 50 Consequently, placebo and sham interventions may be similar to active interventions. Thus, placebo or sham interventions in studying efficacy of vertebral augmentation procedures are met with multiple flaws. Before the sham trials of vertebral augmentation there were multiple failures of placebo interventions with other interventional techniques due to injecting sodium chloride solution into the epidural space and facet joints. 3 , 4 Due to the potential to dilute or wash out inflammatory mediators in these locations, the possibility existed of converting the sham intervention into an active treatment. There is overwhelming evidence of effectiveness of local anesthetics of epidural injections and facet-joint nerve blocks, similar to the effectiveness of local anesthetics with steroids, 3 , 4 , 51 but the effectiveness may not be sufficient to show a significant difference when compared to a de facto active treatment but rendered as sham. The same disadvantages are present in a multiplied format when sham intervention is compared to vertebral augmentation.
Even if a placebo is a completely inert substance, its administration is done within a complex psychosocial context and in conjunction with the rituals of the therapeutic act. Sham intervention exceeds the magnitude of the placebo intervention and is provided not only within a complex psychosocial context, but also within physical and therapeutic contexts.
There is significant confusion between the definitions of placebo and placebo effect for clinicians, methodologists, and scientists. A clinician is interested in any improvement that may take place in the group of patients who either take the inert substance or receive a sham treatment, and this improvement may be attributed to a multitude of factors including spontaneous remission, regression toward the mean, and patients’ expectation of the benefit. Contrary to the clinician, the scientist is typically more interested in the improvement that derives from the patients’ expectations, namely, an active process occurring in the patient’s brain. The methodologist is often mostly interested in assessing the difference in the statistical results without consideration of the clinical or scientific aspects.
In vertebral augmentation trials, a sham intervention is provided with injection of local anesthetic into the skin, followed by stab incisions at the level of the vertebral body, and local infiltration of periosteum and the pedicle with periosteal placement of bone needles. The difference between the augmentation group and the sham group is that augmentation group patients undergo needle placement via a transpedicular approach along with placement of the cement, and the sham patients undergo the first part of the procedure but do not have the cement injected. 18 , 19 , 33
27.3.2 Sham vs. Vertebral Augmentation
In vertebral augmentation trials the simplest design is the one in which patients are prospectively followed after the intervention. This is followed by comparison of patients allocated into two groups, one with no treatment and the other group having undergone vertebral augmentation. These patients may have implied placebo effects because of contextual factors. The next category is the management of the patients with nonsurgical treatment that may also induce placebo effects. The third treatment group is patients receiving a placebo treatment with local infiltration of the skin with local anesthetic followed by a skin incision. This technique will induce both placebo and, to a lesser extent, nocebo effects. In contrast, sham intervention as defined originally by the NEJM trials is the procedure that is much more involved and includes the entire procedure with the exception of the placement of the cement. The rationale for this seeming unusual approach was originally intended to isolate the beneficial effect of PMMA.
27.4 Evidence Based on the Control and the Conduct of the Study
A systematic review performed by Hulme et al 6 in 2006 included a total of 69 clinical studies of which 25 were prospective. Overall, the results showed pain relief in a large proportion of patients in 87% with vertebroplasty and 92% with kyphoplasty.
In contrast, the evidence based on individual RCTs with sham controls including the 2009 NEJM articles as well as the 2018 VERTOS IV study 18 , 19 , 33 showed no significant difference between sham control and vertebroplasty. Apart from sham control being not a placebo treatment, it is very similar to an active control treatment. In fact, facet-joint injections as control group have shown significant effectiveness in decreasing patient’s pain after osteoporotic VCFs at 1 month and 1 year after fracture in patients with severe pain due to vertebral fracture. 23 An author of the INVEST trial, Dr. David Wilson, published results in the European Journal of Radiology that reported 34% of patients with painful VCFs had immediate pain relief after facet-joint injections, and out of 75 patients included in his trial, 61 had facet injections. 52 There were 29 patients who failed facet injections and 24 of these patients underwent vertebroplasty. This study, along with the randomized control trial by Wang et al, shows the effectiveness of facet-joint injections in controlling the pain from VCFs. Compared to facet-joint injections, the sham intervention seen in the NEJM articles and VERTOS IV exceeds a facet-joint injection with a blockade of periosteum as well as the medial branch of the dorsal ramus. Depending on the volume of anesthetic injected (which was not controlled), the dorsal nerve root ganglion and ventral nerve root can be affected. Additionally, these studies utilizing sham control also have faced significant criticism. 17 , 35 Chandra et al 17 and Noonan 35 have eloquently described multiple flaws related to the initial two RCTs published in New England Journal of Medicine. 18 , 19 Apart from various issues related to the recruiting techniques, there were other shortcomings such as a low enrollment rate, selection bias, including worker’s compensation patients, not having uniform diagnostic criteria for fractures, and not having a requirement for physical exam. There was a significantly higher rate of crossover within 3 months from the control group. Both studies were also supposed to be blinded, but the INVEST trial blinding was disrupted by patients receiving procedural bills following their vertebroplasty treatment. 18 Notably, there was also a trend toward a higher rate of clinically meaningful improvement in pain, with 30% reduction in the vertebroplasty group (64% vs. 48%) (p = 0.6). There were also a high proportion of patients with worker’s compensation claims. Due to low recruitment, they resorted to a pain level of 3 as the criterion to be included, which is uncharacteristic of patients with painful VCFs. Sample size was very low even though the authors thought it to be enough patients to produce sufficient power to show a short-term treatment advantage. The final statistical analysis just barely showed any statistically significant difference between vertebroplasty and sham. If the same response rate were present for the originally intended 250 patients instead of the 131 patients that were enrolled, the p value would have been 0.01 in favor of vertebroplasty. Also, if a single patient had had a different response favoring vertebroplasty, the p value would have been 0.04, again a statistically significant result favoring vertebroplasty. The study by Buchbinder et al 19 also was met with multiple deficiencies, not the least of which was that the mean amount of cement injected in the vertebroplasty group was 2.8 mL. This amount was analyzed and determined to be an insufficient amount by Boszczyk et al who found that it “strongly indicates that the treatment arm includes patient who were not treated in a reasonably effective manner.” 53 The VERTOS IV study provided similar results with similar issues related to the conduct of the study. Even though both groups showed no significant difference, pain scales in vertebroplasty group declined to a mean of 2.72 (2.18 to 3.26) from a baseline of 7.72 (7.21 to 8.24) with difference between baseline and 12 months of 5 (4.31 to 5.70) points which is typical of the pain reduction seen with vertebroplasty and is significant compared to the baseline. The sham intervention that they employed for this study, however, also profoundly reduced the pain levels from 7.9 to 3.17 with a prominent reduction of 4.75; thus, the vertebroplasty and the so-called sham treatments showed no significant difference. This lack of difference was not because the vertebroplasty difference was ineffective but because the sham response was profound with a pain decrease magnitude similar to that of very effective active pain treatments. When the VERTOS IV results are compared with the treatment and sham or NSM arms of other trials, the pain reduction curves clearly align with active treatments and treatments that reflect the natural course of the VCF (▶Fig. 27.1 and ▶Fig. 27.2).
These differences cannot be explained on pure placebo response and are not seen in conservative management groups, nontreatment groups, or even in interventional placebo groups with local anesthetic infiltration and incision as shown in the VAPOUR study. 24 In fact, the VAPOUR study 24 utilized a placebo design without sham intervention as we have described above. In this trial, placebo procedure included subcutaneous lidocaine but not periosteal numbing. In addition, only a short needle was passed into the skin incision, but not as far as the periosteum. The results of this trial showed 24 or 44% of the patients in the vertebroplasty group and 12 or 21% in the control group had an NRS pain score below 4 out of 10 at 14 days. Authors of this study concluded that vertebroplasty was superior to placebo intervention for pain reduction in patients with acute osteoporotic spinal fractures of less than 6 weeks duration. Results of this study at 6 months showed that proportion of patients with NRS pain score less than 4 was 69% in the vertebroplasty group and 47% in the control group. Analgesic use declined from 97% to 58% in vertebroplasty group and from 98% to 76% in the placebo group. The placebo used in the VAPOUR trial was representative of the typical placebo response instead of the active treatment response seen in the so-called sham arm of the VERTOS IV trial.
The next category with less rigorous placebo design is the augmentation compared to conservative treatment. Multiple RCTs were performed comparing augmentation with medical management. 26 – 32 These studies showed significant improvement with vertebroplasty and vertebral augmentation compared to NSM. Studies comparing kyphoplasty and vertebroplasty also showed similar outcomes. 21 , 22
Related posts:
4 Physical Examination Findings in Patients with Vertebral Compression Fractures
13 Radiation Exposure and Protection: A Conversation Beyond the Inverse Square Law, Thermoluminescent Dosimeters, and Lead Aprons
16 Cost-Effectiveness of Vertebral Augmentation
14 Appropriateness Criteria for Vertebral Augmentation
25 Postural Fatigue Syndrome
30 Vertebral Augmentation in Instrumented Spinal Fusions
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
