The New England Journal of Medicine (NEJM) has served as a bellwether publication, publishing influential research that heralds in the future of evidence-based practice. In this OE Insight, we will review 11 of the most well-cited orthopedic surgical related randomized controlled trials (RCTs) published in the NEJM over the last decade and explore what they say about 6 important topics. In reviewing these works, where appropriate, we will further report on notable points of contention, as evidenced through correspondence and editorials. 

Bone and joint infections represent amongst the most significant issues within orthopedic practice. Infection constitutes a serious surgical complication which can be difficult to treat and can lead to significant morbidity and mortality. Moreover, as outlined in our recent Insight, the rise of antibiotic resistance, and the need for antibiotic stewardship heightens the need to elucidate optimal anti-microbial strategies. Helping to address this need are two trials in the NEJM. (1,2)

The Oral versus Intravenous Antibiotics for Bone and Joint Infection (OVIVA) trial aimed to challenge the current consensus that complex bone and joint infections require the intravenous administration of antibiotics to effectively fight infection. Supported by a meta-analysis showing little difference between oral and intravenous (IV) in the treatment of chronic osteomyelitis (4), Li et al. (2019) aimed to determine whether oral antibiotics were non-inferior to IV antibiotics when treating bone and joint infections. They conducted a large multicenter randomized open-label non-inferiority trial of 1054 patients comparing the effectiveness of oral antibiotics to IV antibiotics, given within the first 6-weeks of treatment, on definitive treatment failure of osteomyelitis at 1-year. With comparable rates of treatment failure, the primary end-point analysis showed that 6-weeks of initial oral administration was non-inferior to IV antibiotics. 

In an attempt to better steward the use of antibiotics, clinical guidelines have recommended the switch to oral antibiotics where possible (5). This is because intravenous antibiotics, in general, come with the risk of injection-site complications (e.g., infection), and increased length of hospital stay – factors which increase the risk to exposure for drug-resistant bacteria (5,6). Moreover, oral antibiotics have the advantage of reducing patient costs, and relieving hospital resources (i.e., oral antibiotics do not need to be administered by health care professionals) (6). In accordance with these benefits, Li et al. indeed showed that oral antibiotic administration reduced the incidence of catheter-associated complications and median hospital stay. Thus, this work suggests that it would be suitable and indeed appropriate to switch patients to oral antibiotics early on when treating osteomyelitis. Despite the important implications of this finding, as pointed out by Helen Boucher in their editorial, “The limitations included their open-label design (mitigated by use of adjudication committees who were not aware of the treatment assignments), multiple pathogens and treatment regimens (single agent and combination therapies), heterogeneous populations (bone and joint infections with or without hardware, variable surgical interventions…”. In short, in creating a study to mimic the real-world clinical setting, the OVIVA trial introduces a significant degree of complexity, and it is for this reason that Boucher states, “Further studies are needed to confirm these findings and will further inform these strategies and advance stewardship to decrease antimicrobial resistance.”(3) 

One of the highest risk injuries faced by orthopedic surgeons are open fractures, which have a high risk of complications including infection and non-union. A key step for reducing this risk is ensuring effective wound irrigation and debridement. However, despite its importance there remains controversy surrounding an optimal irrigation solution and application pressure. High-pressure irrigation, while thought to clear the wound of debris and bacteria, may instead do the opposite – increase the risk for infection and delay fracture healing (8)(9). Soap is assumed to serve as an effective antiseptic for irrigation, although there is perhaps only limited evidence to support this intuition (7,10). Thus, to identify an optimal irrigation, the Fluid Lavage of Open Wounds (FLOW) investigators conducted a 2 x 3 factorial randomized clinical trial of 2,447 patients comparing the impacts of castile soap versus normal saline irrigation delivered at high, low, or very low irrigation pressures on re-operation risk. Notably, Bhandari et al. found that irrigation pressure had no impact on re-operation, indicating that very low pressure represents a viable low-cost option. Moreover, Bhandari et al. found that a lavage of saline reduced the risk of re-operation when compared to castile soap. Beyond these primary outcomes, through the correspondence, the FLOW trial yielded a third important insight. Precipitated by Parham et al. concerned by the usage of vacuum-assisted closure as a possible source of infection, Bhandari et al. provided additional data supporting vacuum-assisted closure as a risk factor of infection. Despite this, the patients who received vacuum assistance were balanced between treatment groups, and thus did not impact the conclusions of the study (11).

Overall, as illustrated by Dr. Douglass Dirsch in their commentary, “the FLOW trial represents the most substantial contribution to medical knowledge [on the irrigation pressure and solution] to date.” 

 

An outstanding question in the field of orthopedics is determining the optimal treatment for the treatment of spondylolisthesis. Prior to the studies by Ghogawala et al. (2016) and Försth et al. (2016), the most widely cited evidence comparing decompression to decompression plus fusion was a 1991 prospective observational study which showed that decompression plus fusion should be preferred to decompression alone for the treatment of degenerative spondylolisthesis (12). Indeed, according, to Ghogawala et al., in 2016 “to date, there is no class I evidence that laminectomy plus fusion is superior to laminectomy alone for the treatment of degenerative spondylolisthesis” (13). By contrast, a growing body of work was starting to suggest the opposite – that decompression alone was just as effective as decompression plus fusion (14). Under this context, in 2016 the NEJM published two dueling studies (13,15) reporting somewhat disparate conclusions about the efficacy of decompression plus fusion when compared to decompression alone. 

In a randomized study of 66 patients with single-level grade I spondylolisthesis, excluding those with spinal instability, the Spinal Laminectomy versus Instrumented Pedicle Screw (SLIP) trial by Ghogawala et al. (2016) showed that decompression plus fusion improved physical-component summary score, as measured by the Study 36-Item Short-Form Health Survey (SF-36), compared to decompression alone for up to 4-years of follow-up. They further showed that fusion plus decompression had nearly resulted in half the levels of re-operation compared to decompression alone. There was no difference between treatments on the Oswestry Disability Index (ODI) score. These benefits came at a cost, however, as fusion surgery increased the length of hospital stays, and blood-loss. While the authors contend that fusion affords a clinically meaningful improvement for patients with stable spondylolisthesis, the concede it might not be the best default approach, writing, “The well-established higher hospital costs of lumbar fusion may suggest that an overall value assessment might favor decompression alone” (13).

Contrasting this view was the Swedish Spinal Stenosis Study (SSSS) by Försth et al. (2016) who in a randomized trial of 247 patients with and without spondylolisthesis found no difference between decompression alone, and decompression plus fusion in the ODI score or a 6-minute walk test at 2-years follow-up. Furthermore, no difference in re-operation rates was seen. Follow-up at 5 years again found no significant differences between the decompression and the decompression plus fusion groups. By contrast, as with Ghogawala et al., decompression plus fusion increased operating times, blood-loss, and mean hospitalization time (15). 

Reconciling these somewhat disparate results has been difficult with conflicting opinions on the implications and validity of the SLIP and SSSS trials (14). Some have cited differences in regional practices to account for the differing conclusions on re-operation rates. In their commentary, Peul et al. write, “The higher rate of reoperation in the decompression-alone group in the trial by Ghogawala et al. probably reflects decision making about revision surgery in the United States. Patients who have unfavorable outcomes after laminectomy are perhaps offered revision with added instrumented fusion more often in the United States than in Sweden. In contrast, in the United States, physicians have higher thresholds to perform revision surgery after fusion surgery, resulting in lower rates of reoperation than those in the Swedish trial.” (14) 

In responding to these differing results, Försth and colleagues stated, “The current concept of instability as it relates to the selection of the type of surgery is problematic. … There are certainly patients with spinal stenosis with instability who will have better outcomes if fusion is performed as an adjunct to decompression. The actual problem is that we currently do not have firm scientific support to identify these patients.”. In responding to the disagreement, Ghogawala et al. stated, “The major difference between the SLIP study and the SSSS was the patient population. The SSSS included a heterogeneous population of patients with 1 to 2 levels of disease, whereas the SLIP study focused entirely on single-level grade I spondylolisthesis. [Notably], neither trial was appropriately powered to detect a between-group difference in the disease-specific Oswestry Disability Index outcome tool among patients with single-level grade I spondylolisthesis.” It should also be noted that Ghogawala et al., strongly disagree with the editorial by Peul and Moojen, who suggest that lumbar fusion does not add value for patients with spondylolisthesis. (14)

 

Most recently, Austevoll et al. (2021) have weighed in on the debate with the Norwegian Degenerative Spondylolisthesis and Spinal Stenosis (NORDSTEN-DS) trial that aimed to evaluate whether decompression alone was non-inferior to decompression plus fusion (17). Patients with a single-level spondylolisthesis were analyzed, and the primary outcome was the number of patients with a 30% reduction in ODI score. After 2 years of follow-up, they showed that decompression alone was non-inferior to decompression plus fusion in terms of ODI score. Re-operation rates were similar between the evaluated groups. Given that patient entry criteria and study design differed between the SLIP study, and SSSS this data does not directly re-appraise the results of SLIP. Nevertheless, the authors conclude, “Our results were in accordance with the findings of [the SSSS], which involved 133 patients with spondylolisthesis at one or two levels but did not include information about dynamic motion.” (17)

Thus, it would seem the debate surrounding the effectiveness of fusion for spondylolisthesis has not conclusively settled. Nevertheless, perhaps a tenuous compromise suggests that a subset of patients – those who may be at high risk of instability – if they could be identified, may benefit from fusion in addition to decompression. By contrast, most patients may be better served by decompression alone, which is generally a less costly, less complicated, and less risky procedure.

In 2013, the NEJM published two studies focused on clarifying the optimal treatments for degenerative meniscal tears. Previous RCTs published in the NEJM established that arthroscopic surgery was not effective in treating knee osteoarthritis (18,19). Despite this, a common sequelae of knee osteoarthritis is symptomatic meniscal tear, a feature which is typically treated through arthroscopic surgery. Indeed, between 1996 and 2006, despite a dramatic decrease in the rate of knee arthroscopy for treating knee osteoarthritis, use of arthroscopy for treating meniscal tears substantially increased (20). 

The high prevalence of arthroscopic partial meniscectomies led Katz et al. (2013) to ask, even though knee arthroscopy was no better than non-surgical options in treating knee osteoarthritis, could it still be effective in treating knee osteoarthritis related degenerative meniscal tears? Thus, the Meniscal Tear in Osteoarthritis Research (METEOR) trial randomized 351 mild to moderate knee osteoarthritis patients with degenerative meniscal tear to receive physical therapy or arthroscopic partial meniscectomy (22). Knee pain and function were evaluated up to 12 months post-treatment. At all time-points evaluated, there was no significant difference in function between arthroscopic partial meniscectomy and physical therapy in Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) function or Knee Injury and Osteoarthritis Outcome Score (KOOS) pain scores. Thus Katz et al. suggest that arthroscopic partial meniscectomy may be unnecessary and ineffective in knee osteoarthritis patients even for treating degenerative meniscal tears. 

It should be noted that while these results hold for “average” knee osteoarthritis patients with a meniscal tear, there may be specific cases where this doesn’t apply. Several commentors of the article expressed reservations, pointing out how in their experience, patients with mechanical symptoms, including locking, popping, and catching, often realize the most benefit from arthroscopy. This point is encapsulated by Dr. James Strite Jr., who in a comment suggests, “There is a small subset of the knee OA group that clearly benefits from arthroscopic partial meniscectomy. This subset is patients who have mechanical symptoms (catching, locking, giving-way, etc.) secondary to their meniscal tears. You must pay careful attention to all parameters (history, exam, imaging studies) to figure out who they are.” (22) The notion that some patient subsets may yet benefit from surgery may be somewhat encapsulated by the studies intervention crossover population. Illustrated in the commentary by Dr. Rachelle Buchbinder, “Although the subgroup of participants in the current study who crossed over to surgery from the nonoperative group appeared to have worse outcomes before crossover, it is reassuring that by 12 months, their functional outcomes were similar to those of patients who had initial surgery and patients in the control group who did not cross over. This suggests that nonoperative treatment is a reasonable first strategy, with surgery reserved for the minority who do not have improvement.” (21)

Given that the METEOR trial showed that arthroscopic partial meniscectomy was not preferable in patients with knee osteoarthritis, Sihvonen et al. (2013) asked whether, “arthroscopic partial meniscectomy is effective under ‘ideal’ circumstances” (23). To do so, they conducted the Finnish Degenerative Meniscal Lesion Study (FIDELITY), a multicenter RCT of 146 patients comparing partial meniscectomy to sham surgery in patients with symptoms of degenerative meniscal tear without a clinical diagnosis of osteoarthritis (23). At the 12-months follow-up they observed no significant difference between partial meniscectomy patients and placebo controls on Western Ontario Meniscal Evaluation Tool (WOMET) score, and pain after exercise. A subsequent 5-year follow-up analysis showed similar conclusions, indicating no differences between knee function and pain after exercise (24). To the contrary, they noted that arthroscopic partial meniscectomy slightly increased the risk for mechanical symptoms and the risk for developing osteoarthritis (24). Based on this data they concluded that arthroscopic partial medial meniscectomy provides no clinical benefit over sham surgery for treating non-osteoarthritic degenerative meniscal tear.

For some, this conclusion requires some qualifications. In a letter, Krych et al. (2013) noted that, “the declaration of ‘no knee osteoarthritis’ is misleading”, pointing to the fact that most patients in the trial had degenerative or osteoarthritic changes on diagnostic arthroscopy (25). Krych et al. further point out that the efficacy of arthroscopic partial meniscectomy was not tested within patients who experienced traumatic tears or mechanical symptoms, a population highly likely to benefit from surgery. They, alongside others suggest that arthroscopy is likely effective in treating the meniscal tears for a subset of patients. In response, Sihvonen et al. (2013) state, “We disagree that our description of ‘no knee osteoarthritis’ was misleading, since arthroscopically determined cartilage degeneration is not considered osteoarthritis by any currently used clinical criteria.” (25) They further contend, their study adequately included patients with traumatic tears and mechanical symptoms, stating, “Although true traumatic onset of symptoms was an exclusion criterion, in 16% of our patients the onset was sudden or after twisting, and 47% had preoperative mechanical symptoms.” (25)

Ultimately, the work by Katz et al. and Sihbonen et al. show that for many patients with knee osteoarthritis, or degeneration which could lead to knee osteoarthritis, arthroscopic surgery for treating meniscal tears may not be the preferred treatment.

While it has emerged how arthroscopy may not be effective in treating knee osteoarthritis, the question remains, what are effective treatments? Given the lack of evidence for knee arthroscopy, the prevalence of non-surgical treatments for knee osteoarthritis has seen a steady increase in popularity (27). Nevertheless, total knee arthroplasty remains the option of choice for treating cases of severe osteoarthritis (28,29). Despite this, prior to the Medicine Exercise Diet Insoles Cognitive (MEDIC) study by Skou et al. (2015) there was a dearth of high-quality RCT data to support the efficacy of total knee arthroplasty (30). In this context, Skou et al. asked whether total knee replacement could be shown to offer superior clinical outcomes for knee osteoarthritis when compared to non-surgical options. The MEDIC study randomized 100 patients to receive surgery or not. Both groups received 12 weeks of 5-arm non-surgical treatment program provided by trained physiotherapists and dieticians. At 1-year follow-up, the total-knee replacement group showed remarkable improvement compared to the non-surgical intervention group across every facet of the KOOS4 (i.e., pain, symptoms, activities of daily living, quality of life, sports and recreation), and mobility tests. 

Comments on the work suggest that Skou et al. may in fact be underestimating the efficacy of total knee replacement, as patients with severe pain (visual analog pain > 60mm) were excluded. Indeed, in a letter on the work, Teuscher et al. (2016) state “… we are concerned that the exclusion of 117 of 244 otherwise eligible patients (48%) because of severity of symptoms may have led to substantial underestimation of the effect sizes of treatments in both groups, especially in the surgical group” (31). Skou et al. (2016) counter this notion by suggesting their baseline clinical characteristics are highly consistent with previous observation studies, a result which supports it’s generalizability (31). A second contention surrounds the reporting of adverse events.  There were significantly more non-surgically related adverse events (e.g., cancers) in the surgical group compared to the non-surgical group (16 vs. 5), a result which, as reported in a comment by Branden Tarlow are, “much higher than reported in the United States which limits the generalizability of this study … [and] may reflect pre-existing differences in the patient population.” In response Skou et al. indicate the high numbers of non-surgical adverse events are a result of greater supervision of the attending physicians than might typically occur for registry data. Thus, rather than a weakness, this reflects a strength in diligence of the study.

Ultimately, the work by Skou et al. provides high-quality data to cement total knee arthroplasty as the preferred last-line option for knee osteoarthritis, while also clarifying its risks. For many, this serves as an important resource for informing clinical decision making (26).

While recent estimates suggest that total knee replacements can reliably last 25 years or more (33). it is nevertheless conventional wisdom that total knee replacements should generally be reserved until osteoarthritis has progressed to late-stage disease. Thus, determining optimal nonoperative strategies for managing osteoarthritis symptoms is an important and active area of research. Two common options include glucocorticoid injection and physical therapy. In recent clinical guidelines, both glucocorticoid injections, and subsets of physical therapy have received a strong recommendation as a first-line option (34). Despite this, in the US, usage of physical therapy has declined (35), and corticosteroid injection is generally more popular (36). Thus, Deyle et al. (2020) aimed to compare the efficacy between physical therapy and corticoid injection in the treatment of knee osteoarthritis (37). In their trial, 156 patients with definitive osteoarthritis were randomly assigned to receive either physical therapy or glucocorticoids. As measured by the WOMAC score, at 1-year follow-up, physical therapy reduced pain and improved function when compared to glucocorticoid injection. 

The results of this trial raise important implications, and seemingly serve to contradict some recent guidelines against the use of manual therapy (34,38). In exploring this difference, in their commentary Bennell and Hunter (2020) write, “few published trials on manual therapy have generally been of low quality and inconclusive, which probably accounts for the aforementioned recommendations of some medical and research societies against manual therapy for osteoarthritis of the knee.” (32) However, there is concern that the work by Deyle et al. may not fully resolve this discrepancy. Bennell and Hunter write, “the physical therapy program was individualized and included therapist-applied manual techniques combined with home exercises, all of which were based on the clinical judgement of the therapists. … therefore, replication of the findings in the trial may be difficult.” (32)

Overall the MEDIC trial illustrates that physical therapy represents an under-recognized and strong first-line option for managing the symptoms of knee osteoarthritis. 

Hip fractures are serious injuries, and are accompanied with significant morbidities, and a high risk of mortality. It remains debated whether displaced femoral neck fractures should be treated via total hip arthroplasty or hemiarthroplasty. To help answer this question, the Hip Fracture Evaluation with Alternatives of Total Hip Arthroplasty versus Hemi-Arthroplasty (HEALTH) trial randomized 1495 patients across 80 centers to evaluate the risk of reoperation, serious adverse events, and hip function after 2-years of follow-up (34). At 2-years, no significant differences in the rates of revision surgeries were observed, although there was a trend for a greater risk of dislocation for total hip arthroplasty. Moreover, while total hip arthroplasty showed slightly better WOMAC pain, stiffness, function, and total scores, none of these differences exceeded the threshold for a minimal clinically important difference. These results have largely been replicated by a recent meta-analysis of RCTs (40). 

The implications of the HEALTH trial, as discussed in a concurrent editorial by Dr. Jan-Erik Gjertsen, indicate that hemiarthroplasty, as a simpler procedure, may be preferable to total hip arthroplasty for many (39). Arguing against this conclusion, is the view expressed through correspondence, that 2-years follow-up may be insufficient to realize the totality of differences between treatments (41). In response, the HEALTH trial authors write, “[in a meta-analysis of] 16 trials involved a total of 3076 patients and follow-up ranges of 1 to 5 years. Meta-regression analysis suggests that length of follow-up has no significant effect on revision rates or functional outcome.” (41)

Overall, the HEALTH trial provides compelling data to indicate parity between hemi and total arthroplasty for hip fracture, and commences a larger discussion to reshape current standards of clinical practice.

It has long been known that sciatica caused by acute herniation of the lumbar disk can be effectively treated with conservative care (43). Indeed, randomized trials comparing surgery to conservative care for acute sciatica show no benefit of surgery (44). However, it remains unclear whether surgery can benefit patients with chronic sciatica. To answer this question, Bailey et al. (2020) randomized 128 patients with persistent sciatica (i.e., lasting 4-12 months) to receive either microdiscectomy or 6 months of standardized non-operative care, and followed them for 1 year(45). The primary analysis showed that patients who received surgery experienced a significant reduction in leg-pain intensity at 6 months. Secondary analyses measuring pain, function, patient satisfaction, and employment at 6 and 12 months post-treatment indicated the superiority of surgery. However, given the lack of a pre-specified plan to adjust for multiple comparisons, the statistical significance for secondary analyses could not be evaluated. 

Regardless, the implications of this work are profound. Previous work has shown that delaying treatment beyond 6 months increased the risk for poorer clinical outcomes (46,47). Under this context, Bailey et al. show that microdiscectomy represents a viable alternative. Nevertheless, Schoenfeld and Kang in their editorial write, “the trial did not account specifically for the effect of symptom duration in this window or other clinical factors known to influence outcome after diskectomy, such as the size of the disk herniation or the extent of nerve-root compression.” In this way, Schoenfeld and Kang suggest additional analyses are needed to identify whether patient subpopulations exist who might or might not respond to surgery (42).

Overall, Bailey et al. have provided much needed evidence identifying surgery as a viable treatment option for chronic sciatica, this is sure to catalyze future work to identify specific characteristics of patients who may respond better or worse to persistent sciatica. 

Randomized clinical trials are the backbone of evidence-based medicine and clinical practice. Here we reviewed 11 trials that are amongst the most highly cited general orthopedics studies in the New England Journal of Medicine. While, it’s important to acknowledge that any given trial cannot be considered definitive, every trial has to make design choices, and suffers from limitations which will inevitably be challenged. Nevertheless, well-constructed trials, such as discussed in this insight, can serve to challenge conventional thinking on practice. In this way, the RCTs discussed in this trial should not be considered the end of the story, but rather, a call for new ideas, and a catalyst to explore these important questions more deeply.