Surgical risk assessment should be performed by a surgical team seeking a clear perspective not only for possible clinical outcomes but also in order to give patients and families a clear standpoint in terms of expectations [2]. Raw mortality and morbidity, though commonly reported, can be misleading due to differences in preoperative and intraoperative findings of patients [6, 15]. Although the good surgical technique is paramount in reducing adverse outcomes, the final result is a compound where the physiological state of the patient, operative severity, and peri-operative support services intervene [7]. Patient evolution and surgical complications depend on several factors: the quality of the surgical team, physiological status, and surgical stress among others [5]. Even though POSSUM and P-POSSUM scales have been widely used in Europe, their spread through America has not been equal [4].
The P-POSSUM scale was created in order to correct the elevated mortality risk obtained by POSSUM [3]; some studies showed an increased accuracy of P-POSSUM over POSSUM in mortality prediction, with near results with the observed mortality [9,10,11,12, 16]. Given that P-POSSUM has to be correlated to the general condition of the local population for it to be effective [9,10,11,12,13], variability of results obtained is explained by lack of adjustments to population baseline characteristics that generate changes in predicted values. Among these characteristics are malnutrition and economic status; nevertheless literature, as our study, reports P-POSSUM scale has an increased accuracy to predict mortality compared with POSSUM in emergency surgery [12].
We assessed risk prediction in terms of morbidity and mortality by POSSUM and P-POSSUM scores. 61.7% of the patients underwent emergent surgical procedures with a moderate operative score, suggesting an increased risk of morbidity in the first 30 days after surgery. Nonetheless, actual morbidity in our population was significantly lower than predicted (14.2%), most of them classified as Clavien-Dindo II (4.6%) and IIIB (6.6%) were 46.6% of patients required a second surgical procedure, matching Gonzalez-Martinez et al. findings which sustained mortality and morbidity values calculated by POSSUM are overestimated [1, 17]. Stonelake et al. described 86 patients who underwent urgent laparotomy with a 77.9% of morbidity [2, 14], results higher than those found in our population (14.2%). Nonetheless, in terms of mortality, in our study, a higher mortality rate was documented compared to Stonelake et al. results (7.1% vs 5.81%) [2, 14]. This could be due to the not only surgery findings but also to the previous clinical status of the patients involved in our study, reflected in their higher physiological and surgical values.
Melendez et al., analyzed 513 patients for risk prediction using the POSSUM scale finding overestimated morbidity and mortality rates compared with actual values of 17.56% vs 10.33% and 4.5% vs 1.75% respectively [3, 18]. Additionally, P-POSSUM was more accurate, with a predicted mortality value of 1.6% vs an observed value of 1.75% [3, 18]. Mortality and morbidity outcomes in the Melendez cohort showed that the overestimation persisted to a lesser extent, which relates to our findings where values of mortality calculated by POSSUM score overestimated mortality rates and P-POSSUM score estimations had results closer to real (mortality rate 7.1% Vs P-POSSUM predicted mortality 6.31%), with a Pearson's coefficient 0.948 showing clear statistical correlation (p < 0.000). Additionally, most of the studies find discrepancies between expected and observed mortality with the use of POSSUM, but similar with P-POSSUM [1].
Despite the effectiveness of POSSUM and P-POSSUM scores, improvements are still needed because of the overinterpretation in morbidity and mortality, specifically in high complexity surgeries [8, 19,20,21,22]. Among the limitations of this study is its retrospective nature, data was obtained from previously collected clinical charts which may result in missing variables for some of the observations. Thus, more studies are needed to confirm our results.