Introduction: Phantom limb pain is a common complication in people who have undergone limb amputation, with prevalence estimates ranging between 29% and 85.6%. Current systematic-review evidence suggests that recommended treatments are no more effective than placebo for reducing Phantom Limb Pain (PLP). Moreover, there is evidence suggesting that people with amputations may not be getting the treatment they want at different time-points after amputation. In consideration of these points, a research project comprised of a series of interconnected studies aimed to develop an in-depth understanding of the global burden of PLP and patient care priorities after limb amputations, and generate expert recommendations on the best management of PLP in people with amputations. Methods: The research project is comprised of a series of four interconnected studies addressing the four primary aims of the project. A systematic review and meta-analysis were conducted to determine the pooled prevalence estimate and risk factors for PLP in people with amputations. A cross sectional study was conducted to determine the prevalence and risk factors for PLP in people who had undergone lower limb amputations at Groote Schuur Hospital. An expert Delphi study was conducted to reach expert consensus and make recommendations on the effective treatments for PLP in people with limb amputations. Lastly, a patient Delphi study was conducted to generate patient consensus on care priorities for people who have had lower limb amputation for a year or less and for those who have had lower limb amputations for more than a year. Results: The systematic review and meta-analysis of 39 studies revealed a pooled PLP prevalence estimate of 64% [95%CI: 60.01 – 68.05], with a significantly higher prevalence estimate in studies conducted in developed countries 66.55% [95% CI: 62.02 –71.64] than those conducted in developing countries 53.98% [95% CI: 44.79–63.05] (U = 57, p = 0.03). Risk factors that were consistently positively associated with PLP included having an amputation of a lower limb, stump pain, non-painful phantom sensations, persistent pre-amputation pain, proximal site of amputation, and diabetic cause of amputation. The cross-sectional study using a sample of African people with amputations showed a PLP prevalence of 50.78% [95% CI: 41.80 – 59.72] during the week preceding data collection. In this group of patients, persistent pre-operative pain was the only risk factor associated with PLP in the multivariate logistic regression analysis [OR 2.25 (1.03 – 5.05); P=0.04]. In the expert Delphi study, consensus was reached on one pharmacological (amitriptyline) and six nonpharmacological (Graded Motor Imagery, mirror therapy, Cognitive Behavioural Therapy, virtual reality training, sensory discrimination training, use of a functional prosthesis) treatments that were considered effective for managing PLP, and on two treatments [citalopram (60%) and Pulsed Radiofrequency Stimulation of the dorsal root ganglion (70%)] that were considered ineffective. In the patient Delphi study, consensus was reached on 24 short-term care priorities and 12 long-term care priorities. The general consensus among the participants was that pre-amputation, they wanted education support to help them manage their expectations and prepare for life after amputation. In the early stage after amputation, they wanted help with dealing with the psychological trauma of having lost a limb. In the long-term, however, the participants prioritised the need for living a functional and normal life, with respect and dignity like everyone else. Conclusion: The prevalence of PLP in people with limb amputations is high, and awareness of this condition needs to be raised among healthcare professionals to implement evidence-based strategies for alleviating PLP by targeting the relevant underlying mechanisms and modifiable risk factors. Evidence-based medicine indicates that PLP is best managed using non-pharmacological and noninterventional treatments addressing biopsychosocial contributors for PLP. Finally, preparing people for life after amputation and helping them deal with the psychological trauma of having lost a limb may contribute to improved clinical outcomes that may enable them to live a functional and normal life, with respect and dignity.

Background: There is no established costing model for operating theatres in South Africa, yet both sectors have existing charges for operating theatre (OT) time: in the state sector, Uniform Patient Fee Schedule (UPFS) rates, and in the private sector, Rands/minute (R/min) rates for OT time. Understanding the cost of providing the separate components of a health service is important for planning and funding purposes. Objective: The primary objective of this study was to develop a costing model that would allow the calculation of the R/min cost of OT time. The secondary objective was to determine the actual costs, in order to establish the comparable costs that would be included in the R/min charges for OTs in the private health sector. Method: The OTs in a secondary level, state sector hospital in Cape Town were used in this quantitative observational study to develop a top-down costing model for OTs in South Africa. The inclusive costing model was developed in a consultative process with professionals, managers and experts from the state and private sector. The model was then populated with utility measurements (water and electricity) for the month of August 2018, staff salaries, excluding surgeons and anaesthetists, and other costs for the 2018/19 financial year. Results: Costs were considered in the categories of full costs, shared costs and capital or annualised costs. Due to uncertainty in costing of OTs, two models - with different annualisation times assigned to the capital costs - were developed to demonstrate the difference. For shared costs, correction factors were determined using either an activity based (work-load) factor, or a more generic estimation of workload using theatre nursing staff as a percentage of total hospital nursing staff. To determine a R/min cost of creating a minute of available theatre time, all the annual costs were divided by minutes that the OTs are explicitly available, each year, to provide patient care. The model was then populated with costs using the appropriate correction factors. The longer annualisation model costed OT time at R31,46 per minute, and the shorter annualisation model at R33,77 per minute. In both the longer and shorter capital annualisation models, nursing was the largest contributor to costs at 36% and 33% respectively, followed by construction costs at 9% and 11%, and then OT equipment at 8% and 11%. Conclusion: An inclusive, top-down costing model for OTs in South Africa was developed. This costing model will support work to develop costing for individual procedures, the appropriate charge for planned and emergency OT time, and to better determine budgeting for OT services. Meaningful critique of the model will improve its fidelity, and likely increase its utility, especially as SA moves towards universal health coverage.

Introduction Chronic pain affects 10- 25% of the population worldwide. However, studies of people with chronic pain have primarily been conducted in WEIRD (Western, Educated, Industrialised, Rich and Democratic) countries. There is a paucity of data from developing countries with the existing biased data being used to guide treatment of patients in developing countries. To address this knowledge gap, we have analysed the data of 623 patients attending the Chronic Pain Management Clinic (CPMC), at a tertiary facility in Cape Town, South Africa. The profile of the patients was compared to the global data. Methods A cross-sectional study of patients who attended the CPMC was conducted to describe their profile according to age, sociodemographic characteristics, health, gender and use of illicit drugs. Results The typical patient attending the CPMC was of middle age, female gender, of low educational level and less advantageous socio-economic status. They were also more likely to be on a disability grant and suffer from anxiety and depression. Conclusion The profile of the patients attending the CPMC was found to be similar to those from WEIRD countries. These results suggest that we can apply global data of people with chronic pain to patients attending this clinic.

Objective: In this study of patients presenting for non-cardiac, emergency surgery in a resource limited setting, we aimed to evaluate the impact of routine preoperative transthoracic echocardiography on perioperative management. Design: A prospective before- and after-study of adult patients presenting for emergency, non-cardiac, non-obstetric surgery. Setting: The study was performed at an academic hospital in Cape Town, South Africa. Participants: Consenting patients over 18 years of age presenting for emergency surgery enrolled via convenience sampling during working hours over a 10 day period. Interventions: Basic and advanced Focused Assessment Transthoracic Echocardiography (FATE) was performed to evaluate ventricular function, valvular pathology and fluid status. After completing an assessment and treatment plan, the FATE findings were disclosed to the treating anesthetist. A post FATE plan was subsequently completed. Measurements and Main Results: A total of 67 patients were scanned with a change in management detected in 55% of cases. Thirty-nine percent of these alterations were in response to fluid management strategies with 31% of patients scanned being assessed as hypovolemic. There was a statistically significant link between patient volume status and change in perioperative management (p=0.0003). The presence or absence of valvular pathology also led to a significant association with change in management (p=0.020), most commonly in relation to the decision to proceed with surgery or the use of additional monitoring. Conclusion: This observational study of adult patients presenting for emergency surgery in an upper middle-income country demonstrates that routine preoperative transthoracic echocardiography has an impact on perioperative anesthetic management.

The patient-important outcomes of cardiac and non-cardiac surgery are well-recognized but poorly understood. The causes of major morbidity and mortality in patients undergoing non-cardiac are not known. This is not the case in cardiac surgery, which is provided to a homogenous patient population that has been well-described through clinical registries. Recent improvements to the care of cardiac surgical patients have led to dramatic decreases in major morbidity and mortality. However, neurocognitive and functional impairments after cardiac surgery remain the most feared by patients and least understood by clinicians. This thesis comprises 6 chapters that inform these knowledge gaps and establish the basis upon which future research will be based. Chapter 1 is an introduction providing the rationale for conducting each of the included studies. Chapter 2 reports the VISION Mortality study, which explores the relationship between major complications and death within 30-days of undergoing inpatient, noncardiac surgery. Chapter 3 reports a study validating the use of the Standardized Assessment of Global activities in the Elderly (SAGE) scale in patients undergoing cardiac surgery. Chapter 4 presents a pilot observational study that establishes the feasibility of conducting a large, prospective cohort study to determine the relationship between decreases in cerebral saturation during cardiac surgery and postoperative functional decline. Chapter 5 presents a pilot study conducted to inform the feasibility of a large, randomized cluster crossover trial examining whether an institutional policy of restricted benzodiazepine administration during cardiac surgery (compared to liberal administration) would reduce delirium after cardiac surgery. Chapter 6 discusses the conclusions, limitations, and implications of the research presented in this PhD thesis. / Thesis / Candidate in Philosophy

This dissertation uses population-based administrative healthcare data to evaluate the outcomes, processes-of-care and practice variation associated with preoperative medical consultation in Ontario, Canada. First, a multicentre cross-sectional study was conducted to develop a novel algorithm for identifying preoperative medical consultations using administrative data. The optimal claims-based algorithm was a physician service claim for a consultation by a cardiologist, general internist, endocrinologist, geriatrician, or nephrologist within 120 days before the index surgery. This algorithm had a sensitivity of 90% (95% confidence interval [CI], 86 to 93) and specificity of 92% (95% CI, 88 to 95). Second, we conducted a population-based cohort study to evaluate the association of preoperative medical consultation with outcomes and processes-of-care. After adjustment for measured confounders using propensity-score methods, consultation was associated with increased preoperative testing, preoperative pharmacological interventions, 30-day mortality [relative risk (RR) 1.16; 95% CI, 1.07 to 1.25], 1-year mortality (RR 1.08; 95% CI, 1.04 to 1.12), and mean hospital stay (difference 0.67 days; 95% CI, 0.59 to 0.76). These findings were stable across subgroups, as well as sensitivity analyses that tested for unmeasured confounding. Third, temporal trends and practice variation in consultation were evaluated within the population-based cohort. The proportion of patients undergoing consultation remained relatively stable over the study period, at approximately 39%. Although patient-level and surgery-level factors did predict consultation use, they explained only 6.8% of variation in consultation rates. By comparison, inter-hospital differences in rates were substantial (range, 1.9% to 86.8%), were not explained by surgical volume or teaching status, and persisted after adjustment for patient-level and surgery-level factors. Overall, this dissertation highlights the need for research to identify interventions for safely decreasing perioperative risk, define mechanisms by which consultation influences outcomes, examine factors that influence practice variation in medical consultation, and identify patients who benefit most from preoperative medical consultation.

This dissertation uses population-based administrative healthcare data to evaluate the outcomes, processes-of-care and practice variation associated with preoperative medical consultation in Ontario, Canada. First, a multicentre cross-sectional study was conducted to develop a novel algorithm for identifying preoperative medical consultations using administrative data. The optimal claims-based algorithm was a physician service claim for a consultation by a cardiologist, general internist, endocrinologist, geriatrician, or nephrologist within 120 days before the index surgery. This algorithm had a sensitivity of 90% (95% confidence interval [CI], 86 to 93) and specificity of 92% (95% CI, 88 to 95). Second, we conducted a population-based cohort study to evaluate the association of preoperative medical consultation with outcomes and processes-of-care. After adjustment for measured confounders using propensity-score methods, consultation was associated with increased preoperative testing, preoperative pharmacological interventions, 30-day mortality [relative risk (RR) 1.16; 95% CI, 1.07 to 1.25], 1-year mortality (RR 1.08; 95% CI, 1.04 to 1.12), and mean hospital stay (difference 0.67 days; 95% CI, 0.59 to 0.76). These findings were stable across subgroups, as well as sensitivity analyses that tested for unmeasured confounding. Third, temporal trends and practice variation in consultation were evaluated within the population-based cohort. The proportion of patients undergoing consultation remained relatively stable over the study period, at approximately 39%. Although patient-level and surgery-level factors did predict consultation use, they explained only 6.8% of variation in consultation rates. By comparison, inter-hospital differences in rates were substantial (range, 1.9% to 86.8%), were not explained by surgical volume or teaching status, and persisted after adjustment for patient-level and surgery-level factors. Overall, this dissertation highlights the need for research to identify interventions for safely decreasing perioperative risk, define mechanisms by which consultation influences outcomes, examine factors that influence practice variation in medical consultation, and identify patients who benefit most from preoperative medical consultation.