Across the last year, we’ve been tracking worrying headlines about the decrease in hospital quality across the U.S. In fact, the Joint Commission reported a 19% rise in adverse events in 2022. We began to field concerns about the state of healthcare quality today, and what to do about it given the complexities of the workforce crisis, hospital financial pressures, and a backlog of delayed care.

But “quality” is a nebulous term, and without greater specificity about the problem itself, it would be challenging to articulate meaningful next steps. Our research team set out to uncover where exactly our industry’s progress toward quality goals was backsliding, what themes we could surface about why outcomes are struggling, and key priority areas for performance improvement.

We conducted a data analysis based on the Agency for Healthcare Research and Quality’s (AHRQ) open access National Healthcare Quality and Disparities Reports, which aggregated national performance on 127 unique metrics. To identify the areas with the most urgent drops in performance, we created a shortlist of the metrics that were:

• Furthest from their benchmark, on average.
• Worsening over time, on average.
• Marked by racial disparities in performance.

Here’s what our analysis revealed:

• Metrics related to unmanaged chronic diseases dominated the list of 24 quality measures that were furthest from their benchmarks, with 13 metrics performing at least 2x-5x worse than their benchmarks, and one performing 25x worse. Opioid-related deaths topped the list, followed by acute utilization for hypertension, diabetes, heart failure, asthma, and COPD. Other metrics that made the list include hospital admissions for immunization-preventable flu cases, overall new HIV cases, and suicide death — all of which indicate a need to bolster preventive care.
• This trend continued when we pulled the list of metrics where national performance was getting worse over time — but this time, the metrics solely revolved around opioid use and diabetes. Of the eight metrics demonstrating a 5% or greater annual percent change to the negative, five were related to opioid use disorder deaths and acute utilization, and three were related to diabetes acute utilization. In fact, the performance of two opioid-related metrics worsened by 20% from one year to the next.
• When looking at quality performance on a national scale, the data clearly pointed at two clinical focus areas (opioid use disorder and diabetes), indicating a broader challenge with chronic condition management. However, the story was more nuanced when we broke down performance by reported racial group. Of 127 total metrics represented in the data, 109 demonstrated racial disparities. Chronic condition-related measures popped up among the measures with the largest gaps (specifically related to heart conditions), but diabetes metrics weren’t as close to the top, and opioid use disorder metrics didn’t show up at all. Instead, new focus areas rose to the top: obstetric trauma, post-acute care-related patient experience metrics, vaccination-related metrics, and HIV care.

Ultimately, there were only four metrics that showed up in all three of our data cuts (far from their benchmarks, getting worse over time, and marked by racial disparities): diabetes-related hospital admissions with short-term complications, diabetes-related hospital admissions with long-term complications, heart failure-related hospital admissions, and patient experience during discharge planning.

For the specific metrics we identified, and national performance across each, download this tool.

Our analysis revealed three areas where the surface level story differs from the more complex reality going on underneath.

Provider organizations can only develop mature quality improvement arms when leaders begin to dig below the surface — by assessing whether the metrics they collect tell the full story of patient outcomes, proactively uncovering demographic disparities in their data analyses, and ensuring organizational investment goes beyond lip service. But every organization has to start somewhere. Assess your organization’s quality improvement capabilities using this maturity model to chart a path forward.

Julia De Georgeo led the production of this report. Don Malott contributed to its data analysis.

AHRQ’s dashboards feature data from a range of sources, include AHRQ, CDC, and CMS, from various years. The database aggregated national performance over time on 127 unique metrics compared to their benchmarks set by CMS. The metrics tied to various settings (58 were most closely tied with post-acute care, 25 were hospital-based, 34 were primary or specialty care-based, and 10 were miscellaneous), and covered episodic care, prevention, chronic care, and patient experience. Importantly, the database also assessed national performance by racial group when possible (including American Indian/Alaska Native, Asian, Black, Mixed-Race, Native Hawaiian/Pacific Islander, and White).

To identify the areas with the most urgent drops in performance, we aimed to create a shortlist of the metrics that were:
 

1. Furthest from their benchmark on average. The metrics available were reported the form of percents, rates by 1,000 people, and rates by 100,000 people. To “true” the data, we converted all values to rates by 100,000. Then, we calculated the percentage point difference of each metric from its benchmark, and narrowed our list to the worst performers.
2. Worsening over time on average (e.g., for all racial groups). We sorted the metrics trending in the wrong direction by calculating the average annual percent change.
3. Indicating the largest racial disparities in performance. For each metric, we calculated percentage point difference between the highest and lowest performing racial group. In this case, we looked at the metrics reported as percents separately from the metrics reported as rates. We “trued” the metrics represented as rate per 1,000 to rate per 100,000 for accurate comparison.

No dataset is perfect. We can only draw insights from the data we have access to, and there are likely some clinical areas that are not adequately represented by this source, even though it includes 127 measures. And because the AHRQ reports combined a couple of datasets, the metrics themselves vary in the way they are reported (as a percentage of the general population, as rates per 1,000 people, and as rates for 100,000 people), making it challenging to compare apples to apples. In addition, the year the data was gathered ranged from 2015-2020, representing a significant lag to today’s dynamics. Lastly, there’s even less standardization for the reporting of data by racial group — some metrics have performance data across six separates groups, and some only have two.