With the aim of understanding whether we have overused or inappropriately applied OA, we reviewed our series and compared all the EL performed in our center for non-traumatic issues. Three-hundred-twenty patients underwent EL for non-traumatic causes in 7 years: 153 (almost half of them) were treated with an OA procedure.
No significant differences were found in age, sex, and BMI distribution, but most of the other baseline characteristics diverged significantly. A relevant difference existed among indications to surgery: the largest part of OAs was performed in case of GIP (30,7%), for anastomotic leakage or intra-abdominal collections (13,7%). Consequently, the average MPI was 25 and the 47,7% had MPI > 26. On the contrary, the majority of DCs were performed in case of BO.
At univariate and then multivariate Cox regression, CFS (12% more), BI (a risk 5 folds higher), and no-resection (almost 3 folds) resulted being independent risk factor for mortality.
These results are not surprising, since already available in current literature and explored in another cohort study from our institution confirming that CFS relates to poor outcomes and higher mortality in OA [19]. Furthermore, it is well known that intestinal ischemia increases mortality and current guidelines on OA recommend its use in that setting [4, 5, 7]. Patients in our series treated with OA and not resected were those with more severe ischemic conditions: this explains the correlation with the higher mortality.
Considering complications, 5 cases of EAF were reported in the OA series (the 3,9%) and 1 in DC. EAF is considered the most frightening complication of OA and our rate is comparable with current literature. Of the patients complicated with EAF, three had necrotic-hemorrhagic pancreatitis and died during the ICU stay, the other two were reconstructed after prolonged treatment with NPWT fascial system and isolation of the fistula. Some results have been published about the use of OA in abdominal sepsis [20, 21], the factors influencing the onset of EAF and the differences in indications and treatments across different continents [20, 21]. Also on this topic, our rate is comparable with other literature findings [3, 22, 23].
Stratifying our populations by complications, OA did not result affecting mortality. Only transfusion rate and ICU_LOS were associated with mortality, regardless of OA or DC treatment. In our experience, with this result we sustain the safety of OA technique following an emergency laparotomy for non-traumatic issues. On the other hand, we are aware that OA patients spent more time in ICU and ICU_LOS is an independent risk factor for complications. In our center only a minority of patients need ICU hospitalization: stable patients who do not require intensive or hemodynamic support were hospitalized in the ward with major cost-effectiveness and minor risk for complications. This is related to OA not being adopted for physiological instability but to better control the contamination source or to guarantee a relook surgery for ischemic bowel loop. Because of that, most of those patients did not need an adrenergic support or to keep invasive ventilation and ICU assistance.
Concerning the higher rate of transfusions in the OA group, this result could be related to a poorer status of patients who required blood support: in fact, among a total of 36 transfused patients, 14 (39%) had MPI > 26, 20 (56%) were affected by cancer and 26 (72%) were classified as ASA 3/4.
A subgroup analysis was conducted to look for the advantages of OA in term of survival in specific populations. In GIP and MPI > 26 groups, OA did not affect neither improve survival, thus resulting non-inferior to DC. The same, as one could also intuitively deduce, could be concluded for BO.
The most interesting results were obtained from patients with BI. OA management, significatively and independently reduced mortality compared to DC strategy. The effectiveness of OA in BI, in fact, is already stated by the current guidelines [24] and, despite the small number of the sample in this subset, also in our series its biological rationale is well represented: two thirds of patients (24/36) were resected at the index surgery and, of these, one third underwent a further resection at the second surgical look; in addition, 50% of patients who were not resected during the EL received bowel resection during second look.
Looking at 90-days survival it resulted being 50.8% for DC and 60.8% for OA, but with no statistical significance. Our in-hospital OA mortality (around 33%) is in line with the one reported in other studies [25, 26].
Further interesting insights can be extracted from the data of our study. For example, the great majority of TAC techniques used on our patients were NPWT systems (commercial VACs, handcrafted ones and combinations of negative pressure and mesh) and almost 80% reached fascia closure. The median time of OA was 3 days (range:1–29 days), well below the advocated 4–7 days of early closure [27]. The best results in literature are obtained with NPWT associated with dynamic therapy (around 90%) [28, 29]. Beyond materials and re-interventions, OA patients had longer ICU (median of 4 vs 0) and hospital stays (median 24 vs 10, respectively) than DC patients, underlining that this management is resource intensive.
Many studies have investigated the indication for OA in diverticular perforation and related septic shock. An interesting metanalysis selected 7 papers concluding that DCS and OA may increase the anastomosis rate in case of bowel perforation [30]. In our series we are not able to support this assertion. In fact, the ostomy rate was higher in OA group, maybe due to the contamination and BI. Our rate of Hartmann’s procedures was 12,4% in OA vs 4,1% in DC, while the loop ileostomy rate was quite similar amongst groups. Obviously, having these patients most likely a higher risk of leak, the surgeon is forced to make a safety decision and OA is usually chosen as a salvage strategy.
A similar bias could be considered concerning the patient’s age, being riskier the choice to perform an intestinal anastomosis in the elderly. Finally, although our data are not enough to support definitive conclusions, in our experience OA does not appear helpful for reducing ostomy rates.
Further major limits are connected to the retrospective design of this study, and to its monocentric nature that could pose a bias in selection, clinical and surgical management of the patients, although the selection bias mitigation operated by the IPW analysis.