A multiphase observational study was conducted sequentially. To assess the representativeness of the potentially eligible sample of the SINERGY trial, a retrospective observational multicenter approach was used, which compared nonparticipating hospitals with hospitals participating in the SINERGY trial. Subsequently, a pre-post study was conducted in the nonparticipating hospitals in order to identify any changes in perioperative practice, length of functional recovery, and length of postoperative hospital stay. This study was financially supported by the Netherlands Organization for Health Research and Development (ZonMw) (project number: 837003002).
The current study can be seen as an addition to the nationwide SINERGY trial . All the gynecology departments of the 27 hospitals authorized to perform major gynecologic oncologic surgery in the Netherlands were assessed for eligibility. In line with the SINERGY trial protocol , one hospital that referred most of its patients to inpatient rehabilitation facilities after hospital discharge was excluded a priori. In total, 26 eligible hospitals were invited to participate in the SINERGY trial.
A combination of targeted recruitment efforts were undertaken to increase involvement in the SINERGY trial. In mid-2013 local stakeholders were identified and contacted directly by members of the expert team. A personal letter was sent, which contained information about the study project, the implementation strategies being tested, the objectives of the perioperative guidelines, the need for active implementation, and the expected time and personal commitment required. In conformance with the SINERGY trial protocol , and in order to minimize implementation effects before the start of the study project, detailed information about the specific elements of the perioperative guidelines were not disseminated. Local stakeholders were, however, aware of the intentions of the SINERGY trial to reduce time to recovery and length of hospital stay by implementing evidence-based perioperative elements according to the enhanced recovery after surgery (ERAS) approach . Reminders by email and telephone were used consecutively, when necessary. Visits to local organizations to provide additional information were undertaken on request. The study was supported by the Dutch Gynecology Oncology Group (DGOG), which comprises professionals who collaborate to promote research in gynecologic cancer. Funding provided by ZonMw helped to tackle financial disincentives by confining the local costs of implementation to attending meetings and executing local activities. Local stakeholders continued to be approached until a statement of commitment or nonparticipation was provided, resulting in a response rate of 100%. Even though all hospitals acknowledged the need for the nationwide standardization of perioperative practice consistent with ERAS, some decided not to participate in the SINERGY trial (nonparticipation rate 27%). These hospitals did not undergo a baseline audit of perioperative practice to determine if they were eligible for inclusion in the SINERGY trial. Stakeholders were asked about their reasons for nonparticipation. A combination of the compulsory nature of collective multidisciplinary meetings, a lack of time and resources due to high workload, and the lack of financial incentives were reported. In addition, nonparticipants questioned the return on their investment of implementing the evidence-based perioperative elements in improving time to recovery and length of hospital stay.
In total, seven hospitals chose not to take part in the SINERGY trial. Five of these seven hospitals agreed to participate in the current study. These ‘nonparticipating in SINERGY trial’ hospitals were compared with eligible participants of the SINERGY trial in order to address the first research objective of determining sample representativeness. The participant group consisted of hospitals that intended to participate in the SINERGY project (N = 19) . For the second research objective, only the nonparticipating hospitals were analyzed for change in perioperative practice and outcomes following recruitment activities for the SINERGY trial. The comparison with participating hospitals will be made in future research, which will report data of the participating hospitals during the implementation interventions in more detail as well. Patients treated in the period before the recruitment process began, and thus before awareness of the SINERGY trial (pre-awareness group), were compared with patients treated one year after awareness of the trial (post-awareness group). This one-year period was chosen because the implementation interventions for the SINERGY trial were to begin in June 2014 . In this way the effect of awareness was isolated as much as possible.
Nonparticipating and participating hospitals were compared to identify differences in baseline characteristics before the onset of the recruitment process. The total number of beds per hospital, the type of hospital (university or non-university), and the level of specialized health care (secondary or tertiary) were obtained. Effect measures of the implementation of evidence-based perioperative elements were time to functional recovery and length of postoperative hospital stay. The outcomes in these effect measures were assessed in retrospect. Functional recovery was achieved when patients tolerated a normal diet, mobilized independently, and were comfortable on oral analgesia. Day zero refers to the day of surgery.
The outcomes of the second research objective, to explore the effect of awareness, were overall adherence rate to selected evidence-based perioperative care elements, time to functional recovery, and length of postoperative hospital stay in days. The selection of elements was in line with the SINERGY trial protocol  and an overall mean adherence rate was calculated. Selected elements included preadmission counseling and education, no bowel lavage, oral carbohydrate loading before surgery, routine use of prophylactic anti-emetics, nasogastric tube removal after surgery, no use of peritoneal drains, oral fluid intake on day zero, start solid food on the first postoperative day, mobilization on chair at least three times on the first postoperative day, use of epidural analgesia, routine use of oral laxatives, and urinary catheter removal before the third postoperative day. These predefined elements can be retrieved reliably and have a direct influence on recovery. In combination they should reflect an enhanced recovery approach.
In each nonparticipant hospital, retrospective data of two independent, consecutive samples of 30 patients were collected for the pre-awareness (January 2012) and post-awareness (June 2014) group. Baseline characteristics, surgical details, and outcome measures were retrieved from medical records. Patients underwent explorative laparotomies, surgical staging, or cytoreductive procedures for suspected, primary, or recurrent gynecologic malignancies. By selecting this specific group of patients, a comparable case-mix was provided for all hospitals. In addition, hospital characteristics were collected. For the participant group the baseline audit included a consecutive sample of 30 patients treated in 2012. No exclusion criteria were applied at the patient level.
Absolute numbers with percentages were used for categorical data, while means with standard deviations (SD) or medians with ranges, where appropriate, were used for numerical variables. Differences were assessed using the independent-samples t-test or the Mann-Whitney U test for numerical variables, and the chi-square test or Fisher’s exact test, where appropriate, for categorical ones. Both unadjusted and adjusted statistical methods were used to identify differences in implementation endpoints between the nonparticipating and participating hospitals and to assess the overall effect of awareness on perioperative management and outcomes after elective gynecologic surgery. Linear multilevel analysis methods were used to compare the time to functional recovery and the length of hospital stay between the nonparticipating and participating hospitals. Models were adjusted for clustering (random effect) and patient and surgical characteristics (fixed effects). Study group, age, American Society of Anesthesiologists’ classification, (suspected) diagnosis and type of surgery, histological subtype, surgical approach, operation time, type of incision, and type of additional surgical procedure were included in the models as fixed effects. To evaluate the overall effect of awareness, the hospitals included in the nonparticipant study could not be treated as a random sample, but were included in the model as a fixed factor. So, multivariable linear regression models, adjusted for patient and surgical characteristics, were used. Furthermore, we were also interested in the separate outcomes for each hospital. Therefore, subsequent analyses of interaction terms between groups (pre-awareness versus post-awareness) and hospitals were performed to determine separate effects. Adjusted mean differences (MD) and 95% confidence intervals (CI) for results in linear regression models were reported. Multiple imputations (20 imputed datasets) were used to replace missing data on covariates, where appropriate. Since there was only one case with a missing covariate in both the pre-awareness and post-awareness group, complete case analysis was used for the regression models of nonparticipants. In a sensitivity analysis, logarithmic transformation of the data on time to functional recovery and length of postoperative hospital stay was used to analyze the influence of right-skewness of data in the nonparticipant analysis. A p-value ≤0.05 was considered statistically significant. A more than one day reduction in time to functional recovery and length of hospital stay (pre-awareness versus post-awareness) were considered clinically relevant. Analyses were performed using IBM SPSS Statistics for Windows, version 21.0 (Armonk, NY, USA).