BACKGROUND
Occurrence of post-operative complications is used as a surrogate for quality of surgical care.(1–3) The Clavien–Dindo (CD) classification of post-operative complications was introduced to facilitate standardization of reporting of post-operative complications. When the CD classification is used, a complication is graded based on the intervention which is then required in the management of the complication.(1–4) The CD classification categorizes post-operative complications into seven classes. It has been established that the CD classification is feasible for use across surgical specialties.(5–8) Challenges which continue to be identified during the use of CD classifications have necessitated ongoing modifications.(4,5,9–11)
Access to quality surgical care in low- and middle-income countries (LMICs) is limited, which may influence the occurrence of post-operative complications.(12) Patients in LMICs often present late when the disease is advanced. Availability of medical specialists, intensive care unit (ICU) beds and radiology expertise for percutaneous interventional techniques are also limited.(13–19) This study was conducted to report our initial experience following the adoption of the CD classification for use to report post-operative complications across surgical specialities at a tertiary academic hospital in the Gauteng Province in South Africa.
METHODS
An audit of prospectively collected data of morbidity and mortality records of all patients who were 18 years and older and had surgical procedures between 27 February 2017 and 28 February 2018 in divisions and units of Departments of Surgery at Charlotte Maxeke Johannesburg Academic Hospital (CMJAH) was conducted. Permission to conduct the study was received from the Human Research Ethics Committee of University of the Witwatersrand (M170148) and the Research Review Board of the CMJAH.
Consecutive patients who had emergency or elective surgical procedures were included. Data was collected using the Morbidity and Mortality Records of the Department of Surgery on the Research Electronic Data Capture (REDCap) platform. These meetings are held at weekly intervals as part of quality assurance and ongoing training. The meeting is attended by nursing and medical staff of different levels for quality purposes. Informed consent to participate was routinely sought from all patients on admission as part of the REDCap database. Patients who were not able to give informed consent were excluded. Initial categorization of post-operative complications was done by individual surgical specialities and verified during weekly combined morbidity and mortality meetings.
Data collected included patients’ demographics, acuity of presentation (emergency or elective), types of surgical procedures and recorded post-operative complications, including final agreed grades according to CD classification (Table 1).
Clavien–Dindo classification2
| Clavien–Dindo classification | Definition |
|---|---|
| Grade I | Any deviation from normal post-operative course which did not need any intervention except treatment with usual drugs or opening of the wound at the bedside |
| Grade II | Any deviation from normal post-operative course requiring specific pharmacological therapy, blood and blood products transfusion or total parenteral nutrition |
| Grade IIIa | Any deviation from normal post-operative course requiring surgical, endoscopic or radiological intervention without general anaesthesia |
| Grade IIIb | Any deviation from normal post-operative course requiring surgical, endoscopic or radiological intervention under general anaesthesia |
| Grade IVa | Any deviation from normal post-operative course leading to a life-threatening complication with single-organ dysfunction requiring intensive care or ICU admission |
| Grade IVb | Any deviation from normal post-operative course leading to a life-threatening complication with multi-organ dysfunction requiring intensive care or ICU admission |
| Grade V | Any deviation from normal post-operative course leading to death |
Categorical variables were summarized using frequencies and percentages. The mean with standard deviation (SD) was used for the aggregation of continuous data. χ 2-Test or Fisher's exact test was used to assess if there was a relationship between the type of surgical procedures and gender. The strength of associations was measured by Cramer's V and the ϕ coefficient, and the following scale of interpretation was used: high or strong association if coefficient is ≥0.50, moderate for 0.30–0.49, weak for 0.10–0.29 and little or no association if it was <0.10.
The significance of the relationship between the patients’ age and the types of surgery was assessed using Student's t-test or one-way analysis of variance for comparing more than two patient groups. In cases where these tests were not appropriate, the Wilcoxon rank-sum test comparing two groups or Kruskal–Wallis test for more than two groups was used. The strength of associations was measured using Cohen's d for parametric tests and r-value for non- parametric tests. The following scale of interpretation was used: large effect if ≥0.80, moderate if 0.50–0.79, small if 0.20–0.49 and near zero if <0.20.
Comparison of the rate of complications between surgical specialties was done using the z-test with adjustment for multiple comparisons. Data analysis was carried out using SAS version 9.4 for Windows. The level of significance was set at a P value below 0.05.
RESULTS
A total of 250 weekly morbidity and mortality records were studied. Five thousand six hundred and sixty-five admissions with full documentations were noted in the study period. The records contained a total of 3399 surgical procedures of which 1700 (50.0%) were emergency procedures. Elective major and minor operations comprised 1485/3399 (43.7%) and 214/3399 (6.3%) of all surgical procedures, respectively. However, when looking at gender distribution per surgical category, more females had elective and minor procedures accounting for 55.9% and 54.2% for each category, respectively. Of the 1700 emergency operations, 1205 (70.9%) were performed on male patients. The mean age (±SD) of patients who had surgical procedures was 44.3 (±16.7) years.
The majority 689/832 (82.8%) of surgical procedures performed by the Trauma Unit were major emergencies as compared to 307/866 (35.3%), 253/661 (38.3%), 368/777 (47.3%) and 129/263 (49.1%) by the Breast and Endocrine, Vascular, Surgical Gastroenterology Unit and Transplant Unit, respectively. There was a significant but moderate association between gender and admissions in surgical specialties (P < 0.0001; Cramer's V = 0.41). Similarly, there was also a significant but moderate association between the types of surgical procedures undertaken in surgical specialties (P < 0.0001; Cramer's V = 0.36). Moreover, 768/832 (92.3%) of recorded surgical procedures performed by the Trauma Unit were on males compared to 310/866 (35.8%) of procedures in males which were done by the Breast and Endocrine Unit.
One-fifth (n = 178; 20.5%) of Breast and Endocrine Unit procedures were performed as minor procedures. There was a significant but weak association between gender and surgical specialty (P < 0.0001; Cramer's V = 0.27) (Table 2).
Percentage breakdown of patients operated by each surgical specialty
| Parameter | Trauma n = 832 (%) | Breast and Endocrine n = 866 (%) | Vascular n = 661 (%) | Surgical gastroenterology n = 777 (%) | Transplant n = 263 (%) | P value |
|---|---|---|---|---|---|---|
| Gender | <0.0001 | |||||
| Males | 768 (92.3%) | 310 (35.8%) | 391 (59.2%) | 408 (52.5%) | 149 (56.6%) | |
| Females | 64 (7.7%) | 556 (64.2%) | 270 (40.8%) | 369 (47.5%) | 114 (43.4%) | |
| Type of surgery | <0.0001 | |||||
| Minor elective | 0 (0%) | 178 (20.5%) | 0 (0%) | 9 (0%) | 0 (0%) | |
| Major elective | 143 (17.2%) | 382 (44.1%) | 408 (61.7%) | 409 (52.7%) | 134 (50.9%) | |
| Major emergency | 689 (82.8%) | 307 (35.4%) | 253 (38.3%) | 368 (47.3%) | 129 (49.1%) |
The average age of patients who had a surgical procedure following trauma was lower as compared to patients who had vascular operations (Figure 1).
Mean age and 95% confidence interval in years of patients who were operated on in each of the surgical specialties. Key: Trauma = Trauma Unit, 394 = Breast, Endocrine, Sarcoma and Melanoma Unit, 395 = Vascular Unit, 396 = Surgical Gastroenterology Unit, Transplant = Transplant Unit.
Age difference of patients who were operated across specialties was statistically significant (P < 0.0001). Post-hoc tests showed that the mean age of patients who had emergency procedures in the Vascular Unit was lower as compared to those who had elective surgery. However, there was no significant association between patient age and surgery type for patients from the Trauma Unit and Transplant Unit.
Relook procedures were recorded in 480/3999 (12.0%) of the records studied. Some patients required more than one relook. A total of 370/480 (77.1%) relook procedures were performed on males. The mean age (±SD) of patients who had relook procedures was 38.4 (±14.9) years. The Trauma Unit performed 129/480 (26.9%) of the relook procedures, whereas the Vascular Unit performed 22/480 (4.6%) during the study period.
There were a total of 449/3999 (11.2%) complications. Descriptive entry of complications was recorded in 433 of the cases and 56/433 (12.9%) of descriptive post-operative complications stated surgical site infection (SSI). Various descriptions were used to capture SSIs and included wound infection, post-operative wound infection and wound breakdown (Figure 2).
From Table 3, the Surgical Gastroenterology Unit reported the highest overall post-operative complication rate of 147/777 (18.9%), whereas the Transplant Unit reported the lowest at 18/263 (6.8%) (Table 3).
Rate of reported occurrence of post-operative complications in surgical specialties
| Specialty | Overall rate of occurrence of complications n (%) |
|---|---|
| Trauma | 142/832 (17.1%) |
| Breast and Endocrine | 76/866 (8.8%) |
| Vascular | 79/661 (11.9%) |
| Surgical gastroenterology | 147/777 (18.9%) |
| Transplant | 18/263 (6.8%) |
Two hundred and eighty-eight (66.4%) post-operative complications occurred in males. The mean age (±SD) of patients who had complications was 45.4 (±16.9) years.
Of the 462 complications, 426 (92.2%) were graded according to CD of which 131 (31.6%) were categorized as Grade I and 113 (26.5%) as Grade IIIb. The distribution of complications as reported for the entire cohort and each of the specialties is shown in Table 4.
Breakdown of reported complications according to CD classification
| Parameter | Trauma n = 142 (%) | Breast and Endocrine n = 76 (%) | Vascular n = 79 (%) | Surgical Gastroenterology n = 147 (%) | Transplant n = 18 (%) |
|---|---|---|---|---|---|
| Category of complication | |||||
| Infective | 42 (29.4%) | 40 (52.9%) | 45 (56.7%) | 87 (59.0%) | 7 (36.8%) |
| Non-infective | 100 (70.6%) | 36 (47.1%) | 34 (43.3%) | 60 (41.0%) | 11 (63.2%) |
| Relationship of complication with surgical procedure | |||||
| Related | 96 (67.7%) | 63 (82.4%) | 44 (55.4%) | 84 (56.5%) | 14 (77.8%) |
| Not related | 42 (29.9%) | 11 (15.3%) | 30 (37.5%) | 57 (39.1%) | 1 (5.6%) |
| Equivocal | 3 (2.4%) | 2 (2.4%) | 5 (7.1%) | 6 (4.3%) | 3 (16.7%) |
| CD classification | |||||
| Grade I | 25 (17.9%) | 20 (25.9%) | 23 (29.5%) | 72 (48.9%) | 5 (27.8%) |
| Grade II | 36 (25.6%) | 18 (23.5%) | 17 (21.3%) | 20 (13.5%) | 3 (16.7%) |
| Grade IIIa | 33 (23.1%) | 9 (12.9%) | 12 (14.8%) | 17 (11.3%) | 0% |
| Grade IIIb | 40 (28.2%) | 27 (35.3%) | 23 (29.5%) | 30 (20.3%) | 5 (27.8%) |
| Grade IVa | 29 (1.7%) | 1 (1.2%) | 4 (4.9%) | 2 (1.5%) | 4 (22.2%) |
| Grade IVb | 1 (0.9%) | 1 (1.1%) | 0 (0%) | 5 (3.8%) | 0 (0%) |
| Grade V | 4 (2.6%) | 0 (0%) | 0 (0%) | 1 (0.8%) | 0 (0%) |
Using the CD classification, 199/414 (48.1%) of all post-operative complications were infections and 131/199 (65.8%) were directly related to surgical procedures. Reported rate of occurrence of post-operative complications in the Trauma and the Surgical Gastroenterology Units was 17.1% and 18.9%, respectively. There was a significant but weak association between the reported complication type and surgical specialty (P < 0.0001; Cramer's V = 0.25). Trauma Unit and Transplant Unit reported proportionally fewer infective complications compared to the other specialties (Figure 3).
Percentage breakdown of infective and non-infective complications in surgical specialties. Key: Trauma = Trauma Unit; 394 = Breast, Endocrine, Sarcoma and Melanoma Unit; 395 = Vascular Unit; 396 = Surgical Gastroenterology Unit; Transplant = Transplant Unit.
There was a significant but weak association between the grade of complication and surgical specialty (P < 0.0001; Cramer's V = 0.21). The Surgical Gastroenterology Unit had a higher proportion of CD classification Grade I and a lower proportion of Grade II and IIIa complications as compared to the Trauma Unit. The Breast and Endocrine Unit had a higher proportion of Grade IIIb complications as compared to the Surgical Gastroenterology Unit.
Sixty-eight per cent (68.0%) of post-operative complications were directly related to surgical procedures with 82.4% for Breast and Endocrine Unit as compared to 77.8%, 67.7%, 56.5% and 55.4% for Transplant, Trauma, Surgical Gastroenterology and Vascular Unit; respectively. Twenty-nine per cent (29.4%) of post-operative complications for patients from the Trauma Unit were infective as opposed to 52.9%, 56.7%, 59.0% and 36.8% in Breast and Endocrine, Vascular, Surgical Gastroenterology and Transplant Units, respectively. The overall percentage of infective complications was 47.0.
Overall, 437 mortalities occurred during the study period but included patients who did not have any surgical procedure during the index admission. This constituted an overall mortality rate of 7.7%. Two hundred and eighty-seven (287/437; 65.7%) of the deaths were in male patients. The mean age (±SD) of patients who died was 49.1 (±19.0) years.
DISCUSSION
This study is amongst the few studies which investigated the feasibility and usefulness of the CD classification in reporting post-operative complications following a combination of emergency and elective surgical procedures across traditional general surgery specialties.(9) Like reports from other studies, the transition from using word description to the adoption of the CD classification format was seamless in all participating units.(12) It eliminated inconsistencies associated with the use of descriptive entry of post- operative complications which is cumbersome as various terminologies are used to describe some of the complications.(10)
Blanket categorization of complications into minor or major is too broad and not useful for audit and quality improvement plans.(11) Further categorization of post-operative complications into minor and major is reproducible and straightforward when it is done concurrently with CD classification.(10,11)
Occurrence of post-operative complications is used as a surrogate marker of quality of health care. Attributes of quality care include provisioning of safe, timeous, effective, efficient, equitable and patient-centred surgical care. Post-operative complications are part and parcel of surgical services and may not be eliminated. The overall post- operative complication rate of 13.2% in this study is within the reported range from other studies but is lower than what would be expected from a hospital in LMIC setting.(9,10,25) The latter may be due to the under-reporting of surgical complications, known to be a challenge globally.(20,21) The overall mortality rate of 7.7% in this study period would be acceptable bearing in mind that 50% of the procedures were related to emergencies. It either suggests that there were no near-misses or inability to rescue patients who developed post-operative complications.
In-line with global literature, infective complications were the most reported post-operative complication making up to around 48%, two-thirds of which were directly related to the surgical procedures. The most frequently reported class of post-operative complications as per CD classification was Grade I, which is consistent with previous findings.(22) It is expected that over 70% of post-operative complications would be considered minor or moderate, a combination of Grade I and Grade II based on the CD classification.(10,23) However, results from this study showed that Grade IIIb complications were the second most common post-operative complication and were maintained across all surgical specialties despite a significant difference in the demographic and clinical profile of patients.
A higher proportion of CD classification Grade IIIb compared to Grades II and IIIb differs from what has been reported from previous studies.(10,12) The occurrence of Grades I, II and IIIa probably reflected a reporting bias or impact of the availability of resources. Grade I complications were variable between different units, with gastroenterology reporting the highest minor complications, and trauma the lowest in the study group. This might also reflect the different populations noted in these diverse patient populations.
Some of the factors which may influence the occurrence of post-operative complications include gender,(24) age (22) and acuity of the procedure;(9) site of operation;(25) perioperative care; comorbidities, including obesity,(2,23,25,26) surgical expertise (11) and perioperative blood transfusion.(24) There was a significant difference in the age and gender of patients who had surgical procedures across various specialties. Although more than 92% of the patients operated by the Trauma Unit were males, gender representability was a close 50:50 in other specialties. Notwithstanding, over 66% of recorded complications occurred in males.
Emergency surgical procedures were performed in relatively younger patients except in the Vascular Unit, in which majority of the vascular emergency operations were done in patients above the age of 60 years. The mean age of trauma patients was below 32 years, compared to over 55 years for vascular patients. All specialties except Breast and Endocrine Unit focused only on major surgical procedures. Grading of post-operative complications did not factor-in the magnitude of surgical procedures.
LIMITATIONS
The study was reliant on self-reporting by each surgical specialty unit and may not be a true reflection of rate of occurrence of post-operative complications. All deaths were captured in the same platforms, including for patients who were not operated on. Some deaths which occurred in patients who had emergency surgical procedures could not be specifically deemed to have been due to the performance of a surgical procedure. Many factors which may influence the occurrence of post-operative complications were not captured in this study, including the unplanned need for blood and blood products transfusion, initiation of total parenteral nutrition, site of operation and comorbidities.
There were variations in assigning grades for patients who had relook procedures and patients who self-extubated themselves while being ventilated in the ICU. Some patients required ICU care for the management of their complications but were not transferred due to a shortage of beds and this, in turn, may have influenced grading. Additionally, the Breast and Endocrine Unit also covered acute-care surgery after-hours, which predominately involves abdominal procedures, possibly influencing their results.
CONCLUSION
The CD classification was adopted by all specialties studied. The overall rate of post-operative complications was 11%, the majority of which were infections. Reported grades of complications and types of surgery are associated with different specialities. A high number of Grade IIIb complications were recorded than have been reported previously. We recommend a specialty-based follow-up study which would link grading of post-operative complications using the CD classification with intraoperative events.
ACKNOWLEDGEMENT
The research team sincerely appreciates collaboration from members of all surgical divisions which participated, including surgical trainees and medical officers who assisted in entering of data onto the REDCap database and preparation of weekly morbidity and mortality reports. The authors are also grateful to the team from Data Management and Statistical Analysis (DMSA) for their assistance with statistical analysis.