CORR Insights®: What Factors Predict Adverse Discharge Disposition in Patients Older Than 60 Years Undergoing Lower-extremity Surgery? The Adverse Discharge in Older Patients after Lower-extremity Surgery (ADELES) Risk Score

Where Are We Now?

Schaefer et al. [13] sought to create and validate an instrument that predicts which patients will lose independence after lower extremity surgery because of what they call “adverse discharge,” or discharge from the hospital to a long-term nursing home or skilled nursing facility. The tool they developed, the Adverse Discharge in Older Patients after Lower Extremity Surgery (ADELES) score [13], provided good-to-excellent discriminative ability in their sample of patients 60 years and older. The authors also found that the ADELES score was superior to the modified Frailty Index, Charlson comorbidity index, and Procedural Severity Scores for Morbidity and Mortality in predicting adverse discharge [13].

As healthcare systems move away from rewarding volume and toward championing value—defined as health outcomes achieved per dollar spent across the full cycle of care for a given condition [12]—studies like this [13] are especially important. To me, this paper raised two big-picture questions: (1) What are the benefits and shortcomings of risk calculators to predict outcomes that matter to patients, such as discharge disposition, and (2) how can we use information that should be available to providers at the point of care (such as data in the electronic medical record) to support surgeons, patients, and families as they engage in shared clinical decision-making?

Risk calculators can be helpful adjuncts to clinical data and expertise in patient care. A number of such calculators in elective lower extremity orthopaedic surgery have already been evaluated, including tools that predict THA and TKA revision [11], acute periprosthetic joint infection [15], and discharge to a skilled nursing facility or acute rehabilitation center after TKA or THA [3], among others. Additionally, researchers in Australia have created risk calculators to predict discharge destination after isolated lower limb fractures [7, 8]. Schaefer et al. [13] build on this prior work by designing a risk calculator focused on elderly patients only and including the reason for lower extremity surgery (such as TKA or fracture) as a variable in the model predicting adverse discharge destination.

Because shared clinical decision-making improves patient care [2], any tool that supports that process and helps surgeons better set expectations for patients and their families would be valuable. Using the instrument created by Schaefer et al. [13], surgeons can have more-informed discussions about the likelihood of patients needing a higher level of care after discharge simply by considering characteristics found with a quick review of a patient’s chart. This knowledge, which I did not appreciate in an actionable way before reading this paper, can be beneficial in informing patients and families of the possibility that the patient may lose some independence before undergoing any intervention. Indeed, more-precise prognoses on this topic can help families prepare and cope for what most will see as a serious setback, or worse. And, of course, the ability to modify factors we have control over—such as the use of general anesthesia—can only help us take better care of patients.

Where Do We Need To Go?

There are three main areas that future studies of risk calculators should seek to improve on: (1) ensuring accuracy, particularly the accuracy of data inputs and personalized patient information; (2) adding patient-reported outcome measures in risk calculators, whenever possible; and (3) incorporating more-appropriate measures of socioeconomic disadvantage and systemic racism, including implicit bias and generational trauma, as measures become available instead of proxy variables (such as self-reported race or neighborhood income) [1, 9].

Overall, risk calculators are only as accurate as the data inputs provided. In addition, it is imperative that surgeons stay vigilant in the use of risk calculators, ensuring that the populations in which the risk calculator was developed are similar to their own. If not, the tool will likely be inaccurate and may provide false—and misleading—information. Thus, future research is warranted in more patient subgroups. This was appropriately emphasized by the authors of the current study [13].

It also is vital that future risk calculators incorporate measurements of the endpoints that are most important to patients, in particular, patient-reported outcome measures. More-extensive use of patient-reported outcome measures in risk calculators may not only improve model performance but also provide a more-concrete stepping-off point to begin clinical shared decision-making discussions and set patient expectations, because most patient-reported outcomes tools are readily explainable to and easily understood by patients.

Socioeconomic factors have been shown to impact discharge disposition in patients undergoing lower extremity orthopaedic surgery [5, 14]. However, these measurements are only proxy variables for systemic racism, implicit bias, and generational trauma. As we improve our understanding of these ideas and determine how best to measure them, it will be important to include them and revalidate existing risk calculators.

How Do We Get There?

Improving risk calculators requires a commitment by surgeons to collect accurate data, to find ways to gather a wider array of data such as patient-reported outcome measures and accurate measures of socioeconomic disadvantage, and to pledge to push the boundaries of clinical research through new, innovative methods, including through the cautious use of machine learning [4, 6, 10]. Continued improvement of novel risk calculators such as the ADELES score may lead to higher-value, more-personalized care for all patients with orthopaedic conditions.