CORR Insights®: Does a Prescription-based Comorbidity Index Correlate with the American Society of Anesthesiologists Physical Status Score and Mortality After Joint Arthroplasty? A Registry Study

Where Are We Now?

About 85% of patients undergoing total joint arthroplasty have at least one medical comorbidity [3]. These conditions are independently associated with increased postoperative complications, discharge to destinations other than home (such as skilled nursing facilities), and even death after arthroplasty [8].

Over time, we’ve used certain scores like the American Society of Anesthesiologists (ASA) physical status score, the Charlson comorbidity index, and the Elixhauser comorbidity score to analyze the outcomes of patients with comorbidities. These scores are based on clinical assessments and hospital-coded diagnoses [4, 6]. The Rx-Risk score [10] is another measure of comorbidity; it is based on prescriptions dispensed preoperatively. Elixhauser, Charlson, and Rx-Risk comorbidity measures were validated by Inacio et al. [7] as acceptable mortality predictors in THA and TKA patient cohorts. One study [11] found that serum albumin also has a good predictive value for postoperative complications in these patients, with results comparable to ASA. Similarly, the 5-factor modified frailty index is also an independent predictor of postoperative complications and can be used for the same purposes [12].

The authors of the present study [9] analyzed the Rx-Risk score, which uses medication-dispensing records to stipulate the presence of comorbidities that are required to be managed by medications [5, 10] and concluded that ASA and Rx-Risk were weakly to moderately correlated to each other. However, each score was highly associated with the probability of mortality shortly after undergoing a procedure.

The Rx-Risk score is readily available and may be a useful additional measure of a patient’s overall health, with the added benefit of defining the conditions for which a patient is receiving pharmaceutical treatment. Medication information may be useful to further understand risk factors in patients with the potential for poor functional outcome after joint replacement surgery. These may indeed be of great aid in eliminating the confounding factors while evaluating the probability of mortality shortly following major joint arthroplasties. Based on the results of the study, surgeons can do proper risk stratification, which may help when counseling patients and relatives or caregivers as well as when selecting patients for major joint arthroplasties

Where Do We Need To Go?

Surgeons and clinician scientists understand the importance of comorbidity scores when planning surgery and designing research projects; however, some registries have only just begun to record comorbidity scores like ASA, and most registries do not capture scores like Charlson and Elixhauser. In these circumstances, the Rx-Risk score should be an easily available tool that gives us insight into the nature and number of health conditions for which a patient has received a prescription preoperatively. Additionally, the Rx-Risk score is an equally good predictor of mortality at 30 and 90 days after joint replacement surgery as the ASA and 5-year reoperation rates [9]. Although it is weakly related to ASA, when combined with ASA, its concordance is even higher than both the scores used separately [9].

I believe that medication-dispensing record is an indirect or surrogate indicator of underlying comorbidities, whereas other scores consider comorbidities directly. Each score studied in the present paper was highly associated with the probability of mortality shortly after undergoing a procedure, but only weakly to moderately correlated among themselves. This may be the case for several reasons. For example, the Rx-Risk score may be capturing the effect of comorbidities that are missed by ASA. It may also be due to the adverse effect of multiple drugs or their mutual interactions (an effect that may be directly related to the Rx-Risk score) on mortality after THA or TKA, rather than the comorbidities themselves.

Future studies should consider examining the effects of both comorbidities and medications together and aim to come up with one score or prediction tool. This could be done by using logistic regression modeling or other machine learning–based approaches on the available databases. These approaches predict the effect of comorbidities and medications on other factors like age, sex, race, and/or other variables discussed subsequently on mortality and postoperative complications after arthroplasty.

Medical records, specifically the preanesthetic check-up sheets prepared by multidisciplinary preoperative assessment clinics, can give us insight about a patient’s comorbidities and the treatment the patient was taking at the time of the check-up just before going into surgery. For operating surgeons treating a patient with diabetes, knowing how much the sugar is controlled on certain medications is much more useful than only knowing that the patient took a diabetes medication.

Most of the scores predicting postoperative complications depend on the data that have been collected by joint registries or medication-dispensing records. But these scores are not mandatory in most countries, and some others do not collect them at all. Therefore, the findings cannot be corroborated. We need a more accurate scoring system that can be used worldwide, that is easy to understand and interpret, and that has a strong association with patient outcomes after joint replacement surgery.

Although the Rx-Risk score is a good method to assess the comorbidities in a patient and predict mortality, I believe medication-dispensing information has low reliability, particularly in healthcare systems where many patients do not have extended health insurance and cannot afford medications. For example, patients with diabetes who cannot afford insulin would not appear in the dispensing database, but they will have a much higher risk of poor clinical and patient-reported outcomes than someone with treated diabetes. For these reasons, I think studies on the Rx-Risk score need to be validated in settings where health insurance coverage is less robust.

There is also a need for a combined scoring system that complements the information provided by each scoring system, extracts the maximum information regarding existing patient comorbidities and health conditions, and identifies the options to improve the health of patients following total joint arthroplasties.

How Do We Get There?

Tools like the Rx-Risk score may be more system-dependent than generalizable. I would suggest that we not assume that they apply globally. We should test them in not only studies like the current one, but in a variety of healthcare systems.

The leading cause of death after arthroplasty is cardiovascular disease, particularly myocardial infarction (MI). Patients with prior MI are at a higher risk for early mortality after TKA and THA [1, 2]. Similarly, diabetes mellitus is another comorbidity that is associated with an increased risk of mortality after TKA and THA [1, 2]. Some studies evaluated the correlation of age and gender with mortality after arthroplasty, and they concluded that older age and male sex are at increased risk of 30- and 90-day mortality after surgery [1, 2]. In this context, in settings that don’t track medication dispensing consistently, or where patients cannot afford the medications they are prescribed, we should record data preoperatively by means of clinical history and examination and follow the patients postsurgery to assess the mortality.

We can draft a preoperative investigation-based score consisting of important quantitative variables like complete blood count, blood sugar level, serum albumin, serum electrolyte, thyroid profile, liver and kidney function test, coagulation profile, BMI, blood pressure, SpO2, and cardiac function (ejection fraction). This can be easily used across all healthcare settings and will be easily reproducible.

We can conduct a retrospective study of big hospital-based populations in settings that lack joint registries and evaluate the correlation between ASA score and mortality shortly after arthroplasty. From the medical records of the patient, which are collected when a patient receives in-patient care, we can also obtain information about drug history and we can utilize that information in the correlation of the mortality after the arthroplasty. By using this information, we can also correlate ASA score with a drug history.

In countries where joint registries exist, more studies should be conducted in line with the current study to assess the repeatability of the scoring system. But there is also a need for a combined scoring system that complements the information provided by each existing scoring system, extracts the maximum information regarding existing patient comorbidities and health conditions, and identifies the options to improve the health of patients following total joint arthroplasties. For this scoring system, we can use multivariate regression analysis to investigate the effect of confounding factors on the overall score and subscores. Regression modeling may help us define weights for each variable, will serve to reduce the confounding effects, and may also allow us to rely more on the most accurate variables.