Context: The Current State of Inequity in Patient Safety
There has been significant discussion regarding the Social Determinants of Health (SDOH). There is evidence that the impact of social determinants has extended more particularly into any healthcare domain affected by adverse events. Consequently, as called out by our inaugural Virtual Patient Safety (VPS) white paper released in August 2020, Pascal has coined a new term to account for this impact: the Social Determinants of Safety (SDOS).
This blog reviews some of the recent evidence that has emerged, showing that the data generated by voluntary event reporting – the primary method upon which global healthcare has historically relied for identifying adverse events – clearly reflects racial and social biases.
Until the field and the industry proactively seeks to identify any adverse event in every patient all the time – without any racial, social, or other demographically demarcated bias – the healthcare system and the dedicated clinicians will continue to be precluded from delivering care with an equitable approach to patient safety, quality, and risk of preventable injury or death.
The Papers: Approach
There are three papers and two editorials that we will cover here:
- Stockwell et al published in Hospital Pediatrics in January of 2019.
- Halvorson et al – Editorial published in Hospital Pediatrics in January of 2019.
- Thurtle et al published in Hospital Pediatrics in February of 2019.
- Shulson et al published in BMJ Quality & Safety in October of 2020.
- Chin – Editorial published in BMJ Quality & Safety in October of 2020.
- Thomas et al (on harm) published in the Journal of Patient Safety in December of 2020.
- O’Kane et al published by the National Academy of Medicine in Perspectives in September of 2021.
- Thomas et al (on near misses) published in the Journal of Patient Safety in December of 2021.
There appears unsurprisingly to be more interest in studying equity issues as related to adverse events in recent years. The incidence of adverse events impacts patient safety, quality, and risk.
The approach of each was as follows:
- Stockwell et al: This study seeks to understand whether the racial/ethnic and socioeconomic disparities found in the patient safety and quality extend to the pediatric setting. The co-authors used the previously validated Global Assessment of Pediatric Patient Safety (GAPPS) Trigger Tool. The researchers then applied GAPPS to randomly selected pediatric patients from 16 hospitals across four U.S. regions between January 2007 and December 2012.
- Thurtle et al: This study tested the hypothesis that “patient demographic factors such as weight status and race would be associated with safety event reporting in the acute care setting.” Researchers analyzed acute care encounters for patients two to 17 years of age and associated safety events entered in the voluntary event reporting system of a tertiary-care children’s hospital from February 2015 to February 2016. Data used included patient demographics, clinical characteristics, incident description, and reported harm score, and x2 and multivariable logistical regression methods supported the analysis.
- Shulson et al: Researchers tested the hypothesis that patients representing racial/ethnic minorities and other vulnerable populations may be at higher risk of patient safety events. This retrospective cohort study was conducted at a single tertiary care academic medical center. Participants were Inpatient admissions of those aged 18 years or older from October 1, 2014 to December 31, 2018. The primary outcome analyzed was the total number of patient safety events, defined as any event identified by a modified version of the Institute for Healthcare Improvement global trigger tool that automatically identifies potential patient safety events using EHR and voluntary event reporting system.
- Thomas et al: The study’s aim was to determine whether “race differences” exist in voluntarily reported patient harm events in a 10-hospital system on a high reliability organization (HRO) journey. The research used two years of data from July 1, 2015, to June 30, 2017. Inpatients, outpatients, and observation patients were identified as “black,”“white,” or “other” (N = 5038). Comparisons of race proportions were con-ducted to determine differences at the health system level, by hospital, and by event type.
- O’Kane et al: On the 20th anniversary of the publication of To Err Is Human: Building a Safer Health System (IOM, 2000) and Crossing the Quality Chasm: A New Health System for the 21st Century (IOM, 2001), the National Academy of Medicine brought together leaders of seven “prominent” U.S. health care quality organizations to “discuss and author a paper identifying the most important priorities for the health care quality movement in the next 20 years. The authors identiﬁed equity as the area of most urgent and cross-cutting concern for the ﬁeld. This paper summarizes the authors’ conclusions about key barriers and strategies to advancing equity in health care quality.”
- Thomas et al: A follow-on study to the first Thomas et al article aimed to determine whether race differences exist in voluntarily reported near miss events in the same 10-hospital system on an HRO journey. The research used two years of data from July 1, 2015, to June 30, 2017. Inpatients, outpatients, and observation patients were identified as “Black,”“White,” or “other” (n = 39,390). Comparisons of race proportions were con-ducted to determine differences at the health system level, by hospital, and by event type.
The Papers: Findings
Here’s what each of the papers found:
- Stockwell et al: The trigger-based GAPPS tool identified racial and/or ethnic and socioeconomic disparities in rates of adverse events of hospitalized children across geographies and settings. Compared to hospitalized non-Latino white children, hospitalized Latino children experienced almost twice as many adverse events, almost twice as many preventable adverse events, and approximately 50% more high-severity harm. In addition, “compared with privately insured children, publicly insured children experienced higher rates of preventable AEs.”
- Thurtle et al: The co-authors found associations between patient demographic factors and voluntary event reporting in the acute care setting and concluded that, In future studies, “we will compare VER to event identiﬁcation by more objective measures, such as a trigger tool.” Of a total of 22,056 patient encounters, 341 (1.5%) of those had a reported safety event. “In univariate analysis, age, weight category, and race were found to be signiﬁcantly associated with event reporting, whereas sex and insurance provider were not.”
- Shulson et al: This study found that the widespread method used to identify patient safety problems, namely voluntary event reporting, may under detect adverse events in vulnerable populations. The researchers studied 141, 877 hospitalizations, of which 13.6% had any patient safety event. Including analytic adjustments, Asian race/ethnicity was associated with a lower adverse event rate; limited English proficiency (LEP) patients had a lower risk of any patient safety event and voluntarily reported events. Asian and Latino race/ethnicity related events also reflected lower rates of being reported, but there was no difference in the risk of events being identified by the automatic trigger-based method. Black race was associated with an increased risk of automated events as compared to the voluntarily reported approach.
- Thomas et al: “Significant race differences existed: (1) overall with higher proportions of whites and lower proportions of other in a Patient Safety Event Management System; (2) by type across races; (3) in six hospitals across races; and (4) by type and by hospital for blacks and whites.” The co-authors concluded that, “Race differences in harmful events exist in voluntary reporting systems by type and by hospital setting.”
- O’Kane et al: The first conclusion reached for the racial equity agenda is to “embed an equity lens into all quality and safety improvement activities.” The second conclusion related to “quality dashboards” ensuring that equity data are presented to health system leaders. The third conclusion called for more leadership commitment and infrastructural investment. The paper closed by suggesting that, “Perhaps most importantly, it [the health care system] can lead the way for other sectors by establishing a mea-surable and transparent racial equity agenda and hold-ing itself accountable.”
- Thomas et al: “Significant race differences existed: (1) overall across the health care system with higher proportions of events reported for Whites and lower proportions of events reported for Blacks in the Patient Safety Event Management System, (2) by site in 9 of 10 hospitals, and (3) by type.” The co-authors concluded that, “Race differences in near-miss patient safety events exist in voluntary reporting systems by type.”
The Papers: Editorials
Two editorials that appeared commenting on Stockwell et al and Shulson et al were, respectively, Halvorson et al and Chin.
In the Halvorson et al – Editorial published in Hospital Pediatrics in January of 2019, the co-authors remind us that:
- “Strategies to decrease harm are difﬁcult to implement without ﬁrst identifying patients who are at high risk.”
- “Traditionally, the detection of adverse events has relied on voluntary event reporting by providers, which is subject to bias from providers and in which the true incidence of events is underestimated.”
- “Administrative database reviews that include discharge codes have been used in lieu of individual chart reviews in the past, but this automated approach is limited regarding correctly identifying adverse events.”
- “The recently validated Global Assessment of Pediatric Patient Safety (GAPPS) Trigger Tool is used to provide a more objective measure of adverse events in pediatric inpatient care and allows for providers to use focused chart review to identify adverse events in children who are hospitalized.”
One of the most significant statements by Halvorson et al is that “The efﬁcient implementation of preventive strategies requires an understanding of which patients are at the highest risk of harm during hospitalization.”
While seemingly obvious, health systems are generally not measuring with clinically validated adverse event outcomes using EHR data.
Halvorson et al continues by adding that, “Therefore, a second, important next step for the ﬁeld is to develop risk-prediction models for different types of adverse events and identify optimal prevention strategies for each event type and population.”
The key point here is that measurement must be conducted for “different types of adverse events”. Unless healthcare organizations are measuring at the fine-grained level, this data will not be available for measurement and management.
In the Chin – Editorial published in BMJ Quality & Safety in October of 2020, Dr. Marshall Chin forwards a framework for addressing the equity issue in a healthcare delivery context. Salient points included:
- “Hospitals, clinics and health plans are looking inwards to identify organisational biases and discrimination, and developing outward interventions to advance health equity for their patients. Looking in the mirror honestly takes courage; frequently the discoveries and self- insights are troubling.”
- “Schulson et al’s study in this issue of BMJ Quality and Safety, finding that voluntary incident reporting systems may underdetect safety issues in marginalised populations, is an important sentinel event.”
- “Implicit bias in providers and structural bias in safety reporting systems might explain this underdetection of problems.”
- “The patient safety field should move faster, incorporating major advances that have occurred regarding how to reduce health disparities.”
- One of Dr. Chin’s recommendations is to “Examine safety criteria and systems for bias. Design and implement equitable systems for identi-fying, measuring and eliminating safety problems.”
Chin writes that, “…if we are to make progress in reducing adverse events, the field must measure preventable patient harm. How can we target individual adverse events or patterns of harm if we do not have data on where, when, how, and why they are occurring?”
The answer is: we cannot. The field must measure “individual adverse events” if we are to know “where, when, how, and why they are occurring.”
Further, there are those who point to machine learning, AI, or other advanced analytics as the answer – i.e. predicting our way to performance. However, without measuring outcomes, we don’t have the optimal data (i.e. validated outcomes) to train these models. Safety in high risk industries has not been achieved – and will not in healthcare – by relying on improving safety culture alone or by prediction alone. Instead, measurement, monitoring, and “instrumenting the enterprise” with real-time surveillance of risks is critical for robust operational improvement.
Implications for Future Study
Future study should be informed by hypotheses as to where inequities may reside, including but not limited to the following:
- Bias in identification of safety events:
- Bias of the reporter, which leads to the inequitable distribution of voluntarily reported events;
- Overrepresentation of safety events is another, meaning that minority groups experience harm at higher rates than do e.g. white groups;
- Bias in intervening or improving using safety event data:
- Bias of the clinician(s) selecting and prioritizing intervention or improvement initiatives; and
- Bias in resourcing and executing intervention or improvement initiatives.
The Papers: Final Thoughts
Thus far, the literature is showing that racial and social bias is reflected in voluntary event reporting data in both adult and pediatric settings.
Traditional and largely current patient safety programs relying largely if not exclusively on voluntary event reporting to inform operational improvements is highly problematic for three reasons.
- Aside from the clinical and operational superiority of an automated surveillance method covered elsewhere [link], the emerging evidence – which we expect to mount based on real world evidence Pascal has reviewed – shows that voluntary event reporting data contains a significant amount of racial and social bias.
- Common sense would suggest that in an operating environment where even pre-pandemic clinicians were suffering from emotional exhaustion and burnout from demands on their time, including from overwhelming EHR documentation requirements [link], why would a clinician at the end of a long day – or even late at night back at home – not report only the most important actual or potential safety events in order to survive?
- If the most pre-eminent leaders convened by the National Academy of Medicine call to “embed an equity lens into all quality and safety improvement activities,” how is it even possible to rely on inequitable (and unreliable, based on other evidence) voluntary event reporting data? How is it possible to create the recommended equity and quality dashboards with inequitable safety information? Health systems can’t and shouldn’t.
Indeed, a “lens” using AE Outcomes that is accurate, timely, and actionable…and reduces if not eliminates the bias of reporting is required.
We do not fault our dedicated clinical colleagues who day-in and day-out do whatever it takes to deliver the best care possible, particularly since the onset of COVID-19 when the sacrifices became almost unfathomable.
Rather, we recognize that being in the 21st century offers the opportunity to use electronic data to proactively look for all types of preventable patient harm all the time for all patients. It stretches the imagination to suppose that clinicians in any given hospital at all times can serendipitously observe every type of preventable harm or antecedent event that might lead to that harm, leave alone have the time to report that information in a timely manner.
Pascal believes that in five to 10 years, the field will look back on this expectation with incredulity.
What’s worse, it’s inequitable.
It’s time for the field to move from the necessary but insufficient calls to action, changes in mission, and conferences to align to an action-oriented plan of adopting and implementing methods and technologies that are evidenced-based, proven to work, and already reducing if not eliminating the Social Determinants of Safety (SDOS) in harm identification.