A distinctive feature of this study is the strict selection of patients. Some 238 patients were discarded to obtain a homogeneous cohort of pure AAC21,22. Many source papers used by the TG18 suffer from multiple etiologies pooling10,19−21 or only exclude the chronic cholecystitis subgroup8, undoubtedly being a source of statistical noise. Some reports on ACC perform some intents of refined patient selection23, but a strict approach has not been used before, and we believe it is essential.
The current study reports on the experience meticulously registered in a surgical emergency unit where NST has not been adopted as recommended by TG18. In fact, we opted for NST in 5% of patients, whereas 40% of the series were classified as TG grade III. This attitude has been adopted by a significant number of groups with ample clinical experience in ACC management13,24−27.
In the current series, the distributions of TG18 severity classification were clearly more slanted toward severity than the majority of other series of non-selected ACC where the severe proportion oscillates between 5% and 19%3,12,27,28. This plethora of severe is mainly due to higher proportions of Kidney and Neurological dysfunction, and/or international normalized ratio ≥ 1.5 among our patients.
Mortality
According to the TG18, the accepted mortality for acute cholecystitis should be < 1%7. In the current study, the 30-day mortality was substantially higher (3.6%). Large multicenter studies reported a mortality rate from 0.6–13.5% 4,10,29,30. Nevertheless, a fifth of the 26 patients who died in our series did so after the standardized 30 days to calculate postoperative mortality. The mortality rate not including these patients would drop to 2.8%. Furthermore, if chronic cholecystitis and post-endoscopic retrograde cholangiopancreatography cases were included in the total count, as is routinely done in many series, the mortality would have been wrongly reported at 1.8%, missing almost half of the deceased due to complications after ACC was diagnosed and treated. In contrast, our series did not include acalculous or postoperative cholecystitis; two etiologies with very low prevalence but higher mortality and morbidity33.
Advanced age, high ASA score, and CCI are almost universally cited as mortality risk factors, not only for ACC but for any emergency procedure12,34,35. However, not every discrete comorbidity factor of the CCI carries the same relative risk for mortality, their ORs ranging from 1 to 12, as depicted in Fig. 3. That is why we chose to “deconstruct” the CCI and calculate the contribution of each component of the specific population of ACC patients.
Identifying the high-risk patient
Regarding reducing mortality, surgeons have been struggling to identify patients too frail and/or too severely affected, who will be better served without surgery. In this context, we label this patient as a high-risk patient. The standardization and use of therapeutic algorithms that include the preoperative assessment of surgical risk mortality rate of patients admitted with a diagnosis of acute cholecystitis is essential. Hypothetically, in some high-risk patients at admission, their general condition may be improved with medical treatment, and a risk evaluation carried out 24–48 h after admission could reassign the patient to the group of early ST.
Some authors consider high-risk patients as those with failure of at least one organ or multiple organ dysfunction36. Patients with an ASA III or IV score have an expected postoperative mortality of 5–27%, are considered high risk for cholecystectomy37, and have higher morbidity9,38. In contrast, in a prospective observational study, González-Muñoz et al. found that patients with ASA > II and only medical treatment had a mortality of 17%, whereas those operated upon early had no mortality8,11.
In the current series, nearly 40% of patients would have been labeled as high risk. However, we believe that we should not exclude them from early surgery.
Age alone appears in some studies as an independent surgical risk39. However, age by itself does not increase mortality. Decreased functional reserve capacity in addition to comorbidities, usually but not always linked to age, are the main risk factors for mortality39. Consequently, although age per se is not regarded as a contraindication for early cholecystectomy, it has been widely recognized that the use of frailty and surgical risk scores could contribute to achieving the best clinical judgment in elderly people.
There is no consensus on which of the already available surgical risk scores better predicts postoperative mortality in acute cholecystitis. APACHE II > 1540, CCI > 612, and P-POSSUM > 4010,11 have all been used to support ST. None of them were specifically created for ACC; therefore, none of them are particularly advantageous over the others, and none of them offer an outstanding prediction ability.
We found that the individual weight of each variable included in the CCI was quite different when studied in a cohort of ACC patients (Fig. 3).
Acute Cholecystitis Mortality Estimation
By selecting variables with a greater impact on the univariate analysis, we devised the Acute Cholecystitis Mortality Estimation (ACME). The model retained a set of four variables which accurately predicted mortality better than the TG classification (Fig. 4). Using a similar approach, Fagenson et al. chose the modified Frailty Index to identify patients with the worst prognosis after cholecystectomy. They also opted for extracting the more significant CCI components and ended up with a very similar AUC of 83% for predicting mortality14.
Treatment modalities
Patients with initial NST experienced a mortality six times higher than those initially selected for initial ST. Differences in mortality between patients initially ascribed to exclusive antibiotic therapy or cholecystostomy were not significant. Patients who switched from NST to ST (n = 9) had an overall mortality of 11%. Likewise, a massive observational prospective study by Endo et al., compared four strategies (antibiotic therapy, cholecystostomy either as a definitive treatment or as a bridge to surgery, and early cholecystectomy). Their results supported early cholecystectomy over any other strategy, with or without cholecystostomy41. ST is the first option in the management of ACC crystallized in the CHOCOLATE trial, which encouraged early laparoscopia cholecystectomy over cholecystostomy40.
The complexity of the course of treatment, with subgroups of NST patients requiring rescue cholecystostomy and those in turn that finally required rescue cholecystectomy is reflected in Fig. 1. In every non-randomized trial, there are patients migrating from the four initial treatment groups (to be analyzed by intention-to-treat) to treatments finally received (to be analyzed per protocol). This circumstance undoubtedly contributed to the long period of uncertainty surrounding the best therapy option for high-risk patients with ACC.
Propensity Score Matching: Comparison among subgroups of similar severity
This technique has been used occasionally in the cholecystectomy series42,43. In the current study, when comparing high-risk subgroups with the same preoperative morbidity and ACC severity selected by PSM, the mortality was higher in the NST group than in the ST group (26.2% vs. 10.5%), probably reflecting that we reserved the NST for patients with more severe ACC.
Limitations of the study
The retrospective nature of this study is undoubtedly its main limitation. In contrast, the limited application of NST makes this cohort valid to determine risk factors for postoperative complications but less solvent when determining the best strategy for severe ACC in the common range for a series of consecutive patients in a single center 10,38.
The patients in this study had more advanced disease, the severe ACC is being overrepresented in comparison with most other series which can bias the conclusions toward a higher mortality estimation. However, we believe that this population composition makes this analysis more robust in assessing risk factors for mortality.
The follow-up was limited to 30 days after the intervention or until the patient's discharge. More extensive monitoring would likely discover a greater number of complications9. Nevertheless, we reported mortality directly related to the ACC taking place beyond the 30th day of admission, which is routinely missing from other reports circumscribed to 30-day mortality.