Nonphysician Sedation Providers in Africa: What Counts and What Is Being Counted?

See Article, page 250 “Not everything that can be counted counts, and not everything that counts can be counted.” WB Cameron, Informal Sociology: A Casual Introduction to Sociological Thinking, 1963. Does your anesthesia providers’ level of training impact your outcomes? This question has been widely evaluated and debated in the perioperative literature. With increasing demand for surgical and procedural services facilitated by anesthesia care globally, an answer will continue to be sought. Van der Merwe et al1 in their article “Postoperative outcomes associated with procedural sedation conducted by physician and non-physician anesthesia providers: findings from the prospective, observational African Surgical Outcomes Study (ASOS)” published in this month’s Anesthesia & Analgesia, have added to this discussion, with a secondary analysis of data from the African Surgical Outcomes Study (ASOS). Although their study provides some interesting insights into the outcomes of procedural sedation across the continent, our opinion is that the question remains largely unanswered. To date, most of the literature evaluating the association between anesthesia care provider type and outcomes has focused on anesthesia care in highly developed health care systems. Questions have focused on task-shifting, where the responsibility for tasks is shifted from a more highly trained health care provider to health workers with shorter training and fewer qualifications, and task-sharing, where both levels of providers perform the task and may even work closely together. Examples include family doctors in Canada providing unsupervised anesthesia care in community hospitals after adding an additional year of training in anesthesia to their family medicine residency program; certified registered nurse anesthetists (CRNAs), practicing independently in many US states; and French anesthesiologists supervising nurse anesthetists with a 1:2 ratio. Ultimately, the hope is that by shifting/sharing tasks, access to care will improve with less-resource input and with similar (or in the case of task-sharing) even safer outcomes.2 Countries with a gross national income per capita of <$12,696 US dollars (USDs) are often (problematically) lumped together as low- and middle-income countries (LMICs)3 regardless of the profound diversity in this categorization, which contains around 85% of the world’s population.4 There is a critical shortage in human resources for health (HRH) globally, particularly in anesthesia. However, HRH are one of the most complex parts of health systems, with huge international variation in terms of numbers of health care workers, their training, their point of entry into training, their scope of practice, interprofessionalism, resilience, burnout, and retention of health care workers within the system.5–7 Developing a deep understanding of how to most effectively and efficiently provide safe anesthesia care is an urgent priority in improving global surgical outcomes; however, nuances in context make generalizations problematic. Ven der Merwe et al1 aimed to evaluate this question by comparing patient outcomes when procedural sedation was delivered by nonphysician versus physician anesthesia providers. The primary data source, the ASOS, is a landmark study, where investigators collected a large amount of data (11,422 patients) over a relatively short amount of time, with good coverage of a broad geographic area.8 Its largely descriptive statistical analysis has been highly informative of perioperative outcomes in Africa, which appear to be much worse than previously published global data. In contrast, the Van der Merwe et al1 study is a small subset of the primary data (336 patients, ~3% of the full cohort), with a more complex comparative statistical analysis, with the authors concluding that receipt of sedation from a nonphysician provider was significantly associated with increased odds of severe complications. While these results must be interpreted with great caution (as we will outline below), the findings raise important questions about perioperative health care systems in Africa. Van der Merwe et al1’s observational study was designed to answer a question of causation: does receiving care from a nonphysician anesthesia provider cause patients to have more complications? However, when considering findings from any observational study aiming to make inferences on causation, many important factors must be considered. First, one must acknowledge that causal questions ultimately require randomized trial data to provide high certainty that a cause-and-effect relationship exists. Next, one must acknowledge that randomized trials are hard to perform, are resource-intensive, and require specialized infrastructure. The conduct of such a trial would be challenging anywhere in the world,9 but particularly so in the poorest countries. Ironically, this question is arguably most important in the poorest countries because of an increased frequency of harm directly attributable to anesthesia.8,10 Observational studies, especially secondary data analyses studies, are invariably easier to conduct than randomized trials. However, observational studies must be designed considering what a related trial would look like, which Hernán and Robin11 refer to as the Target Trial Framework. In the Target Trial Framework, a key requirement is providing a robust counterfactual comparison. A counterfactual can be thought of as what would have happened to the patients in the absence of the intervention, and in the case of van der Merwe et al,1 we must be confident; the only important differentiating feature between exposed and unexposed patients was the level of training of their provider. Key components of observational design, like valid ascertainment of exposure and outcome status, and meaningful capture of patient-level covariates, are present in the current study and are generally strengths of this study. The exposure of interest is somewhat problematic in that the level of training across the continent varies hugely, particularly in nonphysicians6 and it is not clear whether having a medical degree is the most important factor in “level of training” in anesthesia. However, the foundational issue of finding a valid counterfactual for van der Merwe et al1 appears to be especially challenging in the African context. Reviewing their Supplemental Table (S2), which we summarize here in Table 1, we must acknowledge that most cases delivered by nonphysicians were from just 6 countries that included no physician-delivered sedation cases at all. As most African countries that tend to rely on nonphysician anesthetists have very low number of anesthesiologists and, in turn, these countries tend to have less robust health systems and infrastructure, there is a substantial risk that much of the effect of physician versus nonphysician provider could instead be attributable to the very different health systems present across African countries, which is noted by the authors. One would expect that regardless of provider credentialing, someone receiving care in a less robust health care system would be at greater risk of severe complications. Ultimately, when comparing the outcome of a patient in a well-resourced system under the care of a physician provider with the outcome of a patient in a poorly resourced country with a nonphysician provider, the system could easily confound outcomes, be it through preoperative processes that prepare the patient less effectively for surgery, the availability of essential drugs and equipment, the intraprocedural monitoring, or a multitude of aspects of postoperative care, and, across all phases, the supporting perioperative team. Table 1. - Countries in the Procedural Sedation African Surgical Outcomes Study Substudy, Grouped by Cases Included, Whether the Anesthesia Provider Was a Physician or Nonphysician Categorization of country groups Number of physician delivered cases Number of nonphysician delivered cases Total number of cases included Cases with complications Country R 125 (54%) 3 (3%) 128 (39%) 3 (12%) Countries that included only nonphysicians a 0 (0%) 56 (57%) 56 (17%) 4 (25%) Countries that included only physicians b 41 (18%) 0 (0%) 41 (12%) 1 (6%) Countries that included both physicians and nonphysicians c 67 (29%) 39 (40%) 106 (32%) 8 (50%) aConsisted of 6 countries: B, C, D, F, Q, and S.bConsisted of 4 countries: A, J, M, and N.cConsisted of 10 countries: E, G, H, I, K, L, O, P, T, and U. Moving forward, how could such an obstacle be overcome? Using Hernán and Robin11’s Target Trial criteria, the most important strategy is likely to ensure that patients compared should both be eligible for the same type of care. Strategies, especially pertinent in observational studies of provider type, could include directly match patients with different anesthesia providers within the same hospital, health care system, or at a minimum, country. Other strategies to help appraise van der Merwe et al1’s findings are to consider them in the context of knowledge already in hand. The current study, like most, used a frequentist approach to significance testing, in other words, drawing conclusions based on expectations of what would be expected of these data were the study to be repeated many, many times where we truly believed there to be no difference (the null hypothesis). However, this does not preclude us from thinking about the results probabilistically or in a Bayesian framework, that is, assigning a probability to study findings based on preexisting knowledge interpreted in concert with study data. In other words, conclusions reached would be based on both what our prior knowledge of the subject was, combined with the data from the study. Let us provide 2 applications of this thinking to van der Merwe et al1’s results. First, although data are lacking about provider types and outcomes in LMICs in general and Africa, specifically, robust analyses of US data, such as those from Sun et al,12 suggest no differences in outcomes based on anesthesia care team composition.13 This prior knowledge could reduce our certainty that the results found in the current study are as strong as reported (albeit, and somewhat ironically, with limitations given the very different health systems under consideration). Next, we could consider the plausibility of the results based on postulated mechanisms underlying the association. Of the 15 patients with the primary outcome, 12 had an infectious postoperative complication. While anesthesia care can influence the risk of infectious complications (for example, through intravenous line contamination, temperature management, administration of prophylactic antibiotics, and airway management), the potential causes of infectious complications are numerous, and the extent to which they are mediated by anesthesia care for sedation-only procedures possibly limited. In contrast, were the majority of adverse outcomes serious cardiorespiratory events, the likelihood of a true relationship might seem more plausible. The statement in the abstract of this article that “shifting the task of procedural sedation from physicians to nonphysicians to increase access to health care may be associated with severe postoperative complications and death in Africa” is problematic, particularly considering the existing polarized, political, and opinion-based nature of the discussion of nonphysician anesthesia providers in many jurisdictions. It must be remembered that there is good evidence from high-income countries that anesthesia delivered by nonphysicians can be safe. There is a potential for great harm from avoiding nonphysician anesthesia care particularly in the poorest countries in Sub-Saharan Africa where it is unlikely that anesthesiologists will be available to provide universal access to surgical care for the foreseeable future.14 The Van der Merwe et al1 study is another sobering reminder that outcomes in patients who have been anesthetized or sedated in Africa continue to be unacceptably poor and that there continues to be unacceptable variation in outcome according to the health care system. We strongly agree with the authors’ statement that “research focusing on identifying factors contributing to adverse outcomes associated with procedural sedation is necessary to make this practice safer.” This data also support the idea that there needs to be investment across whole health care systems, of which anesthesia is an essential part. This should include the effective recruitment, training, credentialing, and maintenance of skills of anesthesia providers, regardless of their background, as well as investment in initiatives to improve career progression and retention of these health care providers and to encourage rural practice, where there are particularly few physician anesthesia providers. Physician resistance must be addressed, and this should include involving anesthesiologists in both planning and implementation of task shifting/sharing, avoiding conflict through clear definitions of scope of practice, and considering time-limited local pilots.15 We must also acknowledge the vested interest that anesthesiologists have in avoiding task shifting/sharing: they may have understandable concerns about a potential loss of their own income and status due to task-shifting, which may threaten the sustainability of this already underdeveloped professional group in some African countries. Ultimately, faced with a desperate shortage of anesthesia providers across most of Africa and much evidence that nonphysician anesthesia provision can be safe within high functioning health care systems, the question should not be whether we should use nonphysician anesthetists, but how we can ensure that their practice is safe. DISCLOSURES Name: Yohannes Molla Asemu. Contribution: This author helped write and edit the manuscript. Conflicts of Interest: Y. M. Asemu is a nonphysician anesthetist. Name: Etienne Nsereko. Contribution: This author helped write and edit the manuscript. Conflicts of Interest: E. Nsereko is a nonphysician anesthetist and a Senior Lecturer at the University of Rwanda, with oversight for the training of nurse anesthetists. Name: D. I. McIsaac. Contribution: This author helped write and edit the manuscript. Conflicts of Interest: D. McIsaac is a physician anesthesiologist. Name: M. Dylan Bould. Contribution: This author helped write and edit the manuscript. Conflicts of Interest: M. D. Bould is a physician anesthesiologist and is the immediate past-chair of the Canadian Anesthesiologists Society International Education Foundation, a charity with a primary focus on supporting the training of physician anesthesia providers globally. This manuscript was handled by: Angela Enright, MB, FRCPC.