Abstract
Abstract
Background:
There are no studies reporting the rate of surgical site infection (SSI) after surgery for endometriosis, although this information is valuable when discussing the most appropriate treatment strategy with the patient.
Methods:
We conducted a prospective cohort study in a university hospital and regional reference center for endometriosis. We sought to measure the rate of SSI after endometriosis surgery using prospective SSI post-discharge surveillance data and the hospital information system via an ad hoc algorithm using both diagnosis and procedure code classifications.
Results:
Among 896 consecutive endometriosis surgical procedures, we identified 365 procedures with involvement of the gastrointestinal tract, defined as the deep invasive procedure (DIP) group, 107 procedures with involvement of an ovary, and 424 other procedures. Twelve SSI (all organ/space infections) were observed, all in the DIP group, corresponding to an overall SSI incidence of 1.3% 95% confidence interval (CI) 0.7–2.3, and an SSI incidence in the DIP group of 2.8%, 95% CI 1.5–4.9. The median delay between the procedure and the SSI was 6.5 days (range, 3–23). At least one micro-organism was found in 10 patients (four Escherichia coli, four Enterobacter cloacae, three Enteroccus faecalis, two Bacteroides fragilis, one Pseudomonas aeruginosa, one Candida albicans).
Conclusion:
A low overall rate of SSI after surgery for endometriosis was observed. Nevertheless, procedures with involvement of the intestinal tract were at risk of SSI.
Endometriosis is defined as the presence of endometrial glands and stroma outside the uterine cavity [1]. It affects 5%–10% of women in the general population and may cause pelvic pain and infertility [2,3]. Surgery is a therapeutic option, especially for women with a desire to have children [3]. Surgery may involve several organs (vagina, uterus, ovary) and sometimes the intestinal tract must be opened.
Establishing surgical site infection (SSI) rates for endometriosis surgery via SSI surveillance is difficult. First, current national surveillance systems, which rely on a surgical procedure code to identify the denominator of SSI surveillance [4], are not able to identify surgical procedures performed for endometriosis. This is because the inner logic of surgical procedure coding (in France via the Classification Commune des Actes Médicaux, CCAM) is to describe a surgical procedure using one main act performed on only one organ, whereas surgery for endometriosis includes various types of non-specific procedures, often associated with and involving several organs. Second, the CCAM classification lacks accurate content to describe the most innovative specific procedures. For example, rectal shaving [5], which is used frequently in endometriosis surgery in our center, does not have a corresponding CCAM code. To our our knowledge there are no studies reporting the rate of SSI after surgery for endometriosis, although this information is valuable when discussing the most appropriate treatment strategy with the patient.
The main objective of our study was to assess the incidence of SSI after surgery for endometriosis using SSI data from prospective post-discharge surveillance and surgical procedure data from the hospital information system (HIS). Secondary objectives were to assess if involvement of gastrointestinal tract was a risk factor for SSI and to propose an algorithm to identify surgical procedures performed for endometriosis, via both diagnosis and procedure diagnosis codes in the HIS.
Materials and Methods
A prospective proactive SSI post-discharge surveillance program was set up in the surgery department of our university hospital in 2010. Our hospital is a regional referral center for the treatment of endometriosis. All major gynecologic–obstetric surgical procedures are included in the surveillance program. This proactive surveillance (as opposed to passive surveillance relying on reporting by surgeons) consists of active routine search for possible cases of superficial, deep, or organ/space SSI. The infection control team prospectively detects possible cases using three sources of information: weekly review of bacterial sampling, weekly meetings with the nursing staff, and review of all re-admissions, revision surgery, or emergency department referrals during the 30 days after surgery. Bacterial sampling is performed as required by patient management based on the decision of the surgical team. There is no routine bacterial sampling for the purpose of surveillance.
Once per week, on the same day, the infection control nurse meets with gynecologic ward nurses and outpatient clinic nurses, who provide a list of patients with possible SSI during the previous week. This also allows us to detect superficial SSIs (which do not require re-operation) as well as organ/space SSIs. Each suspected case of SSI is validated with the corresponding surgeon, according to the French Ministry of Health definitions for SSI [6], similar to those of the U.S. Centers for Disease Control and Prevention [7]. Surgical site infection rates are communicated twice per year to surgical teams globally and for each of the main surgical procedure categories.
Identification of surgical interventions for endometriosis
As in most Western countries, funding for public hospitals in France is based on diagnosis-related groups (DRG) [8]. Each patient's stay is associated with one primary diagnosis code according to the International Classification of Diseases, 10th revision (ICD-10) [9]. Other secondary codes may be added. The letter C in the DRG code indicates that the stay includes a surgical procedure, meaning that a CCAM surgical code is associated with the stay. We used our HIS to identify patients who had a surgical procedure for endometriosis using the following search parameters:
Patients admitted to the gynecologic surgery ward between January 1, 2012 and December 31, 2015. A DRG containing the letter C. A primary disease code of endometriosis (ICD-10: N80.x) for the related hospital stay.
Classification of surgical interventions for endometriosis
Our hypothesis was that the extent of the surgical procedure and the depth of organ involvement in the surgical procedure could be a risk factor for SSI. Therefore, we classified surgical procedures in three groups of interventions according to the depth of the organs involved. Because of the multiple procedures performed during one surgical procedure for endometriosis and the limitations of the CCAM mentioned above, we used the nine ICD-10 subcodes, from N80.0 to N80.9 (N80.7 does not exist), which identify organs involved in endometriosis during a given stay to define the surgical procedure groups (Table 1), according to the following algorithm. If one of the diagnosis codes was N80.4 (rectovaginal septum) or N80.5 (intestinal tract), the procedure was classified in the deep invasive procedure (DIP) group; in absence of diagnosis codes N80.4 or N80.5, but presence of N80.1 (ovary), the procedure was classified in the ovarian procedure group; all other procedures were classified in the superficial procedure group.
Inter-Rater Agreement of ICD-10 Classification versus Operative Report (κ = 0.86) of Fifty Endometriosis Surgical Procedures
N80.0: Uterus, N80.2: Fallopian tube, N80.3: Pelvic peritoneum, N80.6: Skin, N80.8: Other, N80.9: No precision.
N80.1: Ovary.
N80.4: Rectovaginal septum, N80.5: Intestinal tract.
ICD-10 = International Classification of Diseases, 10th revision.
To assess the validity of this classification using ICD-10 codes, we drew a random sample of 50 stays and compared the results of classification via ICD-10 codes with classification via the operative report, used as gold standard. The operative report of each patient with SSI identified by post-discharge surveillance between 2012 and 2015 was reviewed to retain cases with a diagnosis of endometriosis. In addition, the following data were collected in the patient's computerized file: patient's age, the micro-organism involved in SSI, resolution of the infection, re-appearance of endometriosis pain, occurrence of a digestive complication, desire for pregnancy, and occurrence of pregnancy after surgery.
Details of surgical interventions for endometriosis
All endometriosis procedures were minimally invasive and none of the operations required a laparoconversion. Bowel surgery was planned routinely before the surgery, the technique to be used (shaving, disc excision, or segmental resection) was chosen on the basis of pre-operative imaging data, and a five-day residue-free diet was administered before the surgery. In rare cases when bowel surgery was required but not scheduled, the procedure on the bowel was postponed, and full imaging assessment of bowel involvement was performed before asking for the patient's informed consent. Because disc excisions were performed using trans-anal staplers, the procedure did not involve intra-abdominal bowel opening [10]. Segmental colorectal resection required an extra two minutes to open the abdominal bowel through a 5 cm suprapubic incision [11].
Each time a procedure requiring bowel suture (either disc excision or segmental resection) was scheduled, antibiotics were administered systematically according to the guidelines of the French Society of Anesthesiology and Intensive Care (SFAR): 30 minutes before the incision, cefoxitine 1 g (to be repeated if surgery longer than two hours). In women with allergies, metronidazole 1 g and gentamycine 5 mg/kg were used. The antibiotherapy was not prolonged after the surgery, except in rare circumstances, such as intra-abdominal bowel content spilling.
Analysis
The inter-rater agreement between classification according to ICD-10 codes and classification according to operative report was assessed through the Cohen κ coefficient. Surgical site infection rates were calculated together with their 95% confidence interval for all endometriosis procedures and for each of the three groups. Because two groups had no SSI events, we used the Clopper-Pearson exact method. All analyses were performed with R statistical software (version 3.4.4, R Foundation for Statistical Computing, Vienna, Austria).
Results
The preliminary assessment of agreement between ICD-10 coding classification and the operative report concluded an almost perfect agreement, with Cohen's κ = 0.86 95% CI 0.73,0.99 (Table 1). Among 896 consecutive endometriosis surgical procedures performed between January 1, 2012 and December 31, 2015, 365 were DIP (involvement of the gastrointestinal tract), 107 were ovarian procedures, and 424 were superficial procedures. The mean age of patients was 37.5 years (standard deviation [SD] 11.8).
Twelve SSIs were observed in 12 patients (Table 2). All were organ/space infections confirmed either by an abscess or collection noted on computerized tomography or during re-operation. The overall incidence of SSI was 1.3% (95% CI 0.7–2.3). All 12 SSIs were found in the DIP group, with an SSI incidence of 2.8% (95% CI 1.5, 4.9) in this group. No SSI was found in the superficial procedure group (95% CI 0, 0.9) or in the ovarian procedure group (95% CI 0, 3.4). The mean age of infected patients was 33.5 years (SD 6.3). The delay between surgical procedure and SSI varied between 3 and 23 days (median, 6.5). At least one micro-organism was found in 10 of the 12 patients. The median follow-up was 1 year (range 0.25 to 3). There were no deaths. Pain re-occurred in one patient and two patients experienced digestive complications (one underwent colostomy without restoration of intestinal continuity for two years and one had constipation). Five patients had a desire for pregnancy after surgery, one of whom became pregnant.
Description of the Twevle Endometriosis Surgical Procedures Resulting in Organ/Space Surgical Site Infections
SSI = surgical site infection.
Discussion
Principal findings
In this large study based on four years of exhaustive prospective post-discharge surveillance of endometriosis surgical procedures in a regional reference center, we report a low rate of SSI. We also showed that the identification of the denominator of SSI surveillance may be automatized using ICD codes.
Surgical site infection was observed only in the group with opening of the digestive tract. Micro-organisms identified in the SSI in our survey were consistent with this route of infection. The rates of SSI we observed in the DIP group were similar to those observed in France [12] for colorectal surgery in patients without comorbidities (National Nosocomial Infections Surveillance system [NNIS]-0). In the ovarian procedure group and the superficial procedure group, which represented two-thirds of our surgical procedures, no SSI was observed suggesting that surgery is a safe treatment in endometriosis populations. In addition, no major adverse events were reported among the 12 patients with SSI, except for the patient who underwent colostomy.
Strengths and limitations
Our study has some limitations. First, it was conducted in a single center and therefore SSI rates should be extrapolated to other centers with caution. However, it is possible that SSI rates may have been slightly overestimated in our work because our university hospital is a regional center caring for the most severe endometriosis cases. Second, we used ICD-10 diagnosis codes to classify surgical interventions, rather than CCAM surgical procedure codes, which were used only to confirm that surgical interventions had been done. However, we demonstrated that our classification was consistent with information available in operative reports and was reliable. This validation of an algorithm using both ICD-10 and CCAM code classifications could allow other centers to monitor SSI rates for endometriosis surgery, in an efficient manner. Third, our method was not able to detect SSIs treated in other hospitals. However, the fact that all patients had a post-operative visit in our hospital scheduled two months after their surgery, and that our hospital is a referral center for endometriosis, makes it unlikely that a SSI was not captured by the surveillance. The strengths of our survey are the large patient set, and the prospective proactive collection of SSI using data from multiple sources.
Conclusion
In this study we have shown that SSI rates were comparatively low after endometriosis surgery and follow-up of patients with infection was favorable, suggesting that the risk of surgical infection is acceptable. Our survey also suggests the need for more precise identification of risk factors associated with occurrence of SSI within the DIP group. We will conduct a nested case control study to identify these risk factors.
Footnotes
Acknowledgments
The authors are grateful to Nikki Sabourin-Gibbs for her help in editing the manuscript.
The datasets analyzed during the current study are available from the corresponding author.