South African Surgical Outcomes Study

Methods -Seven day, South African national multi-centre cohort study of adult (≥16 years) patients undergoing in-patient non-cardiac surgery.

Centres

-We aim to recruit from as many South African centres as possible. Every Medical University in South Africa will participate in SASOS (University of Cape Town, University of the Free State, University of KwaZulu Natal, University of Limpopo, University of Pretoria, University of Stellenbosch, University of the Witwatersrand, and Walter Sisulu University), and most of the hospitals they support and serve will participate in this study.

Ethics approval

• Ethics approval has been obtained from each university centre.
• This study is in effect a large scale clinical audit, and as such a waiver of consent has been approved by seven of the eight medical school ethics committees, similar to the international precedent that was already set, as in the original EuSOS study, where consent was waived in 27 of the 28 European countries participating. Only the University of Witwatersrand requires patient consent for SASOS.

Data collection and collation

• Data will be collected in individual centres on paper case record forms (CRFs) for every patient recruited. Paper CRFs will be stored within a locked office in each centre as they will include identifiable patient data in order to allow follow-up of clinical outcomes. Data will then be pseudo-anonymised by generation of a unique numeric code and transcribed by local investigators onto an internet based electronic CRF. Each patient will only be identified on the electronic CRF by their numeric code; thus the co-ordinating study team cannot trace data back to an individual patient without contact with the local team. A participant (patient) list will be used in each centre to match identifier codes in the database to individual patients in order to record clinical outcomes and supply any missing data points. Access to the data entry system will be protected by username and password delivered during the registration process for individual local investigators. All electronic data transfer between participating centres and the co-ordinating centre will be encrypted using a secure protocol (HTTPS/SSL 3.0 or better).
• Each centre will maintain a secure trial file including a protocol, local investigator delegation log, ethics approval documentation, the participant list, etc.
• A final summary printout of included patients with major variables should be produced for each centre together with final data submission to double check for completeness and accuracy.

Dataset

• A realistic data set will be fundamental to the success of the investigation, and this was confirmed in the EuSOS study where nearly complete data was available on 46 000 patients. We have therefore adopted the EuSOS dataset with minor changes following advice from the EuSOS steering committee, and believe that these key data points will not discourage centres from participating because of an excessive burden of data collection. The reliability of data collection will be analysed formally using K-statistics or intra-class correlation coefficients as appropriate.
• Centre co-ordinators may request the addition of a limited number of data points to support the national SASOS data collection and for subsequent regional analyses. All additional data points must be discussed with the co-principal investigators and if necessary the steering committee.
• Centre specific data will be collected once for each hospital including: secondary/tertiary centre, number of operating rooms, number and level of critical care beds, details about the reimbursement status of the hospital and public holidays or other local factors affecting patient throughput during study period.
• An operating room case record form (CRF) and surgical ward CRF will be completed for every eligible patient who undergoes surgery during the seven day cohort period. Patients will be followed up until hospital discharge. This will be censored at thirty days i.e. patients will be followed up until discharge or for thirty days whichever is the shorter period. If a patient is admitted to critical care at any time during the follow-up period, then a critical care CRF will be completed. The critical care CRF may therefore be completed for patients admitted to critical care after the seven day cohort period is complete.

Sample size calculation

• We intend to recruit as many patients as possible in a seven day period, in order to establish a large, representative sample from all participating South African centres. Based on audit data, we believe that it is possible that the sample size may be close to 7,500 patients. Assuming an overall mortality rate following surgery of 4%, a sample size of 7,500 patients will yield 300 deaths. This will allow the inclusion of at least thirty variables in a logistic regression model for mortality.
• The rate of admission to critical care is likely to vary between centres. Based on an anticipated 7,500 patients this will be provide a sample of up to 300 admissions to critical care after surgery, based on an assumption that critical care admission in South Africa is half that of Europe (4% versus 8% respectively). We expect this to allow a robust logistic regression model for this outcome.
• This dataset should have sufficient generalisability to inform the practice of peri-operative care on a national basis in South Africa.

Statistical analysis

• The data to be collected are all collected as part of routine clinical care. Categorical variables will be described as proportions and will be compared using chi-square tests. Continuous variables will be described as mean and standard deviation if normally distributed or median and inter-quartile range if not normally distributed. Comparisons of continuous variables between groups will be performed using t-tests, one-way ANOVA or equivalent non parametric tests as appropriate. Univariate analysis will be performed to test factors associated with planned and unplanned admission to critical care and / or in-hospital death.
• Generalized linear mixed models using a logit link will be used to identify independent risk or prognostic factors for binary outcomes. These will include one-level models and hierarchical two-level models to account for the expected correlation in outcomes within hospitals. Data will first be analysed by Province, and then at the Country level, using Province as an independent variable. A stepwise approach will be used to enter new terms into the regression models where a univariate p<0.05 will be set as the limit for inclusion of new terms. Results will be reported as adjusted odds ratios (OR) with 95% confidence intervals. A single final analysis is planned at the end of the study.

Primary outcome measure -In-hospital mortality

Secondary outcome measures

• Duration of hospital stay
• Admission to critical care after surgery
• Duration of critical care stay

Organisation

-The Steering Committee will be chaired by Bruce Biccard (BB) and Thandinkosi Madiba (TM). The study management team will be appointed by the Steering Committee and led by BB and TM. The duties of this team will include administration of all project tasks, communication between project partners (including funders, steering committee members, national and local co-ordinators, etc), data collation and management and preparation of reports for individual study sites. The Steering Committee is responsible for the scientific conduct and consistency of the project. The Steering Committee will ensure communication between the funder(s), study management team and co-ordinators as necessary.