“It is remarkable how little difference there is between the REGAIN randomization groups at 1 yr. All estimates, directionally, suggest we need not bend over backward to avoid general anesthesia. Any differences between techniques are small and unlikely to be meaningful to patients.”

Hip fracture is a life-changing event for older patients often followed by poor outcomes. Hip fracture surgery is also particularly amenable to neuraxial anesthesia, as well as general anesthesia. While tremendous strides in the safety of anesthesia care have occurred over the past few decades, identifying long-term survival benefits of one anesthesia type over another would be very important. Nonetheless, traditional endpoints like survival can feel a little thin when a plausible outcome is survival with unacceptable quality of life. Many older adults care more about functional and cognitive outcomes of treatment decisions than mortality. To capture “what matters,” we can and should study function and independence at these distant endpoints. The Regional versus General Anesthesia for Promoting Independence after Hip Fracture (REGAIN) trial was a forward-thinking international, multicenter, pragmatic, randomized trial of neuraxial versus general anesthesia for hip fracture repair with prespecified study of 1-yr outcomes that included both function and survival. The 1-yr results are reported in this issue of Anesthesiology. 

REGAIN’s pragmatic design yielded critical strengths for generalizability. Nearly a third were 85 yr and older, 15% presented with a status of do not resuscitate, 13% had dementia, and greater than 30% required an assistive device for mobility before their fracture. That the investigators enrolled such a highly impaired population is impressive, but this population is typical for hip fracture. Outcomes were selected in collaboration with an advisory board that included patients and care partners to come to a consensus on what matters. Follow-up was remarkably complete (approximately 90% for survival) and additionally explored with sophisticated mathematical methods. Prespecified subgroups—the oldest old, those with baseline dementia, and so forth—ensured that the role of anesthesia type was not different in those hypothesized to be at highest risk. Based on the methods and population, REGAIN’s results are very likely to be applicable to typical hip fracture patients and special populations and should carry sufficient weight to change practice. Thus, we are very excited to have been invited to editorialize these 1-yr findings which suggest…

…that even a major intervention on anesthetic type makes no real difference in long-term traditional or patient-centered outcomes. This comes as something of a shock, if we base our beliefs about anesthesia type on retrospective studies. Retrospective observational studies often find that spinal anesthesia is associated with better outcomes after hip fracture. The fundamental difference between REGAIN, a randomized trial, and retrospective observational work is in the balance of unmeasured confounding. Classic unmeasured confounders in this research area include preoperative cognitive status or delirium (too confused to safely undergo spinal; both delirium and baseline cognitive impairment are notoriously undercaptured by billing/administrative data), and hospital or anesthesiologist quality (those who were most “up on the literature” may have been spinal-ists, leaving their less proactive colleagues to provide general anesthesia along with omission of other evidence-based interventions).

It is remarkable how little difference there is between the REGAIN randomization groups at 1 yr. All estimates, directionally, suggest we need not bend over backward to avoid general anesthesia. Any differences between techniques are small and unlikely to be meaningful to patients. Thus, it should be a thrill to the practicing anesthesiologist to discover that, based on both short- and long-term outcomes, a well conducted anesthetic, whether general or spinal, may be selected according to patient indications and preferences.

Further, REGAIN obliquely demonstrated that patients may have strong preferences about the care they receive. Recall that 3,565 people declined consent before achieving REGAIN’s target n = 1,600 randomized participants. Reasons for declining included 950 who were unwilling to be randomized to spinal anesthesia (even when, at the time this trial was enrolling, spinal anesthesia seemed to be the emerging technique of choice), and 480 who were unwilling to be randomized to general anesthesia. Those two groups made up 40% of those who declined, nearly the size of the trial population itself. Despite the contemporaneous optimism around spinal, twice as many of these older patients explicitly preferred general anesthesia. Patients are, themselves, not agnostic to anesthetic strategy.

Failure of observational studies’ conclusions to replicate in randomized trials is not new. “Armchair Research and the Astigmatism of the Retrospectoscope: Caveat Lector” by John Drummond, M.D., F.R.C.P.C., published in Anesthesia and Analgesia in 2013 cautioned against overinterpretation of observed differences in retrospective data, and much has been written before and since, although perhaps not quite as compellingly titled. REGAIN and a similar trial, Regional Anesthesia vs General Anesthesia (RAGA), in which no sedation was used have offered strong evidence that confounding likely drove many of the findings of postoperative outcomes after neuraxial versus general anesthesia. While those observational studies’ conclusions were—to be frank—probably wrong, it remains valuable that the findings motivated large-scale, well conducted randomized trials. Newer theoretical underpinnings, such as “target trial emulation” designs first articulated in 2016 may help modern observational studies avoid sources of unmeasured confounding.

Older patients care about long-term functional outcomes as much as or more than mortality. Long-term functional and survival outcomes did not differ in this powerhouse of a trial, offering scientific equipoise between the two anesthesia types, but patients themselves may have strong feelings about what is preferred, and if refusal to participate in REGAIN was any indication, many prefer general anesthesia. Practicing anesthesiologists have all experienced the difficulty of positioning a frail patient with cognitive impairment for a safe spinal, perhaps lateral on the fractured side (as for a hyperbaric spinal), perhaps at 2 am. There are real problems with uncritically selecting spinal anesthesia as the “default” type: in REGAIN, 8% of spinal attempts were unsuccessful, placing the patient at risk of known complications of both spinal and general anesthesia when general could have been the technique of choice. Furthermore, those who underwent spinal reported more pain in the first 24 h and had higher rates of prescription opioid use at 60 days. Short- and long-term mortality and patient-centered outcomes do not differ or favor general anesthesia. For older patients with hip fracture, it seems likely that spinal anesthesia should no longer be the default technique, allowing those patients who prefer insensibility to receive care aligned with their values.