WHO SSC : Compliance and Associated Surgical Outcomes in Uganda's Referral Hospitals

Multicentre prospective cohort study in five of Uganda's referral Hospitals in Uganda including the referral private not for profit (PNFPs) hospitals.

Target population: Patients undergoing surgery in Uganda's referral hospitals.

Study population: Surgical operations being performed in Uganda's referral hospitals over a 4-month period.

Sampling method

Purposive sampling of the hospitals was done based on their geographical location, their influence in the areas where they are situated and the presence of an anaesthesiologist in the hospital. The surgeries were enrolled into the study by probability proportional to the size sampling. The sampling unit was surgical operations done in the eligible hospitals. Surgical operations were enrolled into the study systematically during the study period.

Sample size:

Primary objective: To determine the extent of compliance to the WHO surgical safety checklist among Uganda's referral hospitals.

The sample size for this objective was calculated using the Modified Kish \& Leslie method (1965)

N= [Z² p (1-p)] / d² where;

Z= is a percentile of the standard normal value determined by the 95% confidence level (1.96).

N= minimum required sample size

d= desired precision for the study = 0.05

p= is the estimated proportion of hospitals that had poor compliance to SSC = 50%

The proportion used to compute the sample size was the proportion of hospitals which had poor compliance to the SSC. This proportion was obtained from the WHO recommendation when no previous proportions are available = 50% n= 384 participants, since the study was done in several hospitals, adjustment for clustering was done by applying a cluster effect of 2.

Sample size = 768

Adjusting for a 10% loss to follow up, N = n (100+10)/100

Sample size = 845

Variables:

The categorical variables included patient demographics, urgency of the surgery, actual use of the WHO SSC, qualification of staff and the team leader present in the operating room during the use of the checklist, the team member who initiated the checklist, timing of the performance of the different domains and post-operative complications within a 24 hour period after the surgery and daily until discharge or 30 days of in hospital stay, while the ordinal variables included percentage of items checked among the 19 parameters and 3 broad domains of the WHO SSC. These variables were ticked off the questionnaire by the research assistant as they observed the surgery. Compliance was expressed as the percentage of items that has been completed per checklist.

Data collection

Data was collected using standardized sheets and stored safely using waterproof boxes and treated with utmost confidence. All patients undergoing surgical operations were recruited into the study regardless of their age, surgical specialty and the urgency of the surgical operation provided consent to participate in the study was given. The research assistant present in the operating room observed the activities in the room and completed a structured questionnaire regarding demographics, the 19 parameters and 3 domains of the WHO SSC, near misses, qualification of the staff present in the operating room, and intraoperative variables. Technical and intraoperative complications were recorded as well. The patients were followed up and questioned on the occurrence of complications at 24 hours and daily until discharge or 30 days post-surgery whichever came first. The nurses on the ward reported any immediate post-operative complications.

The complications were recorded as defined by the American College of Surgeons National Surgical Quality Improvement Program database ; Bleeding: overt frank bleeding from the surgical site, heavy soiling of the surgical site dressing with blood or need to change dressing within the first 24 hours due to soiling with blood, respiratory complications; acute airway obstruction, inability to saturate above 92% off oxygen therapy within the first 24 hours after surgery with absence of previous hypoxemia on room air and prolonged intubation, post-operative nausea and vomiting, loss of consciousness or altered consciousness within the first 24 hours after surgery based on GCS scores before and after the surgery, cardiac and hemodynamic complications such as hypotension which were defined as a blood pressure of less than 90/60 mmHg, need for unanticipated blood transfusion intraoperatively and 24 hours postoperatively cardiac arrest, pain and death. The patients were then followed prospectively for any other complications in the perioperative period. Any other occurrences that the ward nurses considered complications attributed to the perioperative period were reported to the research assistants and this was recorded in the questionnaire. The patients exited the study at discharge or at 30 days of in hospital stay.

Checklist compliance was defined as "non compliance" (none of the 19 items checked off), "partial compliance" (at least 1 to 9 items checked off) and "high compliance" (10 to 19 items checked off) and was expressed as a percentage of the number of the expected checked off items.

Data safety

Raw data sources were received from the different hospitals and stored in safe waterproof boxes. An electronic data entry tool was formulated and used to store the data on multiple storage devices.

Data management

Data were double-entered in an entry tool created using Epidata software version 3.1. This had in-built checks. The data were validated and inconsistencies cleared accordingly. The data were backed up and a copy was sent to the data analysis program.

Data analysis

Data were analysed using STATA version 13 (StataCorp.2013.Stata Statistical Software: Release 13. College Station, TX: StataCorp LP)

Descriptives: Descriptives were summarised using means and standard deviations for the parametric continuous data. Categorical data were summarised using proportions and percentages.

Comparisons: Comparisons for variables between participants with poor and good surgical outcomes were made. For the parametric continuous data, the student t test was used and for the non-parametric data, Mann-Whitney- U test was employed. For the categorical data, the Chi-Square test of Fisher's exact test was used.

Prevalence of compliance was calculated as follows; the numerator was the number of checked off items on the checklist. The denominator was the total number of expected checked off items. We expressed compliance as a percentage completion of the items comprising the checklist. There is a total of 19 items in the checklist, therefore high compliance involved adhering to at least 10 of them while non-compliance involved not adhering to any of the items. We assigned each of the items of the checklist a score of 1 then calculated a percentage of the compliance by dividing the score attained out the 19 which is the expected score. We presented the mean percentage compliances with their corresponding 95% confidence intervals.

Bivariate analysis: Binary logistic regression model was used to assess for association of each of the predictors with mortality while a linear regression model was used to asses for the association with the length of hospital stay

Multivariate analysis: All factors with p value of 0.2 or less at bivariate analysis were simultaneously entered in multivariate linear and logistic regression models respectively. Confounding was assessed for at a 10% or more difference between the unadjusted and adjusted models. Statistical significance was defined as p < 0.05

Analysed data were presented in figures, tables and text. Results of the linear regression were expressed as coefficient and the ones of logistic regression expressed as OR with a confidence interval of 95%.