Leveraging Clinical Decision Support for Racial Equity: A Sociotechnical Innovation
[ad_1]
For reasons that are structural — i.e., historically driven and embedded within our political, social, legal, and economic systems — health inequities by race and ethnicity are rampant across the United States.1 The problem has attracted growing attention in recent years, and Covid-19 has highlighted the link between racialized structural vulnerability and embodied illness throughout the country. While little progress on racial health inequities has been achieved to date,2 promising race-conscious strategies are now being considered.3,4 One of these, Healing ARC (acknowledgment, redress, and closure), is designed for clinicians and trainees to address documented, institutionally derived racial health inequities (Figure 1).5
Scholarship from critical race theorists demonstrates that racism is pervasive, built into the very fabric of our society.6 Data from our institutions, which we believe is largely representative of U.S. academic medical centers, support this contention. At Brigham and Women’s Hospital, in an initial 10-year study with 1,967 unique patients, we found that self-identified Black and Latinx patients presenting to the ED with a primary diagnosis of heart failure were less likely to be admitted to our specialty cardiology service compared to white patients — even after accounting for other social and health determinants influenced by structural racism, such as area deprivation index, comorbidity burden, insurance, and established outpatient cardiology care.7 Similar to prior studies showing improved outcomes,8–10 cardiology admission led to lower 30-day readmission rates compared with admission to general medicine. We have found similar results in a preliminary analysis at the University of Virginia Medical Center.
Because we intentionally used the Public Health Critical Race Framework6 in our study design, a post-research action component was planned from the beginning. First, however, we sought more information on the specific drivers of this racial inequity, surveying emergency providers to illuminate their decision-making process. While this follow-up study was cut short by the Covid-19 pandemic, the hypothesis-generating results suggest an as-yet poorly described driver of racial health inequities. This analysis of 61 of 135 patients admitted with heart failure (HF) during the study period (October 7, 2019 to March 20, 2020) was based on provider perception of the patient’s racial/ethnic status as Black, Latinx, or white; while it showed that there were no significant differences in admission to a cardiology versus general medicine service based on age, sex, insurance status, education level, or perceived race/ethnicity, white patients in our study were perceived by providers to show greater preference for admission to the cardiology service than Black and Latinx patients (18.9% versus 5.6%) and to self-advocate more strongly when they did so. Importantly, providers acknowledged that these appeals impacted their admission decisions.11
In discussing these findings, we recognized two related problems that required urgent attention: fixing the inequitable admissions process and providing redress for the institutionally derived racial injustice in heart failure care that we had uncovered. Given the impossibility of achieving equity by overlaying race-blind solutions onto a structurally racist system, we ultimately deemed race-blind or equality-based interventions, such as checklists or more detailed admissions criteria, to be insufficient or even potentially counterproductive or harmful.12 There are at least three reasons we instead took a race-conscious approach:
-
There is the inevitable subjectivity that arises when providers attempt to assess illness severity, complexity, or risk of decompensation, particularly in an ED setting when data are often incomplete and there are many competing pressures; this situation remains ripe for bias.
-
We know that risk stratification algorithms are particularly susceptible to racial bias even apart from human interpretation.13
-
We believe that drawing attention to past and present injustices in patient care may serve as a potential antidote to alleviating bias in care.
In light of this understanding, we developed a race-conscious implementation strategy that we hypothesize could offer a solution to both of our problems simultaneously. Utilizing the framework of acknowledgment, redress, and closure (ARC) from leading reparations scholars,14 we are now instituting a Healing ARC program at Brigham and Women’s Hospital and the University of Virginia Medical Center to operationalize equity and provide redress to the communities of color that we have inequitably served for decades. The three-stage process of Healing ARC, which has been discussed in detail elsewhere,5,15 requires health care providers and institutions to acknowledge how racism has contributed to inequitable health outcomes; to redress the harm by providing restitution to the marginalized population (including access to services and care that have historically been denied); and to facilitate closure through reconciliation and agreement that the harm has been redressed.
The Healing ARC program centers the voice of impacted community members through community-based Wisdom Councils. With members recruited by racial equity organizers from local Black and Latinx communities based on health equity interest and experience, these Wisdom Councils are convened to collaborate on the three components of Healing ARC and to ensure that each of them is commensurate with the expectations of impacted communities (Figure 2).
Early on, we advised participants that the Wisdom Council would be implemented exploratively, encouraged critique of the model’s structure and implementation, and readily integrated their feedback. Transparency has been prioritized in all Wisdom Council interactions, from funding and participant stipends to details and internal tensions regarding the discovery of the heart failure inequity to clarity on hospital leadership support.
While the University of Virginia Medical Center is at an earlier stage in the process, at Brigham and Women’s Hospital our Wisdom Council is meeting every other month and currently expanding, as initial members have identified other interested participants. Wisdom Council members are shielded from the logistic burdens of operationalization, with the goal of providing them maximum creative space to provide authentic, informed critiques, viewpoints, and recommendations. These recommendations are taken up by the Wisdom Council organizers who, as hospital employees, are tasked with gathering the resources and permissions necessary for execution, ultimately creating a collaboration that alters-traditional power dynamics. Our first priority with the Wisdom Council is to develop the concept of what acknowledgment might look like. These conversations will provide the basis of a playbook that is both a case study and a guide for other institutions looking to respond to institutional inequities by implementing the Healing ARC model in partnership with local Wisdom Councils.
We have also recognized an opportunity for sociotechnical innovation through the use of computerized clinical decision support systems (CDSS). Because specialist inpatient cardiology care is a limited resource at our institutions that we cannot otherwise immediately resolve, a CDSS-guided admission process to the cardiology service for historically excluded patients is required, at a minimum, for fair redress. Because automatic admission to the cardiology service could create a different set of inequities for patients who might be experiencing additional health issues better served on a general medicine service, we decided to pursue a targeted, race-conscious CDSS intervention. The CDSS language, vetted by community Wisdom Councils, is designed to remind admitting clinicians about our historical racial inequity and to suggest cardiology admission, if there are no other overriding reasons to prefer admission to general medicine.
CDSS have been shown to modestly increase the proportion of patients receiving desired care in a large meta-analysis,16 and for CDSS that are educational in nature, they can exert a “training effect” such that they may eventually no longer be required.17 As significant heterogeneity in effectiveness has been reported, the nature of the alert is crucial and supplementary components may be necessary to maximize potential. This broad range of effectiveness is ascribed mostly to the lack of holistic consideration of the human-computer interface, with studies to date largely failing to report specific effect modifiers including physician training, the number of total alerts given, patient volume, burnout measures, and organizational culture effects.16 We hope to improve upon the methodological rigor by actively collecting these effect modifiers.
Most important, while CDSS have been used in the care of specific vulnerable or marginalized populations, to our knowledge no studies to date have evaluated the ability of CDSS to address inequities in care, racial or otherwise. Instead, equity approaches to clinical decision support tools have primarily focused on how to avoid building bias into CDSS.18,19 Current examples include the increasing number of academic medical centers removing race-based clinical algorithms, including the calculation of estimated glomerular filtration rate that is adjusted for Black patients, that inappropriately biologize race and contribute to inequitable outcomes.20 While eliminating racism from clinical decision support is crucial, we are exploring the untapped potential of CDSS to go one step further in countering institutional racism.
To do so, we have developed a best practice advisory in our electronic medical record system that conforms to the latest CDSS research guidelines21 and is specifically built to counter implicit bias in admission decision-making (Figure 3).
The CDSS pilot implementation began at the BWH early in 2022 with support from key stakeholders in hospital leadership, ED and cardiology operations and management, and informatics. For patients being admitted to the hospital from the ED for a primary concern of heart failure, if admission to the general medical service is selected, providers will receive the best practice advisory — a nudge to positively influence the decision-making process toward equity and lasting practice change. The CDSS notification is: “Patient is from a racial or ethnic group with historically inequitable access to the Cardiology service; consider changing admission to Cardiology unless extreme census or overriding clinical reasons for General Medicine service.” The provider may override this Best Practice Advisory and continue admission to general medicine service by checking any one of three boxes:
-
Cardiology in extreme census
-
Active additional clinical reason making general medicine admission preferable (requires free-text explanation)
-
Other (requires free-text explanation)
The pop-up message is intended to be direct and race-conscious to maximize practice-changing potential for providers and to move toward equity via targeted means.22 The CDSS language and options are also calibrated to avoid automatically switching admitted patients with heart failure to the cardiology service, which could conceivably result in other negative care impacts, understanding that some patients are better served on a general medicine service. Clinicians are always free to make judgments about appropriate triage based on illness severity, and as can be seen in the exceptions listed above, there is flexibility to override the CDSS recommendations where there is a compelling rationale.
In the pilot phase, we are monitoring monthly all patients admitted with heart failure, whether the CDSS fires in each case, how clinicians respond, whether the clinician changed admission location in response to the CDSS, and each patient’s demographics, diagnoses, and hospital course to assess for appropriateness of the choices. While it is too early to draw conclusions, it appears that the CDSS tool is impacting the admission of some Black and Latinx patients with heart failure, who are now being admitted to the cardiology instead of the general medicine service. Determining if or when the CDSS is no longer required, i.e., that we have reached closure in terms of adequate redress and appropriate steps to avoid recurrence, will be based on the admissions data over time and in consultation with the Wisdom Councils.
Given that white physicians are overrepresented in the workforce and that fragility among white providers often manifests as concern around appearing racist,23,24 we suspect that CDSS referencing racial inequities will be more effective than less emotive CDSS topics that are easier to ignore or dismiss. With the long-standing lack of organized training on racism in medical education curricula, providers may also feel less confident in their initial decision compared with other CDSS topics. By combining the best practice advisory with additional antiracism trainings, we expect to maximize the CDSS practice-shifting impact while providing a blueprint for similarly primed U.S. academic medical centers aiming to advance antiracism at their institutions.
As with all approaches seeking to tackle large and historically embedded problems, ours has limitations and potential unintended consequences. Due to insufficient numbers in our patient populations, we were unable to draw conclusions about patients identifying with other racial or ethnic groups beyond Black, Latinx, and white. The lack of documented evidence is not evidence of absence, however, and a significant limitation of any intervention that applies only to statistically documented inequities is that it inevitably excludes patients from racial or ethnic groups with smaller numbers in a given catchment, including American Indian and Asian American patients in our case, who may be experiencing undetected bias or discrimination.
Because concerns about limited resources arise when attempting to expand access in a targeted way, we plan to monitor the heart failure outcomes of this approach for all patients by race, ethnicity, and other demographics. It is important to note that all patients with severe or complex cardiac problems are already triaged to the cardiology service, restricting the possibility that such a program will limit beds for severely ill cardiac patients. We are not aware of any cases where expanding access for patients experiencing marginalization hinders access for other groups, and we are confident that white patients will continue to receive the same standard of heart failure care at our institutions.
A notable concern of utilizing Healing ARC to address documented racial inequities is the possibility of its distortion and manipulation for political ends, something we have already experienced.15 This is an unfortunate but common risk of engaging in racial equity work as health professionals in the United States, particularly for Black physicians.25
As the disaggregation of data by race and ethnicity to monitor health care inequities becomes more widespread,26 academic medical centers will find themselves grappling with myriad examples like ours — if they have not already. By providing a pragmatic tool to harness existing technologies for equity and to engage impacted communities in developing reparative interventions, Healing ARC can guide health professionals and institutions on the journey from documenting inequity to antiracism action. As an organizational process, Healing ARC opens the door to further exploration of the policies that led to institutional inequities in the first place and stimulates creative strategies to address them. Most important, its insistence on transparency, fair material compensation, and community involvement provides an opportunity for a new era in institutional accountability and trustworthiness, one that is long overdue.
[ad_2]
Source link