George C Velmahos

Massachusetts General Hospital, Boston, Massachusetts, United States

Are you George C Velmahos?

Claim your profile

Publications (465)1409.86 Total impact

  • [Show abstract] [Hide abstract]
    ABSTRACT: On April 15, 2013, two improvised explosive devices (IEDs) exploded at the Boston Marathon and 264 patients were treated at 26 hospitals in the aftermath. Despite the extent of injuries sustained by victims, there was no subsequent mortality for those treated in hospitals. Leadership decisions and actions in major trauma centers were a critical factor in this response. The objective of this investigation was to describe and characterize organizational dynamics and leadership themes immediately after the bombings by utilizing a novel structured sequential qualitative approach consisting of a focus group followed by subsequent detailed interviews and combined expert analysis. Across physician leaders representing 7 hospitals, several leadership and management themes emerged from our analysis: communications and volunteer surges, flexibility, the challenge of technology, and command versus collaboration. Disasters provide a distinctive context in which to study the robustness and resilience of response systems. Therefore, in the aftermath of a large-scale crisis, every effort should be invested in forming a coalition and collecting critical lessons so they can be shared and incorporated into best practices and preparations. Novel communication strategies, flexible leadership structures, and improved information systems will be necessary to reduce morbidity and mortality during future events. (Disaster Med Public Health Preparedness. 2015;0:1-7).
    Disaster Medicine and Public Health Preparedness 06/2015; DOI:10.1017/dmp.2015.42 · 1.14 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: : Graduated Drivers Licensing (GDL) programs phase in driving privileges for teenagers. In 2007, Massachusetts implemented a stricter version of the 1998 GDL law, with increased fines and education. This study evaluated the impact of the law on motor vehicle crash (MVC)-related health care utilization and charges.
  • Haytham M A Kaafarani, George C Velmahos
    World Journal of Surgery 06/2015; DOI:10.1007/s00268-015-3119-0 · 2.35 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: Fulminant Clostridium difficile colitis (fCDC) is a highly lethal disease with mortality rates ranging between 12% and 80%. Although often these patients require a total abdominal colectomy (TAC) with ileostomy, there is no established management protocol for post-operative antibiotics. In this study we aim to make some recommendations for post-operative antibiotic usage, while describing the practice across different institutions. Multi-institutional retrospective case series including fCDC patients who underwent a TAC between January 1, 2007, and June 30, 2012. We first analyzed the complete cohort and consecutively performed a survivor analysis, comparing different antibiotic regimens. Additionally we stratified by time interval (antibiotics for ≤7 d, or ≥8 d). Primary outcome was in-hospital mortality. Additional secondary outcomes included hospital length of stay (HLOS), ICU LOS, number of ventilator-free days, and occurrence of intra-abdominal complications (proctitis, abscess, sepsis, etc.). A total of 100 fCDC patients that underwent a TAC were included across five institutions. Four different antibiotic regimens were compared; A (metronidazole IV+vancomycin PO), B (metronidazole IV), C (metronidazole IV+vanco PO and PR), and D (metronidazole IV+vancomycin PR). The combination of IV metronidazole with or without PO vancomycin showed superior outcomes in terms of a shorter ICU length of stay and more ventilator-free days. However, when comparing metronidazole alone vs. metronidazole and any combination of vancomycin, no significant differences were found. Neither the addition of vancomycin enema, nor the time interval changed outcomes. Patients, after a TAC for fCDC, may be placed on either IV metronidazole or PO vancomycin depending upon local antibiograms, and proctitis may be treated with the addition of a vancomycin enema (PR). There was no data to support routine treatment of more than 7 d.
    Surgical Infections 06/2015; DOI:10.1089/sur.2013.153 · 1.72 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: A negative computed tomographic (CT) scan may be used to rule out cervical spine (c-spine) injury after trauma. Loss of lordosis (LOL) is frequently found as the only CT abnormality. We investigated whether LOL should preclude c-spine clearance.
    The American Journal of Surgery 06/2015; DOI:10.1016/j.amjsurg.2015.04.008 · 2.41 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: Little evidence currently exists regarding the clinical or financial impact of intraoperative adverse events (iAEs). We sought to study the additional health care charges attributable to the occurrence of an iAE. The administrative and ACS-NSQIP databases at our tertiary academic medical center were linked for all patients undergoing abdominal surgery (January 2007-October 2012). The ICD-9-CM-based Patient Safety Indicator "accidental puncture/laceration" was used to screen the linked database for potential iAEs. All iAEs were confirmed subsequently through standardized review of all flagged medical records. Multivariate analyses controlling for demographics, comorbidities/laboratory values, procedure type, and approach and complexity of surgery were performed to assess the increase in health care charges independently predicted by the occurrence of iAEs. Of 9,111 patients, 183 were confirmed to have iAEs. Patients in the iAE group had higher median total charges ($27,169 [IQR, 17,302-44,952] vs $13,312 [IQR, 8,586-22,012]; P < .001), direct charges ($17,808 [IQR, 11,520-28,930] vs $8,738 [IQR, 5,686-14,227]; P < .001) and indirect charges ($9,396 [IQR, 5,932-16,144] vs $4,568 [IQR, 2,887-7,824]; P < .001) when compared with patients without iAEs. Multivariate analyses demonstrated that iAEs independently predict an increase in total hospitalization charges by 41% (95% CI, 30-52%; P < .001). Specifically, the direct, indirect, operating room, laboratory/radiology, and alimentation/medical therapy charges increased by 42, 39, 27, 54, and 48%, respectively (all P < .001). In addition to the morbidity incurred by patients, the occurrence of an iAE is associated with major additional health care charges. In an era of value-based health care, understanding and preventing iAEs can lead to major cost savings alongside improvements in patient safety and surgical quality. Copyright © 2015 Elsevier Inc. All rights reserved.
    Surgery 05/2015; DOI:10.1016/j.surg.2015.04.023 · 3.11 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: Macronutrient deficit in the surgical intensive care unit (ICU) is associated with worse in-hospital outcomes. We hypothesized that increased caloric and protein deficit is also associated with a lower likelihood of discharge to home vs transfer to a rehabilitation or skilled nursing facility. Adult surgical ICU patients receiving >72 hours of enteral nutrition (EN) between March 2012 and May 2014 were included. Patients with absolute contraindications to EN, <72-hour ICU stay, moribund state, EN prior to surgical ICU admission, or previous ICU admission within the same hospital stay were excluded. Subjects were dichotomized by cumulative caloric (<6000 vs ≥6000 kcal) and protein deficit (<300 vs ≥300 g). Baseline characteristics and outcomes were compared using Wilcoxon rank and χ(2) tests. To test the association of macronutrient deficit with discharge destination (home vs other), we performed a logistic regression analysis, controlling for plausible confounders. In total, 213 individuals were included. Nineteen percent in the low-caloric deficit group were discharged home compared with 6% in the high-caloric deficit group (P = .02). Age, body mass index (BMI), Acute Physiology and Chronic Health Evaluation II (APACHE II), and initiation of EN were not significantly different between groups. On logistic regression, adjusting for BMI and APACHE II score, the high-caloric and protein-deficit groups were less likely to be discharged home (odds ratio [OR], 0.28; 95% confidence interval [CI], 0.08-0.96; P = .04 and OR, 0.29; 95% CI, 0.0-0.89, P = .03, respectively). In surgical ICU patients, inadequate macronutrient delivery is associated with lower rates of discharge to home. Improved nutrition delivery may lead to better clinical outcomes after critical illness. © 2015 American Society for Parenteral and Enteral Nutrition.
    Journal of Parenteral and Enteral Nutrition 04/2015; DOI:10.1177/0148607115585142 · 3.14 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: Heart rate complexity, commonly described as a “new vital sign,” has shown promise in predicting injury severity, but its use in clinical practice is not yet widely adopted. We previously demonstrated the ability of this noninvasive technology to predict lifesaving interventions (LSIs) in trauma patients. This study was conducted to prospectively evaluate the utility of real-time, automated, noninvasive, instantaneous sample entropy (SampEn) analysis to predict the need for an LSI in a trauma alert population presenting with normal vital signs.
    Journal of critical care 03/2015; 30(4). DOI:10.1016/j.jcrc.2015.03.018 · 2.19 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: Computed tomography angiography (CTA) has been increasingly used in traumatic brain injury (TBI) patients to uncover vascular lesions that might have preceded the trauma and caused the bleed. This study aims to evaluate the usefulness of head CTA in the initial care of blunt TBI patients. We conducted a retrospective case-control analysis of adult TBI patients, admitted to our Level I trauma center from January 1, 2012 to December 31, 2012. The patients were grouped as those with and without a CTA of the head. The primary outcomes included a change in management after the findings of head CTA and secondary outcomes included rate of admission to the ICU, ICU length of stay, hospital length of stay, discharge disposition, and mortality. Six hundred adult patients had blunt TBI and underwent head CT as a part of their evaluation. Of these 600 patients, 132 (22%) underwent head CTA in addition to CT. Only one patient had altered management after the CTA results; the patient had a diagnostic angiogram that was negative. Ninety-eight patients did not have any additional findings on CTA. Of the remaining 33 patients with additional CTA findings, 12 had incidental vascular malformations, which showed no acute pathology and were not related to the injury. In the matched comparisons, patients with CTA had a longer hospital stay, higher rate of ICU admission, and longer ICU stay. There was no significant difference in mortality and discharge disposition between the 2 groups. Head CTA is commonly used after blunt TBI but does not alter management and should be abandoned in the absence of clear indications. Copyright © 2015 American College of Surgeons. Published by Elsevier Inc. All rights reserved.
    Journal of the American College of Surgeons 03/2015; 220(6). DOI:10.1016/j.jamcollsurg.2015.03.002 · 4.45 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: Benchmarking the quality of intraoperative care by comparing the rates of intraoperative adverse events (iAEs) necessitates adequate risk-adjustment. We sought to identify the patient- and procedure-related risk factors for iAEs.
    Journal of the American College of Surgeons 03/2015; DOI:10.1016/j.jamcollsurg.2015.03.045 · 4.45 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: Graduated driving licensing (GDL) programs phase in driving privileges for teenagers. We aimed to evaluate the effect of the 2007 GDL law on the incidence of total motor vehicle crashes (tMVCs) and fatal motor vehicle crashes (fMVCs) among teenagers in Massachusetts. The Fatality Analysis and Reporting System, the Missouri Census Data Center, and the Massachusetts Department of Transportation databases were all used to create and compare the incidence of tMVCs and fMVCs before (2002-2006) and after (2007-2011) the law enactment. The following three driver age groups were studied: 16 years to 17 years (evaluating the law effect), 18 years to 20 years (evaluating the sustainability of the effect), and 25 years to 29 years (control group). As a sensitivity analysis, we compared the incidence rates per population and per licenses issued. tMVCs decreased following the law for all three age groups (16-17 years, from 7.6 to 4.8 per 1,000 people, p < 0.0001; 18-20 years, from 8.5 to 6.4 per 1,000 people, p < 0.0001; 25-29 years, from 6.2 to 5.2 per 1,000 people, p < 0.0001), but the percentage decrease in tMVC rates was less in the control group (37%, 25%, and 15%, respectively; both p's < 0.0001). The rates of fMVC also decreased in the age groups of 16 years to 17 years (from 14.0 to 8.6 per 100,000 people, p = 0.0006), 18 years to 20 years (from 21.2 to 13.7 per 100,000 people, p < 0.0001), and 25 years to 29 years (from 14.4 to 11.0 per 100,000 people, p < 0.0001). All of these results were confirmed in the sensitivity analyses. The 2007 Massachusetts GDL was associated with a decreased incidence of teenager tMVCs and fMVCs, and the effect was sustainable. This study provides further support to develop, implement, enforce, and maintain GDL programs aimed at preventing MVCs and their related mortality in the young novice driver population. Epidemiologic/prognostic study, level III.
    Journal of Trauma and Acute Care Surgery 02/2015; 78(2):265-271. DOI:10.1097/TA.0000000000000512 · 1.97 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: Translation of evidence to practice regarding adherence to published guidelines for transfusion of red blood cells (RBCs) in critically ill patients is sometimes suboptimal. We sought to use a multimodal intervention founded on peer-to-peer feedback and monthly audit to increase adherence to restrictive RBC transfusion guidelines. We conducted a prospective interventional study with a preintervention and postintervention comparison in our tertiary care center. For the 6-month preintervention period (January 1, 2013, to June 31, 2013) and the 6-month postintervention period (October 1, 2013, to March 31, 2014), all RBCs transfused in the surgical intensive care unit (SICU) were evaluated for pretransfusion hemoglobin (Hgb) trigger (TRIG). During the intervention, if stable low-risk patients were transfused outside of restrictive guidelines, the clinicians received e-mail notification and education from a surgeon colleague within 72 hours of transfusion. The mean TRIG, percentage of transfusions with TRIG greater than 8.0 g/dL, and rate of overtransfusion (posttransfusion Hgb > 10) were compared before and after intervention. For stable, low-risk patients, mean TRIG decreased from 7.6 g/dL to 7.1 g/dL (p < 0.001) and percentage of transfusions with TRIG greater than 8.0 g/dL decreased from 25% to 2% (p < 0.001) The overtransfusion rate decreased from 11%to 3% (p = 0.001). Total 6-month transfusions decreased from 284 U to 181 U, a 36% decrease. There were no significant differences in median SICU or hospital lengths of stay. Although SICU discharge Hgb and hospital discharge Hgb were significantly lower in the intervention period (8.4 vs. 8.6 [p = 0.037] and 8.6 vs. 9.0 [p = 0.003]), 30-day readmission and mortality rates were similar. A blood management program based on peer e-mail feedback was effective in improving adherence to guideline recommendations for transfusion of RBCs in stable, low-risk SICU patients. Therapeutic/care management study, level IV.
    01/2015; 79(1):65-70. DOI:10.1097/TA.0000000000000683
  • [Show abstract] [Hide abstract]
    ABSTRACT: To date, no studies have reported nationwide adoption of acute care surgery (ACS) or identified structural and/or process variations for the care of emergency general surgery (EGS) patients within such models. We surveyed surgeons responsible for EGS coverage at University Health Systems Consortium hospitals using an eight-page postal/e-mail questionnaire querying respondents on hospital and EGS structure/process measures. Survey responses were analyzed using descriptive statistics, univariate comparisons, and multivariable regression models. Of 319 potential respondents, 258 (81%) completed the surveys. A total of 81 hospitals (31%) had implemented ACS, while 134 (52%) had a traditional general surgeon on-call (GSOC) model. Thirty-eight hospitals (15%) had another model (hybrid). Larger-bed, university-based, teaching hospitals with Level 1 trauma center verification status located in urban areas were more likely to have adopted ACS. In multivariable modeling, hospital type, setting, and trauma center verification predicted ACS implementation. EGS processes of care varied, with 28% of the GSOC hospitals having block time versus 67% of the ACS hospitals (p < 0.0001), 45% of the GSOC hospitals providing ICU [intensive care unit] care to EGS patients in a surgical/trauma ICU versus 93% of the ACS hospitals (p < 0.0001), 5.7 ± 3.2 surgeons sharing call at GSOC hospitals versus 7.9 ± 2.3 surgeons at ACS hospitals (p < 0.0001), and 13% of the GSOC hospitals requiring in-house EGS call versus 75% of the ACS hospitals (p < 0.0001). Among ACS hospitals, there were variations in patient cohorting (EGS patients alone, 25%; EGS + trauma, 21%; EGS + elective, 17%; and EGS + trauma + elective, 30%), data collection (26% had prospective EGS registries), patient hand-offs (56% had attending surgeon presence), and call responsibilities (averaging 4.8 ± 1.3 calls per month, with 60% providing extra call stipend and 40% with no postcall clinical duties). The potential of the ACS on the national crisis in access to EGS care is not fully met. Variations in EGS processes of care among adopters of ACS suggest that standardized criteria for ACS implementation, much like trauma center verification criteria, may be beneficial.
    Journal of Trauma and Acute Care Surgery 01/2015; 78(1):60-8. DOI:10.1097/TA.0000000000000492 · 1.97 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: Excessive crystalloid administration is common and associated with negative outcomes in critically ill trauma patients. Continuous furosemide infusion (CFI) to remove excessive fluid has not been previously described in this population. We hypothesized that a goal-directed CFI is more effective for fluid removal than intermittent bolus injection (IBI) diuresis without excess incidence of hypokalemia or renal failure. CFI cases were prospectively enrolled between November 2011 and August 2012, and matched to historic IBI controls by age, gender, Injury Severity Score (ISS), and net fluid balance (NFB) at diuresis initiation. Paired and unpaired analyses were performed to compare groups. The primary endpoints were net fluid balance, potassium and creatinine levels. Secondary endpoints included intensive care unit (ICU) and hospital length of stay (LOS), ventilator-free days (VFD), and mortality. 55 patients were included, with 19 cases and 36 matched controls. Mean age was 54 years, mean ISS was 32.7, and mean initial NFB was +7.7 L. After one day of diuresis with CFI vs. IBI, net 24 h fluid balance was negative (-0.55 L vs. +0.43 L, P = 0.026) only for the CFI group, and there was no difference in potassium and creatinine levels. Cumulative furosemide dose (59.4mg vs. 25.4mg, P < 0.001) and urine output (4.2 L vs. 2.8 L, P < 0.001) were also significantly increased with CFI vs. IBI. There were no statistically significant differences in ICU LOS, hospital LOS, VFD, or mortality. Compared to IBI, goal-directed diuresis by CFI is more successful in achieving net negative fluid balance in patients with fluid overload with no detrimental side effects on renal function or patient outcome.
    Journal of Emergencies Trauma and Shock 01/2015; 8(1):34-8. DOI:10.4103/0974-2700.150395
  • Haytham M A Kaafarani, George C Velmahos
    Surgery 01/2015; 157(1):6-7. DOI:10.1016/j.surg.2014.10.001 · 3.11 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: Objective: We discuss the strengths of the medical response to the Boston Marathon bombings that led to the excellent outcomes. Potential shortcomings were recognized, and lessons learned will provide a foundation for further improvements applicable to all institutions. Background: Multiple casualty incidents from natural or man-made incidents remain a constant global threat. Adequate preparation and the appropriate alignment of resources with immediate needs remain the key to optimal outcomes. Methods: A collaborative effort among Boston's trauma centers (2 level I adult, 3 combined level I adult/pediatric, 1 freestanding level I pediatric) examined the details and outcomes of the initial response. Each center entered its respective data into a central database (REDCap), and the data were analyzed to determine various prehospital and early in-hospital clinical and logistical parameters that collectively define the citywide medical response to the terrorist attack. Results: A total of 281 people were injured, and 127 patients received care at the participating trauma centers on that day. There were 3 (1%) immediate fatalities at the scene and no in-hospital mortality. A majority of the patients admitted (66.6%) suffered lower extremity soft tissue and bony injuries, and 31 had evidence for exsanguinating hemorrhage, with field tourniquets in place in 26 patients. Of the 75 patients admitted, 54 underwent urgent surgical intervention and 12 (22%) underwent amputation of a lower extremity. Conclusions: Adequate preparation, rapid logistical response, short transport times, immediate access to operating rooms, methodical multidisciplinary care delivery, and good fortune contributed to excellent outcomes.
    Annals of Surgery 12/2014; 260(6):960-6. DOI:10.1097/SLA.0000000000000914 · 7.19 Impact Factor
  • Elie Ramly, Haytham M.A. Kaafarani, George C. Velmahos
    [Show abstract] [Hide abstract]
    ABSTRACT: Age-related anatomic, physiologic, and immunologic changes to the pulmonary system, as well as a high prevalence of chronic pulmonary diseases, puts the geriatric patient at an especially high risk for postoperative pulmonary complications. Successful perioperative respiratory care of the geriatric patient relies on careful risk assessment and optimization of pulmonary function and support. The success of such efforts aimed at preventing and/or mitigating pulmonary complications in the geriatric patient depends on a thorough, individualized, yet standardized and evidence-based approach to the care of every patient. Copyright © 2015 Elsevier Inc. All rights reserved.
    Surgical Clinics of North America 11/2014; 95(1). DOI:10.1016/j.suc.2014.09.009 · 1.93 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: Main concern during the practice of selective non-operative management (SNOM) for abdominal stab wounds (SW) and gunshot wounds (GSW) is the potential for harm in patients who fail SNOM and receive a delayed laparotomy (DL). The aim of this study is to determine whether such patients suffer adverse sequelae because of delays in diagnosis and treatment when managed under a structured SNOM protocol. 190 patients underwent laparotomy after an abdominal GSW or SW (5/04-10/12). Patients taken to operation within 120 min of admission were included in the early laparotomy (EL) group (n =153, 80.5 %) and the remaining in the DL group (n =37, 19.5 %). Outcomes included mortality, hospital stay, and postoperative complications. The median time from hospital arrival to operation was 43 min (range: 17-119) for EL patients and 249 min (range: 122-1,545) for DL patients. The average number and type of injuries were similar among the groups. Mortality and negative laparotomy were observed only in the EL group. There was no significant difference in the hospital stay between the groups. The overall complications were higher in the EL group (44.4 vs. 24.3 %, p =0.026). DL was independently associated with a lower likelihood for complications (OR 0.39, 95 % CI 0.16-0.98, p =0.045). Individual review of all DL patients did not reveal an incident in which complications could be directly attributed to the delay. In a structured protocol, patients who fail SNOM and require an operation are recognized and treated promptly. The delay in operation does not cause unnecessary morbidity or mortality.
    World Journal of Surgery 10/2014; 39(2). DOI:10.1007/s00268-014-2813-7 · 2.35 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: Current guidelines recommend cholecystectomy (CCY) during the index admission for mild to moderate biliary pancreatitis as delayed CCY is associated with a substantial risk of recurrent biliary events. Delayed CCY is recommended in severe pancreatitis. The optimal timing of CCY in necrotizing pancreatitis, however, has not been well studied. We sought to determine the safety of single-stage CCY performed at the time of necrosectomy and its effectiveness in preventing subsequent biliary complications.
    Gastroenterology 10/2014; 19(1). DOI:10.1007/s11605-014-2650-x · 13.93 Impact Factor
  • Annals of Emergency Medicine 10/2014; 64(4):S7–S8. DOI:10.1016/j.annemergmed.2014.07.042 · 4.33 Impact Factor

Publication Stats

10k Citations
1,409.86 Total Impact Points


  • 2008–2015
    • Massachusetts General Hospital
      • • Department of Surgery
      • • Division of Trauma, Emergency Surgery and Surgical Critical Care
      Boston, Massachusetts, United States
  • 2006–2015
    • Harvard Medical School
      • Department of Surgery
      Boston, Massachusetts, United States
  • 2005–2014
    • Harvard University
      Cambridge, Massachusetts, United States
  • 2013
    • University of Michigan
      • Department of Surgery
      Ann Arbor, Michigan, United States
    • Boston University
      Boston, Massachusetts, United States
  • 2012
    • University of Massachusetts Medical School
      • Department of Surgery
      Worcester, Massachusetts, United States
  • 1996–2009
    • University of Southern California
      • Department of Surgery
      Los Angeles, California, United States
  • 2000–2007
    • Keck School of Medicine USC
      Los Ángeles, California, United States
    • California State University, Los Angeles
      Los Ángeles, California, United States
  • 1996–2007
    • University of California, Los Angeles
      • Department of Surgery
      Los Ángeles, California, United States
  • 2004
    • Oregon Health and Science University
      • Department of Surgery
      Portland, OR, United States
  • 2002
    • California State University, Sacramento
      Sacramento, California, United States
  • 1999
    • Santa Barbara Cottage Hospital
      Santa Barbara, California, United States
  • 1994–1998
    • University of the Witwatersrand
      • Department of Surgery
      Johannesburg, Gauteng, South Africa
    • Chris Hani Baragwanath Hospital
      Johannesburg, Gauteng, South Africa
  • 1993
    • University of Johannesburg
      Johannesburg, Gauteng, South Africa