While exploring why patients receive intensive and intermediate care after surgery,1,2 the lack of objective quantitative measures of patients' pre-operative medical conditions that can be applied to all adult patients undergoing the many types of modern surgery became evident. The commonly used pre-operative assessment tool, the American Society of Anesthesiologists (ASA) Physical Status Classification,3 is used to assess all types and ages of surgical patients. It utilizes a simple categorical score to evaluate the severity of non-surgical pre-existing disease. Despite being a subjective score with a fair degree of inter-observer variability, it has repeatedly been found to be capable of appropriately classifying the physical condition of pre-operative patients and assessing their risks for post-operative morbidity and mortality.4,5 However, the ASA classification has limitations, especially, its inability to adequately classify patients without serious pre-existing illnesses who are suffering from acute surgical illnesses or injuries. A classic example is the seriously injured, but otherwise healthy, young trauma victim, who is classified ASA Class IE. However, this designation does not reflect the victim's multiple major injuries.
The ASA classification's limitations were the impetus for the development and evaluation of a new assessment tool, the Pre-operative Therapeutic Intensity Score (P-TIS). The P-TIS is based on the concept pioneered by Cullen et al.6 of evaluating the intensity of therapeutic interventions. The intended use of the P-TIS is mainly as a health services research tool for assessing pre-operative resource utilization. To evaluate the use of the P-TIS, the goal of this study was to examine the hypothesis that the P-TIS can be used to evaluate whether a patient would receive post-operative intensive or intermediate care, that is the greater the number and intensity of pre-operative interventions, the greater the need for post-operative ICU or intermediate care. In addition, the P-TIS was compared with currently available evaluation methods, specifically the ASA classification3 the Charlson Co-Morbidity Index7 and the Revised Cardiac Risk Index.8 These comparisons with existing scores, which were primarily designed to assess surgical risk, aimed to determine whether the P-TIS complements or duplicates these other pre-operative assessment methods.
Materials and methods
The initial phase involved developing P-TIS, whereas subsequent phases evaluated the score in the clinical arena.
The goal was to develop a scoring system that objectively quantifies the extent of pre-surgical therapeutic interventions, such as medication use, physiological monitoring and resuscitation procedures. Its major intended use is as a health services research tool to quantify pre-operative resource utilization. The score was to be applicable to all types of adult surgical patients and procedures and to be usable across institutions. P-TIS was not intended to be used for estimating operative risk or predicting peri-operative outcome.
P-TIS uses the same scoring scheme as the Therapeutic Intervention Scoring System (TISS), assigning 1–4 points to commonly used therapeutic interventions based on care intensity.6 Low-intensity interventions are assigned 1–2 points, whereas higher intensity interventions are assigned 3–4 points. The criteria for including specific interventions, diagnostic procedures and therapies were their importance for maintaining homeostasis. This included drug therapies (e.g. for hypertension, heart disease and diabetes), electrical therapies for cardiac disease (e.g. pacemakers, implantable automatic cardiac defibrillator, recent cardioversion), physiological monitoring (e.g. continuous electrocardiographic monitoring) and emergent biochemical testing (e.g. arterial blood gases). In addition, the extent of assistance with activities of daily living (ADLs) was scored.
The development process used a modified Delphi method that began with an initial draft prepared by the investigators that included some elements of the TISS6 and Intraoperative Therapeutic Intervention Score (I-TIS9), plus interventions important in pre-operative assessment, for example use of oral anti-hypertensive agents. The draft score was reviewed by a multi-institutional group of attending anaesthesiologists, internists and surgeons who commented on the interventions and their point values. Their input was also sought concerning additional interventions to be included in the score. To assist with this assessment, the group was provided with a number of scenarios to score. Examples included a healthy patient scheduled for a laparoscopic cholecystectomy, an elderly gentleman with significant cardiac and pulmonary disease scheduled for an elective colectomy and a teenager with multiple injuries (abdomen, lower extremities and head) in a motor vehicle accident. Disagreements were discussed in group meetings in order to reach a consensus. The preliminary versions of the P-TIS were field tested by investigators individually scoring data from actual patients and then reviewing the scores as a group. On the basis of these field trials, the score was modified and then retested. A number of drafts were necessary until the group was satisfied that the score met the development specifications.
Approval was obtained from the Hadassah Medical Organization Institutional Review Board. Informed consent was waived because of the study's observational nature.
The final P-TIS version was evaluated in the clinical arena following approval of the IRB. Data were collected on two groups of patients: group 1, a convenience sample of elective surgery patients entering the post-anaesthesia care unit (PACU) or directly admitted to an ICU [general ICU (GICU), neurosurgery and cardiothoracic ICUs]; group 2, a convenience sample of emergency surgery patients admitted to the PACU or an ICU after surgery. Patients at least 14 years of age were included, as they were treated by the adult surgical services.
The data were collected from the patients, their responsible physicians and their charts before and following surgery in group 1 and after surgery in group 2. Data included demographic information, such as age and sex, and clinical details, (surgical diagnosis, operative procedure, location of initial post-operative care, in-patient mortality and lengths of ICU and hospital stays). Also determined were ASA Physical Status,3 Revised Cardiac Risk Index,8 Charlson Co-Morbidity Index,7 Physiologic and Operative Severity Score for the Enumeration of Mortality and Morbidity (POSSUM)10 and a modified ASA classification that also considers the effects of the surgical condition on systemic well being.11 The surgical complexity score modified from the POSSUM score quantified the complexity of the planned operation.10 Minor (one point): for example dilation and curettage, tooth extraction, cataract extraction, incision and drainage of rectal abscess. Moderate (two points): for example, appendectomy, cholecystectomy, mastectomy, transuretheral resection of prostate, open reduction and internal fixation of hip or long bone fracture. Major (four points): for example laparotomy, bowel resection, cholecystectomy with choledochotomy, peripheral vascular procedure, major amputation, posterior spine fusion and abdominal hysterectomy. Major plus (eight points): for example aortic procedure, abdominal–perineal resection, pancreatic or liver resection, oesophagogastrectomy, craniotomy, cardiac surgery, anterior–posterior spine fusion/instrumentation.
The hospital has no formal surgical intermediate care unit, so intermediate care patients remain in the PACU for extended (>12 h) stays. After 12–24 h in the PACUs, these patients are transferred to a floor or ICU bed. Anaesthesiologists and surgeons can pre-operatively and intra-operatively request that patients remain in the PACU for extended intermediate care stays. Patients are admitted for extended intermediate care stays, not because GICU beds are unavailable, but because the patient's surgeons and anaesthesiologists think they need a lower intensity of care than that provided in the GICU, but higher than that provided on the floors. When GICU beds are unavailable, GICU patients are physically admitted to the PACUs but are considered ICU patients and cared for by the GICU team. There are written admission criteria for extended intermediate care PACU stays and GICU admissions based on the American College of Critical Care Medicine Practice Parameters which are the framework, but not the absolute guidelines, for physician decisions.12,13
Inter-rater reliability of the P-TIS was examined using kappa statistics and regression analysis. Reasons for the lack of concordance were sought and evaluated for each version. Parametric variables are presented as mean ± SD and non-parametric variables as medians. An a priori decision was made to analyse elective and emergency surgery patients separately.14 Emergency patients were defined as patients not on the elective surgical schedule at 0800 h on the day of surgery. Differences between the two groups were examined using two-tailed t-tests for parametric variables and the Wilcoxon rank–sum test for non-parametric data. P less than 0.05 was considered statistically significant. Multi-variate logistic regression examined the association of both acute and chronic and total P-TIS with the type of post-operative care. Two sets of logistic regressions were performed: ICU versus intermediate/floor care and ICU/intermediate versus floor care. The results were used to identify the better model which was then used in a third set of analyses adding the surgical complexity score. The logistic regressions are reported as odds ratios and 95% confidence intervals. Additionally, the area under the receiver operating characteristic (ROC) curve, specificity, sensitivity and Hosmer–Lemeshow statistic are reported. The relationships between the various scores and classification systems were examined by calculating Pearson (relationships between two continuous variables) and Spearman [relationships between continuous and categorical/ordinal variables [e.g. between the P-TIS and ASA classifications)] correlation coefficients. To reduce the possibility of a type 1 error due to the many correlations, a Bonnferoni correction set significance as P less than 0.001.15 Statistical analyses were performed using Systat 12 (Systat Software, San Jose, California, USA).
Pre-operative Therapeutic Intensity Score development
Developing P-TIS involved assessing and testing six versions. During initial field trials, the developers observed that many patients had interventions within 2 days of surgery that appeared to influence the extent and intensity of intra-operative and post-operative care. The score was thus separated into acute (<48 h before surgery) and chronic (>48 h before surgery) components, so that the final version has acute and chronic sub-scores, plus a total score (Table 1).
The final version of the P-TIS has 40 interventions in the chronic component and 82 in the acute one, with a maximum possible score of 302 points (85 for the chronic portion and 217 for the acute one). As some patients undergoing surgery do not have any pre-operative therapeutic interventions (e.g. a healthy individual undergoing cosmetic surgery), they received a P-TIS of 0. During the evaluation of the final version of the score, 35 randomly selected patients were scored by the same three individuals yielding a kappa statistic of 0.81. It was possible to score a patient in 6–7 min after acquiring some experience.
Group 1 consisted of 716 elective surgery patients (cardiothoracic surgery, 10%; general surgery, 30%; neurosurgery, 8%; OB/GYN, 13%; orthopaedic surgery, 13%; urology, 10%, vascular surgery, 8%, others, 8%). Group 2 included 289 emergency surgery patients (cardiothoracic surgery, 4%; general surgery/multiple trauma, 43%; neurosurgery, 9%; OB/GYN, 7%; orthopaedic surgery, 25%; other, 12%). Table 2 shows their demographics, clinical characteristics and outcomes. The acute versus chronic P-TIS distribution patterns were different for elective and emergency surgery patients (Fig. 1a and 1b), with the latter having higher acute P-TIS scores.
The acute P-TIS was especially good for distinguishing between ICU and intermediate care/floor admissions in the emergency group, but less able to distinguish between post-operative ICU/intermediate and floor care patients. Adding the surgical complexity score improved the analysis's sensitivity and specificity in the elective surgery group.
Both the acute and chronic P-TIS increased along with ASA class (Table 3). In general, the chronic P-TIS, but not acute P-TIS, increased in tandem with the Revised Cardiac Risk Indexes or Charlson Co-Morbidity Scores (Fig. 2). There were differences in acute and chronic P-TIS scores between trauma and other emergency surgery patients. In trauma patients (n = 52) classified as ASA 1E (age 26 ± 11 years), the acute and chronic P-TIS scores were 27.0 ± 13.1 and 0.2 ± 0.1, respectively [31% motor vehicle accidents, 19% gunshot wounds, 25% bomb blasts and 25% others (falls, industrial accidents, etc.)]. Fifty-eight percent went to an ICU after surgery. This contrasted with the other ASA 1E patients (n = 44, age 33 ± 18 years) who had acute and chronic P-TIS scores of 9.9 ± 6.0 (P = 0.0008 versus trauma) and 0.1 ± 0.1, respectively. Only 11% (P = 0.0011 versus trauma) went to the ICU.
There were good correlations (r > 0.7) between the chronic P-TIS and the ASA, Charlson and Revised Cardiac Risk Scores in both groups. The total P-TIS did not correlate as well (r < 0.6) because of the influence of the acute P-TIS. The modified ASA classification was the only score that correlated well (r > 0.6) with the acute P-TIS and total P-TIS. The physiologic component of the POSSUM correlated well (r > 0.68) with the total P-TIS.
A Pre-operative Therapeutic Intervention Score that uses diagnostic and therapeutic interventions as its variables was developed and evaluated. A unique aspect of the P-TIS is its acute and chronic components. The chronic component scores compared favourably with other pre-operative evaluation methods, such as the ASA classification,3 Revised Cardiac Risk Index8 and Charlson Co-Morbidity Scores.7 The acute P-TIS, which reflected care provided for serious illness or injury within 48 h prior to surgery, was not accounted for by the other scoring systems. Furthermore, the higher the acute P-TIS, the greater the odds that patients received post-operative ICU care in both the elective and emergency surgery groups (Table 4). These results indicate that peri-operative, intervention-based scoring is a paradigm that deserves further refinement and investigation.
Post-operative intensive and intermediate care requires much resource utilization. Therefore, the hypothesis that the P-TIS, which quantifies pre-operative resource utilization, would be closely related to the receipt of post-operative intensive care was proven true by logistic regression analysis. The analysis demonstrated that the higher the acute P-TIS, the greater the odds of receiving intensive rather than floor/intermediate care immediately after surgery. The acute and total P-TIS were better in differentiating between intensive care and floor/intermediate care patients than in differentiating between intensive/intermediate and floor patients (as demonstrated by the Hosmer–Lemeshow test). Among the emergency surgery patients, the sensitivities and specificities were of similar magnitude, whereas among the elective surgery patients, the sensitivities were lower than specificities. It was surmised that this was due to some elective surgery patients not receiving much care prior to surgery, but then receiving post-operative intensive care because of the complexity of the surgery. Examples include elective abdominal aortic aneurysm repair and combined anterior–posterior spine surgery in which patients are often admitted on the day of surgery but are admitted to the ICU for post-operative care.16 Therefore, in the elective surgery group, a high acute P-TIS or total P-TIS was highly associated with post-operative ICU care, although not all patients who receive post-operative ICU care had high acute P-TIS or total P-TIS. The importance of surgical complexity in predicting the type of post-operative care among elective surgical patients was shown by increased sensitivities and specificities when a measure of surgical complexity was added to the analysis. Therefore, a high acute P-TIS and/or a complex operation results in ICU admission after surgery. Adding the surgical complexity score did not change the sensitivities or specificities in the emergency surgery group, as in these patients, the degree of physiologic derangement is the major determinant of the location of post-operative care.16 The importance of surgical complexity in determining the receipt of post-operative ICU care in elective, but not emergency, surgery is consistent with previous studies.16,17
The subsequent portion of this study explored how the P-TIS compared with commonly used pre-operative evaluation methods. The chronic P-TIS correlated relatively well (r > 0.7) with the ASA score, Charlson Co-Morbidity Index and the Revised Cardiac Risk Index among both elective and emergency surgery patients. However, the correlations were no better than r is equal to 0.8 because all these scores use different pre-existing conditions as variables. For example, the Charlson Co-Morbidity Index and the Revised Cardiac Risk Index are diagnosis-based, whereas the P-TIS is intervention-based.
Unlike the other scores, the P-TIS score evaluated the acute pre-operative situation. Not surprising, only the modified ASA classification correlated well with the acute P-TIS, as it takes into account the patient's total condition and not just pre-existing situations like the unmodified ASA score. In the elective surgery group, the acute P-TIS increased in concert with increases in the original ASA classification (classes 3 and 4), indicating that these sicker patients received more interventions before surgery than ASA 1 and 2 patients (Table 3). Similarly, in elective surgery patients, the acute T-PIS increased as the RCRI increased (Fig. 2a). Trauma victims were also examples of the acute P-TIS providing much additional information. Although, many of these patients were healthy prior to their injury and, thus, were ASA Class IE as demonstrated by their low chronic P-TIS, many required post-operative ICU care. Therefore, the acute P-TIS, unlike the ASA classification, reflected the severity of their conditions. This demonstrated that the concept of using acute, in addition to chronic, interventions for pre-operative evaluation, appears to fill a gap in the quantitative methods used for pre-operative assessment.
The strength of this study is its introduction of a new concept, therapeutic intensity scoring, for pre-operative evaluation. The chronic P-TIS correlated well with existing pre-operative assessment tools. However, whether P-TIS predicts peri-operative outcomes (mortality and morbidity) was not directly explored because it would require much larger samples, as peri-operative mortality is very low. Instead, the P-TIS was compared with other scores, such as the Charlson Co-Morbidity Index, which were previously validated against outcomes, especially morbidity and mortality.18–20 Additionally, the P-TIS, especially total P-TIS, correlated well with estimates of morbidity and mortality made by the POSSUM audit score,10 which was found in many, but not all, comparison studies to accurately estimate peri-operative morbidity and mortality.21–24
The currently constituted P-TIS has limitations. It is cumbersome in the clinical arena because of its many variables, although after becoming familiar with them, investigators scored patients in 6–7 min. Nevertheless, a version with fewer variables would be more practical, simplifying and expediting scoring. However, the main intended use of P-TIS was as a health systems research tool to measure pre-operative resource utilization. For such a use, scores with many variables are often better measures of resource utilization, especially in heterogeneous populations. The original TISS, still used in ICUs today, has 128 variables.6 Numerous studies have simplified TISS by reducing the number of variables, something that could be accomplished with P-TIS through a multi-centre evaluation.25,26 Another limitation of P-TIS, particularly its chronic component, is that it assumes that the patients received appropriate prior medical care and, thus, are receiving therapy for their significant underlying medical problems.
In summary, this study demonstrates the feasibility of using a score measuring the intensity of pre-operative therapeutic and diagnostic procedures for evaluating whether patients will receive post-operative ICU care. The P-TIS needs further multi-centre evaluation to determine its utility and validity as a tool for quantifying resource utilization, performing clinical pre-operative assessment and predicting the intensity of post-operative care. In addition, the specific situations in which this score would be useful need to be ascertained, for example in a trauma or tertiary care centre with many ICU-bound post-operative patients. It should also be determined whether anticipated surgical complexity should be routinely included when using P-TIS to predict whether post-operative ICU care will be needed. Multi-centre evaluation should also provide data to reduce the number of variables, thus, simplifying the score and making it easier to use.
This work was supported by a grant from The Israel National Institute for Health Policy and Health Service Research.
1 Friedman B, Jian J, Elixhauser A, Segal. A hospital inpatient costs for adults with multiple chronic conditions. Med Care Res Rev
2 Weissman C. The enhanced postoperative care system. J Clin Anesth 2005; 17:314–322.
3 New classification of physical status. American Society of Anesthesiologists. Anesthesiology
4 Menke H, Klein A, John KD, Junginger T. Predictive value of ASA classification for the assessment of the perioperative risk. Int Surg 1993; 78:170–266.
5 Wolters U, Wolf T, Sttzer H, Schroder T. ASA Classification and perioperative variables as predictors of postoperative outcome. Br J Anaesthesiol 1996; 77:217–222.
6 Cullen DJ, Civetta JM, Briggs BA, Ferrara LC. Therapeutic intervention scoring system: a method for quantitative comparison of patient care. Crit Care Med 1974; 2:57–60.
7 Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis 1987; 40:373–383.
8 Lee TH, Marcantonio ER, Mangione CM, et al
. Derivation and prospective validation of a simple index for prediction of cardiac risk of major noncardiac surgery. Circulation 1999; 100:1043–1049.
9 Klein N, Weissman C. Evaluating intraoperative therapeutic and diagnostic interventions. Anesth Analg 2002; 95:1373–1380.
10 Copeland GP, Jones D, Walters M. POSSUM: a scoring system for surgical audit. Br J Surg 1991; 78:355–360.
11 Halaszunski TM, Juda R, Silverman DG. Optimizing postoperative outcome with efficient preoperative assessment and management. Crit Care Med 2004; 32(4 Suppl):S76–S86.
12 Guidelines on admission and discharge for adult intermediate care
units. American College of Critical Care
Medicine of the Society of Critical Care
Medicine. Crit Care Med
13 Guidelines for ICU admission, discharge and triage Task Force of the American College of Critical Care
Medicine. Crit Care Med
14 Weissman C, Klein N. The importance of differentiating between elective and emergency postoperative critical care
patients. J Crit Care 2008; 23:308–316.
15 Warner RM. Applied statistics: from bivariate through multivariate techniques. Thousand Oaks, CA: Sage; 2008. p. 278.
16 Weissman C, Klein N. Who receives postoperative intensive and intermediate care
? J Clin Anesth 2008; 20:263–270.
17 Kheterpal S, O'Reilly M, Englesbe MJ, Rosenberg AL. Preoperative and intraoperative predictors of cardiac adverse events after general, vascular, and urological surgery. Anesthesiology 2009; 110:58–66.
18 Goffi L, Saba V, Ghiselli R, et al
. Preoperative APACHE II and ASA scores in patients having major general surgical operations: prognostic valve and potential clinical applications. Eur J Surg 1999; 165:730–735.
19 Singh B, Bhaya M, Stern J, et al
. Validation of the Charlson comorbidity index in patients with head and neck cancer: a multiinstitutional study. Laryngoscope 1999; 107:1469–1475.
20 Krenn H, Dorn C, Fitzgerald RD, et al
. Prognostic value of APACHE II and III for preoperative evaluation of emergency surgery patients. Eur J Anaesthesiol 2000; 17:529–530.
21 Bann SD, Sarin S. Comparative audit: the trouble with POSSUM. J R Soc Med 2001; 94:632–634.
22 Bollechwiler E, Lubke T, Monig SP, Holscher AH. Evaluation of POSSUM scoring system in patients with gastric cancer undergoing D2-gastrectomy. BMC Surg 2005; 5:8–14.
23 Midwinter MJ, Tytherleigh M, Ashley S. Estimation of mortality and morbidity risk in vascular surgery using POSSUM and the Portsmouth predictor equation. Br J Surg 1999; 86:471–474.
24 de Cassia Braga Ribeiro K, Kowalski LP. APACHE II, POSSUM and ASA scores and the risk of perioperative complications in patients with oral or oropharyngeal cancer. Arch Otolaryngol Head Neck Surg 2003; 129:739–745.
25 Castillo-Lorente E, Rivera-Fernandez R, Rodriguez-Elvira M. Tiss 76 and Tiss 28: correlation of two therapeutic activity indices on a Spanish multicenter ICU database. Intensive Care Med 2000; 26:57–61.
26 Moreno R, Morais P. Validation of the simplified therapeutic intervention scoring system on an independent database. Intensive Care Med 1997; 23:640–644.