Tools and scores for perioperative pulmonary, renal, hepatobiliary, hematological, and surgical site infection risk assessment: an update

ABSTRACT Introduction: perioperative risk assessment is essential to mitigate surgical complications, which suggests individual and collective interest since the number of surgical procedures in Brazil has been expanding steadily. The aim of this study was to summarize and detail the main calculators, indexes and scores regarding perioperative pulmonary, renal, hepatobiliary, hematological and surgical site infection risks for general non-cardiac surgeries, which are dispersed in the literature. Method: a narrative review was performed based on manuscripts in English and Portuguese found in the electronic databases Pubmed/MEDLINE and EMBASE. Results: the review included 11 tools related to the systems covered, for which the application method and its limitations are detailed. Conclusion: the non-cardiovascular perioperative risk estimation tools are beneficial when disturbances are identified in the preoperative clinical examination that justify a possible increased risk to the affected system, so the use of these tools provides palpable values to aid in the judgment of surgical risk and benefit as well as it identifies factors amenable to intervention to improve outcomes.

This article aims to address the dispersion of non-cardiovascular perioperative risk assessment tools by synthesizing the main calculators, indices, and scores regarding perioperative pulmonary, renal, hepatobiliary, hematological, and surgical site infection risks for general, non-cardiac surgeries.

METHODS
This is a narrative review carried out by searching the Pubmed/MEDLINE and EMBASE electronic databases for manuscripts in English and Portuguese.
We chose this method because it has the advantage of allowing the aggregation of different elements -that is, the risk assessment tools for several systems -within a single text.However, we should note that the format is susceptible to subjectivity.

Pulmonary Risk
The assessment of the risk of pulmonary complications for a long time remained undervalued to the detriment of cardiac risk in surgeries.This has changed in recent decades, since it has been identified that pulmonary complications can occur with a frequency similar to cardiac complications and determine a longer hospital stay, in addition to both often occurring concomitantly [6][7][8] .
Perioperative pulmonary complications can be of different nature [8][9][10] , but the most clinically relevant are atelectasis, pneumonia, respiratory failure, and exacerbation of underlying chronic lung disease 11 .For these perioperative outcomes, multiple risk factors have already been demonstrated, namely, type of surgery, advanced age (over 60 years), chronic obstructive pulmonary disease, smoking, heart failure, functional dependence, ASA classification, obesity, impaired consciousness (confusion, delirium, but not dementia or chronic mental illness), abnormal findings on chest examination, alcohol use, and weight loss 9,11,12 .
Several indices and risk calculators for perioperative complications in this system have been developed.
They differ in the variables considered, outcomes, and populations studied and no tool is suitable for all situations [13][14][15][16][17][18][19][20] .Due to their good clinical applicability and risk estimation capability, we highlight the Respiratory Failure Risk Index 18 , the Postoperative Pneumonia Risk Index 19 , and the Assess Respiratory Risk in Surgical Patients in Catalonia (ARISCAT) 20 , described below.
The Respiratory Failure Risk Index was proposed in 2000, developed from a prospective cohort of 81,719 patients undergoing noncardiac surgery, and validated on a set of 99,390 patients.The outcome was postoperative respiratory failure, defined by use of mechanical ventilation for more than 48 hours after surgery or reintubation with mechanical ventilation after extubation.This outcome is relevant insofar as, although uncommon, it incurs a substantial increase in 30-day mortality -in this study, 27% of mortality occurred in patients who had postoperative respiratory failure and 1% in those who did not.
The main limitations of the study were the exclusion of female patients, a population with a high prevalence of comorbidities, and the absence of pulmonary function tests prior to surgery.The index is performed by scoring each risk factor, and the sum categorizes the patient into one of five classes with increasing risk of postoperative respiratory failure, as seen in Tables 1 and 2 18 .Although the index considers the score of only some types of surgery as risk factors for respiratory failure, all types of non-cardiac surgery performed under general, spinal, or epidural anesthesia were included in its development, except for transplants and procedures with very low mortality, such as dental procedures, endoscopy, and central venous catheter insertion.The outcome is postoperative pneumonia, which is relevant due to the increase in 30-day mortality -in the study, patients with postoperative pneumonia had a 21% mortality rate versus 2% in patients without it.Among its main limitations is the population with a high prevalence of comorbidities and low participation of female patients (3.2%), which may limit its usefulness for more heterogeneous and healthy populations.In addition, the list of risk factors that score in the index is lengthy, which can make its use difficult.All surgeries in which there was general, epidural, spinal, or local anesthesia were included, excluding only transplants and surgeries with very low mortality.Possibly the small proportion of participants undergoing orthopedic procedures may have underestimated the risk of this type of surgery.The index is determined by scoring risk factors and categorizing the patient into 5 classes (Tables 3 and 4) 19 .5 and 6 20 .
Acute kidney injury (AKI) is not uncommon in the perioperative period.It occurs in around 12% of major elective surgeries, ranging from 1% to 18.7% depending on the type of procedure, patient risk factors, and perioperative management [25][26][27][28][29] .Postoperative AKI contributes substantially to the increase in hospital expenses 30 and, even when mild, constitutes a risk factor for prolonged stay in the intensive care unit (ICU) 25 .Renal risk indices that have been proposed for general non-cardiac surgery focus on predicting the probability of postoperative AKI 29,31 .
We highlight two of these tools here.The General Surgery Acute Kidney Injury Risk Index was proposed in 2009, based on data from 75,952 surgeries in the United States, found 11 independent risk predictors (grouped into nine items), and estimated the risk in five classes, as seen in Table 7 29 .Among its limitations are the exclusion of vascular, urological, ophthalmological, obstetric, and cardiac procedures, and the lack of data on intraoperative hydration.
In 2019, the Simple Postoperative AKI Risk (SPARK) was published, based on data from 79,518 patients, seeking to predict AKI.The SPARK uses a score according to Tables 8 and 9 and is suggested for surgeries in which the patient is stable and there is no specific risk that requires a more detailed assessment (nephrectomy, for example) 31 .This tool is limited to include only general, orthopedic, gynecological, obstetric, urological, and neurosurgical procedures, with reduced performance in the last two.It was also developed in an Asian population, in South Korea, which may be a limitation when applied to more heterogeneous populations.

Kidney risk
Renal dysfunction, generally measured by serum creatinine and estimated glomerular filtration rate, increases perioperative complications and mortality, especially cardiovascular complications 21,22 .The use of dialysis 23 and the presence of acute renal failure 24 also contribute to an increased risk of perioperative complications.
Both the General Surgery AKI Risk Index and the SPARK excluded patients with previous chronic kidney disease and did not consider the perioperative use of nephrotoxic agents.

Hepatobiliary risk
The literature on perioperative hepatobiliary risk estimation focuses on the identification of increased risk of morbidity and mortality in cirrhotic patients [32][33][34][35][36][37][38][39][40] , given that the safety of elective surgeries in patients with mild chronic liver disease has already been reported 41,42 .Acute hepatitis substantially increases the perioperative risk, such that in these cases clinical treatment is recommended, and surgery is postponed whenever possible [43][44][45] .In cirrhotic patients, perioperative mortality is around 7-9% in elective surgeries 33,38 .Two tools initially developed for different purposes were successfully adapted to estimate the risk of perioperative mortality in cirrhotic patients: the Child-Turcotte-Pugh score and the Model for End-stage Liver Disease (MELD).
The Child-Turcotte-Pugh (CTP) score was initially used to identify the level of liver dysfunction and the severity of portal hypertension in cirrhotic patients 46 .
Despite being old (1973), the score started to be used for risk estimation and remained with good accuracy over time 44 .It is composed of five stratified parameters, to which a score is assigned, as shown in Tables 10 and   11.Among its limitations, the CTP score has the possible variation between evaluators (in the encephalopathy and ascites items) and the small sample in the studies that used it as a risk estimation method.Studies that identify the CTP score as a risk estimation model or as a factor associated with increasing mortality according to patient classification include different types of surgery: general abdominal surgery 33,37 , neurosurgery, head and neck, ophthalmologic, facial, thoracic, vascular, urological, and gynecological surgery 35 .
MELD was proposed in 2000 with the aim of predicting mortality after transjugular intrahepatic portosystemic shunt (TIPS) 47  with MELD <8, to 54%, when >15 33,36,38 .The guidelines of the American Society of Gastroenterology define that MELD <16 expresses a reduced risk in relation to higher values 48 , and MELD is associated with long-term postoperative mortality 36 .Among the types of surgeries considered to identify the relationship between MELD and perioperative mortality are general abdominal 33,36,38 , orthopedic, and cardiovascular 36 surgeries.
increased perioperative bleeding without the need for additional tests 49 .Thus, the main hematological condition that benefits from risk stratification by scores is venous thromboembolism (VTE).

The incidence of VTE (deep vein thrombosis -
DVT -and pulmonary thromboembolism -PTE) has had little variation over the years despite the evolution in treatment and prophylaxis 50,51 .Its occurrence substantially increases the costs of hospitalized patients 51 , and it is the leading cause of preventable in-hospital death 52,53 .

Hematological and thromboembolic risk
In the hematological evaluation of the pre--surgical patient, one mainly seeks to identify anemia and coagulation disorders.Anemia is directly clarified by the blood count and other complementary tests, without the application of specific risk scores.A structured clinical history is usually sufficient to exclude the risk of Table 13.This sum will classify the patient at low, medium, or high risk of VTE, as seen in Table 14 64 .Despite being a well-structured formulation, this tool has a lengthy application and lacks external validation 58 .

Surgical site infection
Surgical site infections (SSI) represent an important fraction of nosocomial infections in surgical patients and may be responsible for 38% of these infections depending on the scenario analyzed, as exposed by Malone et al. in a study with 6,301 North American individuals, mostly male (95%), undergoing non-cardiac surgery 65 .SSI occur in 1.2-3.9% of surgeries [65][66][67] , which may vary depending on external factors such as adequacy of the antibiotic prophylaxis protocol [68][69][70] , surgeon's experience in specific procedures 71 , and operation site, with a tendency towards a higher incidence in developing countries 72 .15.Each risk factor correspond to 1 point to the index and patients with 0 points are considered low risk, patients with 1 point, moderate risk, and high risk with 2 points or more.The study states that the index is capable of adequately predicting 90% of SSI cases, with greater precision than the traditional wound classification system 74 .
In 1991, an adaptation to the SENIC project index was proposed, also in the United States: the Surgical Site Infection Risk Index, based on the National Nosocomial Infections Surveillance (NNIS) program.In this index, like the previous one, each risk factor corresponds to 1 point to the patient's total score.The factors considered are: 1) the ASA rating (1 point if ≥3); 2) wound classification by the traditional system (1 point if contaminated or infected) and; 3) duration of surgery (1 point if above the 75 th percentile -the study provides a table with cut-off times for each type of procedure, in hours).For patients with 0, 1, 2, and 3 points, the incidence of SSI estimated by this tool is 1.5%, 2.9%, 6.8%, and 13%, respectively 73 .
More recently, in 2013, the Surgical Site Infection Risk Score (SSIRS) 67

FINAL CONSIDERATIONS
Preoperative risk assessment is usually focused on the cardiovascular system, may benefit from extending the analysis to other systems.In practice, these other   The identification of factors considered in the composition of scores and calculators can also guide perioperative interventions aimed at risk reduction.For example, smoking is one of the factors that comprise the calculation of the risk of surgical site infection according to SSIRS 67 .Therefore, in elective surgeries, previous smoking cessation, in addition to the known systemic benefits, may reduce the incidence of surgical site infection.
This way, the physician who evaluates the patient in the preoperative period can choose to use the most appropriate tool, as deemed necessary, considering its method of application herein described, and its ad-vantages and limitations, as also presented throughout this work and summarized in Table 16.Great attention should be paid to the surgical specialties that were excluded from the development of each tool to know the applicability of the risk estimation method for the type of surgery to be performed.
One of the limitations of this study is the subjectivity inherent to the narrative review model.The model was chosen because it allows the aggregation of different topics about the use of tools for non-cardiac perioperative risk assessment in a single text.However, it is impossible to exclude some subjectivity in the selection and interpretation of the bibliography contained herein.
systems' risk estimation tools will be beneficial in patients in whom, by clinical examination, alterations are identified that raise the possibility of increased risk.For exam-ple, in cirrhotic patients, the help of the CTP 46 or MELD 47 tools is of interest insofar as they provide tangible values for surgical decision making, considering risk and benefit.Tool (year of publication) Benefits Limitations

Pulmonary risk
Respiratory Failure Risk Index (2000) 18 Simple application (sum of points).Considers type of surgery.
Only included male patients.Evaluates single outcome Postoperative Pneumonia Risk Index (2001) 19 Simple application (sum of points).Considers type of surgery.
Validated for population with high prevalence of comorbidities and low female participation.Evaluates single outcome ARISCAT (2010) 20

Simple application (sum of points). Evaluates multiple outcomes
Relatively small sample.Validated for specific population (Catalonia) Kidney risk AKI Risk Index in General Surgery (2009) 29 Simple application (sum of points).Considers the operated site.
Vascular, urological, ophthalmological, and obstetric procedures excluded.Did not evaluate the use of nephrotoxic agents and intraoperative hydration.

Tool (year of publication) Benefits Limitations
Simple Risk of Postoperative AKI -SPARK (2019) 31 Simple application (sum of points).Uses several factors that can be preoperatively adjusted.

Includes obstetric procedures
Reduced performance in urological and neurosurgical procedures.Did not evaluate the use of nephrotoxic agents.Developed in a specific population (South Korea).
Possible variation between evaluators (degrees of ascites and encephalopathy).
Few types of surgery (abdominal, orthopedic, cardiovascular) Hematological and thromboembolic risk Caprini (2005) 59 Cut-off point indicating chemoprophylaxis for VTE.Considers the type of surgery.
Lengthy application (37 items to consider) Rogers (2007) 64 Considers the type of surgery.
Lengthy application (23 items to consider).
Old, developed from hospital records from the years 1970-1976 in the United States.
SSIRS (2013) 67 Simple application (digital calculator).Targets the specific type of surgery to be performed, including all specialties.
Risk in unclean wounds can be underestimated by the use of antibiotics.Of the 14 factors considered, only 2 are potentially modifiable (smoking and surgery time)

A
well-established and old way of classifying operative wounds was developed by the US National Research Council in 1964, grading into clean, potentially contaminated, contaminated, and infected.Despite widely known, this traditional wound classification system has limited accuracy in risk estimation, especially as it does not consider factors intrinsic to the patient 73 .Thus, some tools for SSI risk estimation were developed.In 1985, an American study with 58,498 patients was published proposing a risk index for surgical site infection, as part of the Study on the Efficacy of Nosocomial Infection Control (SENIC) project.This index estimates the risk of superficial or deep SSI based on four risk factors, as shown in Table was published, based on a derivation and validation samples from approximately 180,000 patients.This model estimates the risk of superficial or deep surgical site infection up to 30 days after the procedure.The risk is estimated through a calculator considering the characteristics of the patient and the surgery, available in English at: http://www.ohri.ca/SSI_risk_index/Default.aspx.The calculator asks for information on smoking (yes or no), weight, height, medical history (peripheral vascular disease, metastatic cancer, use of corticosteroids for at least 10 days, systemic inflammatory response syndrome, or sepsis in the last 2 days), and surgery (inpatient or outpatient, emergency or not, wound classification, ASA class 1 , general anesthesia or not, additional procedure, surgery time, and type of surgery).

Table 1 -
18sk factors and postoperative Respiratory Failure Risk Index scores18.
published in 2001, developed from medical records of 160,805 patients, and validated by data from 155,266 individuals, all of whom underwent noncardiac surgery.

Table 2 -
18asses, scoring and estimated risk by postoperative Respiratory Failure Risk Index18.

Table 3 -
19sk factors and scores for the Postoperative Pneumonia Risk Index19.

Table 4 -
19asses , scoring and estimated risk by the Postoperative Pneumonia Risk Index19.

Table 5 -
20stoperative pulmonary risk factors and scores according to the ARISCAT index20.
ARISCAT: Assess Respiratory Risk in Surgical Patients in Catalonia.
ARISCAT: Assess Respiratory Risk in Surgical Patients in Catalonia.

Table 8 -
31eoperative risk factors for renal risk and their SPARK scores31.

Table 9 -
Classes and estimated Renal risk according to SPARK Index31 .
. It is calculated by the following formula, approximating the result to the nearest whole number: MELD = 3.78 x ln(serum bilirubin in mg/dL) + 11.2 x ln(INR) + 9.57 x ln(serum creatinine in mg/dL) + 6.43.INR is the International Normalized Ratio obtained from the prothrombin time.The literature indicates that the perioperative 30-day mortality predicted by MELD ranges from 5.7%,

Table 13 -
64sk factors for venous thromboembolism according to the Rogers score64.

Table 14 -
64sk levels for venous thromboembolism according to the Rogers score64.

Table 15 -
74rgical site infection risk index of the SENIC project, 198574.

Table 16 -
Main advantages and limitations of risk assessment tools by system.
ARISCAT: Assess Respiratory Risk in Surgical Patients in Catalonia; AKI: acute kidney injury; SPARK: Simple Postoperative AKI Risk; MELD: Model for End-stage Liver Disease; VTE: thromboembolism venous; SENIC: Study on the Efficacy of Nosocomial Infection Control; SSIRS: Surgical Site Infection Risk Score.