The American Society of Anesthesiologists (ASA) Physical Status classification system celebrates its 80th anniversary in 2021. Its simplicity represents its greatest strength as well as a limitation in a world of comprehensive multisystem tools. It was developed for statistical purposes and not as a surgical risk predictor. However, since it correlates well with multiple outcomes, it is widely used—appropriately or not—for risk prediction and many other purposes. It is timely to review the history and development of the system. The authors describe the controversies surrounding the ASA Physical Status classification, including the problems of interrater reliability and its limitations as a risk predictor. Last, the authors reflect on the current status and potential future of the ASA Physical Status system.
ANESTHESIOLOGY emphasizes quality patient care, safety, and scientific discovery.1,2 Eighty years ago, leaders of the American Society of Anesthesiologists (ASA; Schaumburg, Illinois) saw the need for a simple tool to collect and tabulate statistical data related to patients’ status before anesthesia.3 This initiative was the framework for the development of the ASA Physical Status classification system.4
It was recognized from the time of its introduction that the severity of coexisting diseases and the functional status of the patient are vital parameters to stratify patients before surgery. This system was not designed, developed, or intended to predict perioperative risk.5,6 Nevertheless, many subsequent publications have shown that the ASA Physical Status classifications are strongly correlated with outcome, either independently or in conjunction with other information.7–9 When combined with other factors (e.g., age, type and complexity of surgery, frailty, nutritional status, end-organ dysfunction, among others), the ASA Physical Status classification can contribute to the assessment of risk and outcomes, at least with aggregate data.7–9 However, its utility as a predictor for the individual patient is dubious.
Despite its familiarity, considerable variability exists between anesthesiologists when assigning an ASA Physical Status score to specific patients.10,11 Misclassification may have several negative consequences, ranging from inappropriate staff assignments to misaligned billing codes, and even skewed measures of outcome.12–18 One attempt to address this variation was made by the ASA with the addition of case examples, approved first in 2014 and updated in 2020. However, it appears that these examples have only marginally improved reproducibility.19,20
Given these issues, we will review the history and evolution of the ASA Physical Status classification system and examine its strengths and limitations, specifically the variability of ASA Physical Status assignments. We will discuss its application to special populations (e.g., pediatrics, obstetrics, trauma surgery, and the potential role of age and frailty), its use as a risk predictor particularly as compared with more comprehensive risk stratification systems, and its use for other administrative purposes (including billing). Finally, we will consider approaches that could impact its future value.
In 1940, the ASA established a committee consisting of Meyer Saklad, Ivan Taylor, and Emery Rovenstine (fig. 1) “to study, examine, experiment, and devise a system for the collection and tabulation of statistical data in anesthesia.” In the May 1941 issue of Anesthesiology,3 Saklad proposed six designations. Classes 1 to 4 described a patient’s preoperative “physical state,” whereas classes 5 and 6 were reserved for patients scheduled for emergency surgery. Saklad also provided a series of examples to illustrate the various grades (table 1).
The 1941 publication articulates the authors’ vision. Saklad elaborated on the concept of “operative risk,” and recognized that its assessment is multifactorial, situational, and complex. Factors such as physical condition, surgical procedure, surgical skill, postoperative care, and the experience of the anesthetist required consideration. As a result, operative risk can vary tremendously, even for comparable patients. Given that the goal of the system was to allow the collection of data for statistical analysis, the assignment of a class solely based on the patient’s “physical state” required fewer data and was easier to implement. A quote from the manuscript summarizes this thought process:
The attempt to determine a patient’s “Operative Risk” may be of value in prognosis, but such grading of patients is useless from a statistical point of view. It is useless from several standpoints: the excessive number of variables to be considered, the tremendous degree of variation in different clinics and different physicians and the complete lack of agreement as to definition of terms.
Saklad’s proposal was apparently adopted by the ASA, although the timeline remains elusive. We assume that the classification system was implemented around April 1945 with publication of a document titled “Codes for the Collection and Tabulation of Data Relating to Anesthesia Inhalation Therapy and Therapeutic & Diagnostic Block” (fig. 2, A and B). A 1978 editorial21 by Arthur Keats corroborates this concept and reports that the House of Delegates did, sometime before 1945, accept the Saklad plan and also added a seventh category for the “moribund patient not expected to survive with or without surgery.” Figure 2C shows a document from the papers of Paul M. Wood, M.D. (Judith A. Robins, Wood Library-Museum of Anesthesiology, Schaumburg, Illinois, January 2021, written communication) in which definitions of physical states 1 through 7 are listed (including a misprint as physical state 6 is listed twice). Unfortunately, the date of this document is unknown. Figure 3 shows a sample paper anesthesia record approved by the ASA in 1950 with the physical status codes in the upper right corner.
In 1961, Dripps et al. published “The Role of Anesthesia in Surgical Mortality.”22 Their aim was to define the contribution of anesthesia to perioperative deaths. Among the data collected was the patient’s “physical status”:
A normal healthy patient
A patient with a mild systemic disease
A patient with a severe systemic disease that limits activity but is not incapacitating
A patient with an incapacitating systemic disease that is a constant threat to life
A moribund patient not expected to survive 24 h with or without operation
The physical status classification differed slightly from the 1941 system. Specifically, Dripps et al. added that “Physical Statuses 1, 2, 3, and 4 correspond to the classification proposed by the American Society of Anesthesiologists. In the event of emergency operation, the only change in classification was to precede the number with the letter E. PS 5 corresponds to the original ASA Classification 7.”
Since it appears that members of the society raised questions about the definition and application of Saklad’s system, Richard Ament proposed the adoption of the five-category Dripps et al. version via Resolution No. 6 to the 1962 House of Delegates (fig. 4).23,24 The resolution was approved, and the new classification system was published in the ASA newsletter and the January–February 1963 issue of Anesthesiology.25 The ASA used the same wording as in the Dripps et al. publication and only added, “In the event of an emergency operation, precede the number with an E.”
The Dripps et al. version remained unchanged for the next 18 yr. But in 1981, a small modification was made; ASA Phyisical Status 5 was changed to “a moribund patient who is not expected to survive without the operation.”26 However, this change does not appear to have been widely publicized, and multiple sources continued to cite the older definition. This ambiguity was nicely summarized as late as 2011 by Thackray and Gibbs.26
In 1983, the ASA added a sixth category (ASA Physical Status 6) to denote a brain-dead patient scheduled for organ retrieval.27 * Again, this revision was criticized for a lack of clear dissemination. In a 2001 letter (18 yr later), Stone described asking nine consultants and seven junior anesthetists to grade a brain-dead patient undergoing organ harvest. None of the 16 providers assigned the correct class.27
On October 15, 2014,28 the next revision of the ASA Physical Status included the provision of a number of illustrative patient examples for each category, similar to Saklad’s original manuscript. The categories (I through VI and E) remained unaltered except for a change to Roman numerals. This was a significant modification to the position taken by the ASA since 1963 when examples or other elaboration on the individual categories were expressly avoided. The change was driven by the growing use of the ASA Physical Status system by nonanesthesiologists and administrative bodies, in many cases for purposes that were far different from the primary intent of the system, but which nevertheless had a major impact on the practice of anesthesiology and general medical care.29
In 2019, the ASA House of Delegates added two items to the text surrounding the 2014 version. Specifically, they suggested that “anesthesiology departments may choose to develop institutional examples to supplement the ASA approved examples” and that “the final assignment of Physical Status classification is made on the day of anesthesia care by the anesthesiologist after evaluating the patient.”
The most recent revision from December 13, 2020, included the addition of pediatric and obstetric patient examples for each category (table 2).30 Despite the changes aimed at either simplification of the system or improvement of interrater reliability, we find it striking that the current ASA Physical Status system bears such close resemblance to the original version by Saklad and colleagues from 80 yr ago.
Strength of ASA Physical Status
A major strength of the ASA Physical Status system is its simplicity. Saklad and team endeavored to create a tool that would need few variables and no complex computations. This has made it possible for the ASA Physical Status system to be used internationally and across multiple medical specialties. It is freely available and does not require data abstractors or computerized modeling.30 It is time-tested and obviously durable, as evidenced by iis survival for 8 decades.3,28 None of the later modifications required any fundamental alteration of its original concepts.
The intended use of the ASA Physical Status system is to allow preoperative health status of surgical patients to be summarized and compared. For example, someone might wish to compare postoperative length of stay at hospital A versus hospital B. But someone argues, “We take care of sicker patients at hospital A.” Examining the distribution of the assigned ASA Physical Status scores of the patients at the two hospitals helps address this conundrum. As a result, the system has been widely adopted as an administrative tool to compare the performance of hospitals or as a covariate for measuring the outcomes of specific surgical populations.
Also consistent with the originators’ goals, the ASA Physical Status system has become a ubiquitous component of clinical investigations.31 Publications related to anesthesia and surgery almost always contain the ASA Physical Status class distribution of the subjects under investigation. These help define the demographic characteristics of subjects and document similarities (or differences) between groups. It may also be used as a covariable for a wide range of study outcomes. Indeed, the ASA Physical Status classification system has even appeared in the veterinary literature (at least as applied to dogs, cats, pigs, and rabbits).32
Limitations of ASA Physical Status
The ASA Physical Status system has also been used for purposes that go far beyond its original intent. For example, it has been used to make triage decisions regarding assignment of inpatient versus ambulatory care for individual patients. Some hospitals use it to decide whether or not to transfer patients to other facilities.33 It may be used to aid in assigning appropriate anesthesia providers to a given case,29,34,35 and to determine staffing ratios when anesthesiologists cover multiple anesthetizing sites.36 It has been used by nonanesthesiologists to determine whether anesthesia professionals are required for out-of-operating-room procedures with sedation.37 Finally, ASA Physical Status is used as a billing modifier by many insurers. However, these “unintended” uses may be problematic.
The first examination of interrater reliability appeared in 1978. Owens et al. created a list of 10 hypothetical patients with two of the authors defining a consensus ASA Physical Status designation for each.38 They then surveyed 235 ASA members. Just over 60% of respondents agreed with the consensus classifications for six or more of the cases. Four cases were the major sources of disagreement. One healthy patient was “misclassified” (in the opinion of the authors) as ASA Physical Status II based on age alone (75 yr). Another healthy 24-yr-old was “misclassified” as ASA Physical Status II based on a hemoglobin of 9.5 g/dl (and some respondents even assigned an ASA Physical Status III score). Another patient with a previous, uncomplicated myocardial infarction (7 months earlier) received scores ranging from I to IV. Similar inconsistency was seen for an otherwise healthy 37-yr-old woman, based on her body mass index of 47.5 kg/m2. The authors note that many anesthesiologists have difficulties in separating “preoperative physical status” from perceived “anesthetic risk” and concluded that the ASA Physical Status “suffer(s) from a lack of scientific definition.”
Multiple other investigators, using similar methods, have found similar results.39–41 Clinician agreement ranged from 40% to over 90%, but complete agreement was not achieved in any study; in some cases, ASA allocations spanned at least three and occasionally five grades. In perhaps the largest such study, Cuvillon et al. extracted the ASA Physical Status scores assigned by attending anesthesiologists in 1,554 cases.42 The records were presented—along with the original ASA Physical Status assignment—to other experienced anesthesiologists. Each either agreed with the original score or assigned a different one. The lowest and highest discordance rates were found for ASA Physical Status I (24.5%) and IV (83.3%), with a Cohen’s kappa coefficient of 0.53, indicating weak to moderate agreement.
Better concordance is seen when direct patient contact by all raters is involved. Sankar et al. compared scores assigned by anesthesiologists in the preoperative assessment clinic with scores given to the same patients by different providers on the day of surgery.43 Sixty-seven percent of patients were assigned the same scores in the clinic and operating room. Nearly 99% of assignment pairs either matched or were within one ASA Physical Status class of each other, producing a kappa score of 0.61. Interrater disagreements were associated with age, selected comorbidities (particularly hypertension and malignancy), and the commonly mistaken belief that the complexity or duration of the surgical procedure influences ASA Physical Status.
Value of Case Examples to Interrater Reliability
Three years after the 2014 ASA Physical Status modification, Hurwitz et al. examined the value of the new patient examples.44 A questionnaire with 10 hypothetical ASA Physical Status I, II, and III patients were sent to 779 anesthesia-trained providers (anesthesiologists, fellow/resident anesthesiologists, certified registered nurse anesthetists, and anesthesia assistants) and to 110 non–anesthesia-trained physicians and nurses who assign ASA Physical Status for sedation purposes. In the first part of the survey, the ASA Physical Status category was decided based on the words of the category alone. In the second part, the same cases were presented in a different order, and with the new ASA-approved examples added. The responses were compared to consensus scores determined by the investigators. Significantly improved agreement (greater than 70% of responders agreed) was found in 9 out of 10 scenarios when examples were provided. However, both overclassification and underclassification occurred despite the examples. One hypothetical patient had several coexisting diseases that may well have med the criteria of the ASA Physical Status III category, and the majority of the responders assigned the patient to that category in the first part of the survey. However, none of those comorbidities were listed in the examples, and therefore both the investigators and the majority of the responders in the second part of the survey scored the same patient as ASA Physical Status II. Clearly, patient examples may have positive and negative consequences.
However, another survey found the opposite result.10 Eight hypothetical scenarios ranging from ASA Physical Status I to V were presented to providers with differing degrees of experience. Both the ASA Physical Status classes and the examples were provided to the respondents. Interrater reliability remained low despite the examples, with more experienced anesthesiologists “incorrectly” scoring the ASA Physical Status classes more often than residents and recent graduates.
In the largest effort to assess this issue, Fielding-Singh et al. examined over 2 million records of patients from the American College of Surgeons (Chicago, Illinois) National Surgical Quality Improvement Program database from 2011 to 2017, with four specific comorbidities.19 In patients with septic shock and acute renal failure on dialysis (ASA Physical Status IV per definition) the underclassification rates were 24% and 38.7%, respectively, in the 2011 to 2014 cohort. This remained virtually unchanged between 2015 and 2017, after the examples were introduced. In patients with a body mass index from 30 to 39.9 kg/m2 (ASA Physical Status II) and body mass index over 40 kg/m2 (ASA Physical Status III), the rates of underclassification were 4.8% and 27.6%, respectively, between 2011 and 2014, with a significant but very small improvement noted after 2014.
Special Populations and Additional Uncertainties
Before the 2020 amendment,30 the ASA Physical Status document included the disclaimer, “The examples in the table […] address adult patients and are not necessarily applicable to pediatric or obstetric patients.”28 The only listed pediatric-specific example was a premature infant with postconception age less than 60 weeks (noted as ASA Physical Status III).28
Similar to the interrater variability of ASA Physical Status in adults, studies focusing on children show only weak45,46 to moderate agreement.47,48 Tollinche et al. assessed children with cancerand found kappa scores of only 0.042, which is equivalent to random chance.46 However,
Jacqueline et al. reported better results from a postal questionnaire sent to members of the Society for Pediatric Anesthesia (Richmond, Virginia),48 reporting an overall kappa coefficient of 0.47. This improved to 0.52 when cardiac cases were excluded. Similar to studies of adults, the best agreement was noted at the extremes of the ASA Physical Status categories. In 2019, Leahy et al. proposed a Pediatric ASA Physical Status scoring system that was richly populated with clinical examples.49 Ferrari et al.50 tested the utility of this new proposal by asking teams of three pediatric anesthesiologists to reclassify 120 pediatric surgical cases based on this proposed system. Reassignment most often resulted in an upgrade to the ASA Physical Status (e.g., 42% of the original ASA Physical Status I group were reclassified ASA Physical Status II), while a quarter of the ASA Physical Status IV children were downgraded to III. As with previous studies, the lowest concordance was noted in ASA Physical Status II and III patients and was best at the extremes.50
Based on this and related work, the ASA included pediatric-specific examples in the 2020 amendment of the ASA Physical Status.30 Whether the new pediatric examples will produce the desired effect remains to be determined.
The 2020 update of the ASA Physical Status guideline recognizes that normal pregnancy is not a disease, yet it is categorized as ASA Physical Status II due to the significantly different physiologic status of a parturient.30 However, there is little guidance as to how to adjust for the many complications of pregnancy. This challenge is compounded by the paucity of published papers that address the appropriate stratification of more complicated pregnant patients.51,52
Thus, it is not surprising that interrater inconsistency occurs. Nearly 400 members of the Society for Obstetric Anesthesia and Perinatology (Lexington, Kentucky) were asked to assign a ASA Physical Status to “a healthy 24-year-old woman, G1P0 [gravida 1, para 0, a woman who is pregnant for the first time and has not yet delivered], in active labor, requesting labor epidural analgesia.”51 Respondents were nearly evenly split (44% ASA Physical Status I vs. 55% ASA Physical Status II). Because of this, some obstetric experts sought to introduce a pregnancy specific modifier by adding a “G” to the baseline ASA Physical Status score.51 Opponents argued that this does not reflect both specific derangements (e.g., placenta previa vs. gestational diabetes) and disease severity (e.g., mild vs. severe pre-eclampsia) of more complicated pregnancies. Unfortunately, there is no consensus among obstetric anesthesiologists regarding the optimal modification of the traditional ASA Physical Status system as it applies to pregnant patients before surgery.
The new 2020 case examples attempt to capture some of the nuances encountered with more complicated pregnancies. However, as with the new pediatric examples, formal study will be needed to evaluate the impact of these changes.
Emergency Surgery, Trauma, and Critical Illness.
The original 1941 physical classification system stated, “An emergency operation is arbitrarily defined as a surgical procedure which, in the surgeon’s opinion, should be performed without delay.”3 Since 1963, the addition of “E” to any ASA Physical Status class denotes emergency surgery. ASA now defines an emergency as existing when delay in treatment of the patient would lead to a significant increase in the threat to life or body part.53 While deceptively simple, these definitions are, in fact, quite imprecise. The “E” designation does not specify the duration or the severity of the emergency diagnosis; there are obvious differences between a comatose patient with an acute epidural hematoma and a patient with early acute appendicitis. What time increment constitutes a “delay in treatment”—minutes, hours, or days? And how are clinicians to quantify a “significant increase in the threat”?
The surgical literature provides one approach. Kluger et al. published a proposal for the ideal time to initiate surgery.54 The authors define bleeding emergencies (predominantly trauma-related) as “immediate surgery” and nontraumatic emergencies (such as perforated viscus, incarcerated hernia) as “surgery within an hour.” Both groups would certainly warrant “E” designation within the ASA Physical Status system. In addition, some institutions utilize an “A, B, C” triage system, where a class A emergency would be surgery within an hour, a class B emergency might safely wait 3 or 4 h, and a class C might wait as long as 8 h. However, whether all these categories (other than the most urgent) warrant the “E” designation is debatable and undefined, and how such a classification might be incorporated into the ASA Physical Status is unknown.
There is also ambiguity concerning a patient’s preinjury state, their current clinical state, and the assignment of an ASA Physical Status. The current ASA Physical Status system provides no specific guidance as to how the severity of an injury translates to the standard definitions of the preoperative state (e.g., what kind of injury correlates with “severe systemic disease”). But examination of Saklad’s notes and examples provides guidance.3 Saklad described how a previously healthy 20-yr-old patient who sustained both head trauma and significant abdominal hemorrhage is to be designated physical state class 6 (indicating a trauma-triggered transition from the initial physical class 1 state to the emergency category for a class 3 or 4 patient). While not covered in the current ASA examples, this patient would probably be an ASA Physical Status V-E since survival would be unlikely without surgery. But classification of less catastrophic emergencies is more controversial.
Last, the impact of preoperative intensive care unit interventions such as mechanical ventilation, dialysis, extracorporeal membrane oxygenation or left-ventricular assist devices on ASA Physical Status class designation is undefined.55 Patients receiving such treatments can be of all ages and preintervention conditions, and their physiologic condition can range from near moribund to extremely “stable,” even ambulatory. However, there is little guidance on how to integrate advanced life-support devices into an ASA Physical Status assignment.
Age, Chronic Disease, and Frailty.
The ASA has consistently refrained from considering age as a part of the physical status system, except for the very young. Chronological age was intentionally omitted since it usually does not correlate with functional status. While older patients suffer greater perioperative morbidity and mortality,56–58 a retrospective study in 159 nonagenarians (mean age of 91.8 yr) found that it was not age per se, but ASA Physical Status and emergency surgery that were key drivers of their outcomes.59 In clinical practice, anesthesiologists almost certainly upgrade their ASA Physical Status class assignment based on the extremes of age, possibly independent of “systemic disease” and probably based on the belief that a chronic disorder(s) combined with extreme age warrants such upgrading. While it is likely that this adjustment to the ASA Physical Status score reflects a valid product of clinical intuition and experience, it also reflects the inherent subjectivity of the ASA Physical Status (e.g., what constitutes a “mild” vs. “serious” systemic disease) and is unlikely to be resolved by any simple modification to the ASA Physical Status categories.
In recent years, “frailty” has been increasingly recognized as an important comorbidity and factor in surgical outcomes.60,61 Various scales are available to measure frailty before surgery. They all examine a patient’s activities of daily living (such as the need for assistance with bathing, dressing, and so forth), strength, weight loss, cognitive function, among others.62,63 Frailty is mentioned as a complement to the ASA Physical Status classification system in the ASA’s preamble on the current website but is not included in the table of definitions or examples.30 Thus, there is no formal guidance or current means to integrate frailty into the ASA Physical Status. Nonetheless, it would seem that an objective determination of “frail” could—by itself—constitute at least a “serious systemic disease” (ASA Physical Status III) and in some patients could be a “threat to life” (ASA Physical Status IV). Therefore, we assert that frailty should become a standard part of a comprehensive preoperative evaluation and could well be incorporated into future amendments to the ASA Physical Status system.
ASA Physical Status, Risk, and Outcome.
ASA Physical Status is not intended and should not be used as predictor of operative risk, certainly not in individual patients.5,6,43,64 It has nevertheless evolved into just such a system in both appropriate and inappropriate ways. It is intuitive that a patient’s “physical status” will be associated with that person’s postoperative outcome as preexisting disease may be the most common cause of death.65 No one was surprised when Dripps et al. reported a strong correlation between ASA Physical Status and postoperative mortality attributable to anesthesia.22 However, while Dripps et al. reported a statistical association, the positive predictive value for any individual ASA Physical Status value was very low (i.e., only 3% for ASA Physical Status IV and V combined), meaning that the overwhelming majority of even the sickest patients do not die as a result of their anesthetic care (although all-cause mortality was not reported).
Similar univariate correlations with multiple outcomes have been shown many times. Davenport et al. examined the outcomes of a random group of 5,878 patients undergoing noncardiac surgery at a single center.66 All-cause 30-day mortality, morbidity, costs, and length of stay increased progressively with increasing ASA Physical Status with P values of less than 0.0001. Mortality increased from 0 for ASA Physical Status I to 70% for ASA Physical Status V. They also compared the predictive value of ASA Physical Status against a larger series of National Surgical Quality Improvement Program risk factors—and noted an exceptionally high correlation (C-statistic approximately 0.9). Hackett et al. examined the records of 2,297,629 noncardiac surgical patients in the National Surgical Quality Improvement Program database67 where the ASA Physical Status values had been assigned by anesthesia providers. There was a clear and strong relationship between ASA Physical Status and both complications and all-cause deaths (table 3). Koo et al. published a meta-analysis of 77 studies containing over 165,000 patients.68 The results of individual studies were converted into 2 × 2 contingency tables, based on mortality and whether the patient had an ASA Physical Status I or II versus ≥ III. Their pooled estimate of sensitivity was 0.74, meaning that if a patient died, they had a 74% chance of having ASA Physical Status III or greater. Specificity was 0.66, meaning that if a patient did not die, they had a 66% chance of having ASA Physical Status I or II. But again, the positive predictive value for an ASA Physical Status III or greater was very low; less than 4% of patients with ASA Physical Status III or greater actually died.
Multiple other studies have shown similar unadjusted and statistically adjusted relationships. These include patients with endometrial cancer,58 urinary tract cancers,69 noncardiac thoracic surgery,70 hip arthroplasty,9,71 major spine surgery,72,73 hip fractures,74 and trauma.75–77 Such associations are not, however, universal. Thomas et al. found that ASA Physical Status was not predictive of postoperative morbidity in patients undergoing surgery for chronic subdural hematomas,56 but seven of the top eight most common complications were neurologic events that might be unrelated to preoperative “physical status.” ASA Physical Status did not predict postoperative prosthetic joint infections,78 nor was it associated with mortality in gastric cancer patients.57 Shahrokni et al. found a univariate relation between ASA Physical Status and postoperative intensive care unit admissions, but not with 6-month mortality.79
Several of the previously cited studies support the concept that the predictive value of the preoperative physical status can be improved when additional data are used. Perhaps the earliest application of this approach was by Goldman et al., who constructed a multivariate statistical model including the ASA Physical Status classification, along with other cardiac signs and symptoms (e.g., S3 gallop, previous myocardial infarction, among others).80 ASA Physical Status has also been used as part of the National Surgical Quality Improvement Program Surgical Risk Scale. Davenport et al. examined the records of 183,069 patients having general or peripheral vascular surgery.81 Over 50 preoperative and intraoperative risk factors and laboratory tests were used to construct a predictive model. Twenty factors remained in the final model—with ASA Physical Status having the highest single odds ratio (for ASA Physical Status [IV and V] vs. [I and II]).
It is clear that the ASA Physical Status scale is limited in predicting risk as a stand-alone tool and given the issues of interrater reliability and poor positive predictive values, it should never be used as a predictive tool for individual patients (with the possible exception of ASA Physical Status V and V-E). A subjective 5-point scale can never incorporate the myriad of factors that influence outcome in a single individual. As noted above, more elaborate risk prediction models have been developed—but they also require far more information and are more demanding to apply. A qualitative systematic review in 2013 compared the performance of eight leading surgical risk predictors and concluded that the Portsmouth Physiological and Operative Severity Score for the enumeration of Mortality and Morbidity and the Surgical Risk Scale were the most consistently accurate predictive tools for predicting postoperative mortality and morbidity.82
ASA Physical Status and Billing.
The ASA Physical Status system is used as a billing modifier by some private payors. One additional unit may be added to the charge for an ASA Physical Status III patient, 2 units for an ASA Physical Status IV patient, and 3 units for an ASA Physical Status V patient. No added units are allowed for ASA Physical Status I or II. Other factors related to the ASA Physical Status may also generate additional units, such as the extremes of age or emergency surgery (so-called Qualifying Circumstances codes).83 While these codes are not recognized by Medicare or Medicaid, the ASA’s Commercial Conversion Factor Survey reports that 80% of responding anesthesia group private contracts include reimbursement for ASA Physical Status and Qualifying Circumstances codes. On July 1, 2021, the U.S. Departments of Health and Human Services, Treasury, and Labor (Washington, D.C.), and the Office of Personnel Management (Washington, D.C.) issued their first rule related to the “No Surprises Medical Bill Act” that recognizes anesthesiology’s unique payment methodology including the physical status modifier.†
The aforementioned variation in ASA Physical Status assignment amplifies the potential for upcoding or downcoding. Unintentional upcoding may simply occur as a result of the “subjectivity” associated with ASA Physical Status assignments, variation in interpretation of published patient examples, and perhaps the specific experience or practices of providers.
In 2016, Nie et al. showed a progressive increase in the fraction of gastrointestinal endoscopy patients coded as ASA Physical Status III or greater (from 11.6% in 2005 to 18.9% in 2013).84 After adjusting for recorded comorbidities, the probability of medically comparable patients receiving higher ASA Physical Status scores increased almost fivefold, suggesting (to the authors) the likelihood of upcoding. In contrast, Schonberger et al., working with the National Anesthesia Clinical Outcomes Registry, attempted to determine whether there was a discontinuity in the distribution of ASA Physical Status scores assigned to patients below and above age 65 yr.17 Since patients 65 yr and older are likely to be covered by Medicare, which does not pay for these ASA Physical Status modifiers, this would theoretically remove any incentive to upcode for seniors not covered by Medicare. They were unable to find such a discontinuity, although they acknowledge that they cannot determine whether this was “because of virtue, lack of knowledge, or a combination thereof” or because the incidence of upcoding was too low to detect. Evidence for downcoding, perhaps by nonanesthesia providers seeking to avoid the need for mandated anesthesia services, is even less compelling. In an abstract published in 2015, Romano et al. noted that ASA Physical Status scores assigned by nonanesthesiologists for patients requiring sedation were significantly lower than those assigned (retrospectively) to the same patients by anesthesiologists.85 However, it is impossible know if this represents “intentional” downcoding or is related to other unknowable factors. We have been unable to identify any other published data related to this issue. While it remains possible that upcoding or downcoding occurs, there is absolutely no evidence to support any belief other than that practicing anesthesiologists consistently strive to assign accurate ASA Physical Status classifications.
It should be noted that both upcoding and downcoding involve the use of the ASA Physical Status for the classification of individual patients, not large aggregate groups of patients—something that is problematic because of interrater issues. However, with improvement of interrater reliability, potential coding inaccuracies may become less prevalent and hence less of an issue.
The Future: Potential Modifications and Conclusions
ASA Physical Status works extremely well for its intended purpose, which is to allow comparison of large groups of patients.48,52,86–88 We believe that most ASA members and ASA leadership are comfortable with the current use. Every version of the system has been approved by the ASA’s most representative body, the House of Delegates, including in 2020 without dissent. However, on its 80th anniversary, it is timely to ask if the system needs further refinement. To answer that question, we must reflect on the purpose of any such change. Do we simply want better interrater reliability? Do we wish to refine the definitions of ASA Physical Status II, III, and IV patients (which represent the greatest degree of assignment variability) along with additional focus specific to frailty and emergency patients (particularly trauma)?48,86,87 Do we need more categories? Do we want to improve its use as a risk predictor? Do we want to improve its regulatory applications or its use as a billing modifier? Each goal would require a different approach—and some would mandate major changes to the current structure.
In the opinion of the authors, creating an outcome prediction system for use in an individual preoperative patient would go far beyond a “modification” of ASA Physical Status, and would require a more sophisticated multivariate approach—as has already been done and validated by others. We believe that efforts would be better directed at improving category clarity and interrater consistency—and actively disseminating the most recent version of the ASA Physical Status system to the worldwide anesthesia community. A first step would be to identify those clinical scenarios that seem to result in the most common assignment discrepancies. This was, of course, the origin of the added examples. Further reiterative efforts along these lines, continuing to focus on identified “problems” (e.g., ASA Physical Status III and IV), might yield improvements. However, given the unavoidable fact that different providers will judge the health status of the same patient differently—even if both providers personally perform histories and physical examinations—perfect consistency when binning the enormous spectrum of human disease into a small number of ordinal categories is likely unattainable. Hence, any group striving to improve interrater reliability will need to pre-emptively establish their definition of “sufficiently good reliability.”
The end product of such efforts might involve a few more ASA Physical Status categories or further revision of patient examples. In the 21st century, basic technology could readily assist the ASA Physical Status assignment process. A simple option might be to provide the definitions of the current ASA Physical Status categories alongside key “check boxes” on the anesthesia record, or to create “pop-up” screens with both the definitions and examples. There has been one published effort to “predict” ASA Physical Status using a huge number of variables,89 but this is clearly not feasible for the busy clinician. In a world of anesthesia information management systems, it is possible to imagine automated decision-support algorithms that could be applied using information already contained in the medical record. However, to date, no such effort has been reported. One could envision a semimanual system (perhaps a smartphone-based application) that asks a provider to answer a series of simple yes/no questions to “recommend” an ASA Physical Status class. Regardless of the approach, it needs to retain the simplicity and ease of use of our current system. Moreover, continued improvements to the current (and future) ASA Physical Status systems also require that the ASA actively disseminate and educate both American and international anesthesia providers regarding precise ASA Physical Status definitions and case examples.
We envision continued utilization of this bedside tool by both anesthesia professionals and others for years to come. Ultimately, its future will be determined by the medical needs of clinicians, the research needs of investigators, the oversight needs of regulators—and most importantly, the goals of the ASA leaders and its membership. In order to stay current and meet these diverse demands, the system warrants periodic re-examination, consistent with its long-standing history. For now, the ASA and its members should be proud of a system that has stood the test of time.
The authors thank Judith A. Robins (Wood Library-Museum of Anesthesiology, Schaumburg, Illinois) for her help with the search for rare historical documents related to the American Society of Anesthesiologists Physical Status classification system and its authors.
Support was provided solely from institutional and/or departmental sources.
Dr. Todd was the Editor-in-Chief of Anesthesiology from 1996 to 2006. Dr. Todd was also the awarded the 2016 Excellence in Research Award by the ASA. Dr. Cole is vice president of the Anesthesia Patient Safety Foundation (Rochester, Minnesota), a foundation of the ASA, and past president of the ASA. Dr. Prielipp is a former member of the Board of Directors of the Anesthesia Patient Safety Foundation. He also serves on the speakers’ bureau for Merck Co., Inc. (Kenilworth, New Jersey) and as an opinion leader for 3M (Minneapolis, Minnesota). The other authors declare no competing interests.
It is not clear how the current ASA Physical Status VI category applies to so-called “non-heart beating donors” who were not previously declared brain-dead.
Message from the ASA President, Friday, July 9, 2021.