Insight
July 3, 2020

Patient mortality at safety net hospitals

Why do safety net hospitals tend to score lower on the Lown Index for patient mortality? This pattern is not due to just one reason; it is the product of many factors.

By Vikas Saini, Shannon Brownlee, and Judith Garber

Key Takeaways

  • On the Lown Institute Hospitals Index, safety net hospitals have lower average rankings for risk-adjusted overall mortality compared to non-safety net hospitals.
  • Safety net hospitals’ patients are more likely than those cared for at non-safety net hospitals to be insured by Medicaid or Medicare. This means safety net hospitals receive lower reimbursements on average and are often under-resourced, which may affect patient outcomes.
  • Safety net hospitals are often located in underserved areas, which makes it more likely that patients arrive at the hospital with more advanced cases of illness compared to patients living in wealthier communities—thus impacting clinical outcomes.

Can hospitals “do it all”? Can they achieve excellent patient outcomes while also avoiding unnecessary care and serving as civic leaders? The results of the Lown Index show that some hospitals are performing well in all three categories—but such hospitals are the exception.

In general, we found that hospitals with the highest scores in our patient outcomes category are not the hospitals with the highest scores in civic leadership, a category that includes the level of community benefit provided by the hospital, the ratio of CEO pay to average worker wages, and how inclusive the hospital was regarding its patient population. In fact, there is no overlap between the top 100 hospitals in these two categories. While many well-known hospitals tend to achieve high rankings in patient outcomes, few of them excel in civic leadership; likewise, the safety net hospitals and public hospitals at the top of our civic leadership category tend to have below average scores in patient outcomes, specifically mortality outcomes (see Table 1).

Number of hospitals Average clinical outcomes rank In-hospital mortality (avg percentile rank) 30-day mortality (avg percentile rank) 90-day mortality (avg percentile rank) 1-year mortality (avg percentile rank)
Not safety net 2668 1460 56.51 54.47 56.50 57.09
Safety net 692 2537 24.90 32.75 24.92 22.65

Why do safety net hospitals score lower on the Lown Index for patient mortality compared to non-safety net hospitals? This pattern is the product of many factors.

Resources at safety net hospitals

Differences in clinical resources available to safety net hospitals compared to non-safety net hospitals may affect clinical outcomes. Safety net hospitals have been historically under-resourced and are more likely than most hospitals to care for patients who are under- or uninsured, which limits their ability to bring in revenue.

Even with the additional payments safety net hospitals receive from state and local agencies, some struggle to stay open. According to a recent report from America’s Essential Hospitals, safety net hospitals operate with margins about a third that of other hospitals but spend more than non-safety nets on uncompensated care. All hospitals that serve mostly Medicare, Medicaid, and uninsured patients receive lower reimbursements than hospitals that take mostly privately-insured patients, and thus may be punished financially for taking on poorer patients.

It would not be surprising if these more inclusive hospitals have fewer resources and less-skilled staff than many of the hospitals that achieve better scores on our clinical outcomes metrics. In fact, disparities in outcomes for patients with Covid-19 in certain hot spots have been attributed in part to systemic lack of funding and resources at safety net hospitals.

If we are to get serious about tackling structural racism and population health, then both safety net hospitals and the communities they serve will need more support and resources to achieve health equity."

What risk adjustment misses

We believe it is also likely that safety net hospitals are taking care of patients who are sicker than those at hospitals that score well on patient outcomes. The principal reason for this belief has to do with the limitations of risk adjustment. For all of our clinical outcomes measures, we use a risk adjustment tool called the Risk Stratification Index (RSI), to adjust for patient conditions. All risk-adjustment methods, including ours, help create an “apples to apples” comparison of outcomes among hospitals, because we are measuring outcomes for patients with the same conditions against each other in our metrics.

While RSI has better discriminatory accuracy than other risk-adjustment tools, even the best risk adjustment tools do not yet adequately take health disparities and differences in community conditions into account. This limitation is likely one reason why the outcomes at safety net hospitals are lower. For example, a patient with diabetes who has access to regular health care and nutritious food, and who can afford their medications, will likely have a better outcome than a patient with diabetes who does not have access to these benefits, regardless of how well that second patient is cared for in the hospital.

Safety net hospitals tend to be located where poorer community conditions are common, such as inadequate access to good food. These conditions are compounded by a lack of community-based health care infrastructure, such as primary, specialty, and urgent care. Further challenges to accessing health care (needing to take a day off work or transportation issues, for example) make it even more likely that patients arriving at safety net hospitals are sicker than patients coming to hospitals from wealthier communities.

Measuring mortality in the short and long term

Another factor to consider is the unique scoring system of the Lown Index for clinical outcomes. Unlike other hospital rankings, our clinical outcomes metric includes rates of 90-day and 1-year mortality for patients, as well as in-hospital and 30-day mortality, which are more commonly used (although the longer-term mortality metrics were weighted less in the overall clinical outcomes score).

Our addition of these longer-term mortality metrics to our Index has the effect of shining a harsh light on existing disparities in social and environmental conditions across communities. We are aware that these metrics may penalize some hospitals, such as safety nets, because the community conditions around them are poor. We incorporated these longer-term outcomes to acknowledge recent efforts encouraging hospitals to take more responsibility for what happens to the patient after discharge.

Rates of mortality at 90-days and 1-year are likely affected by community conditions, some of which are not easily in the hospital’s control, such as the availability of safe housing, healthful food, and help with self-care. However, some factors, such as access to primary and specialty care, are in hospitals’ control. Theoretically, hospitals that wish to take responsibility for longer-term health outcomes of their patients and communities could invest in the availability of community-based health care by situating primary care clinics in underserved neighborhoods. They could also contribute to improving population health by donating to organizations that increase access to housing and healthy food, investments that could have a positive impact on longer-term mortality rates of their patients. However, the benefits of such proactive behavior would take years to show up in data and may be difficult to achieve for hospitals that are struggling with short term, day-to-day challenges. Moreover, safety net hospitals may not have the resources to make such investments.

When life expectancy is determined by your zip code more than your genetic code, it is clear that safety net hospitals face an enormous uphill battle in overcoming higher mortality rates in the communities they serve."

Conclusion

Our results indicate that hospitals’ inclusivity and mortality scores are inversely related. This is not surprising. When life expectancy is determined by your zip code more than your genetic code, safety net hospitals face an enormous uphill battle in overcoming the higher mortality rates in the communities they serve. Conversely, we don’t know whether the excellent patient outcomes seen at many wealthy hospitals, which score high on most other hospital rankings, are truly due to excellent care, or are the result, at least partly, of more privileged community conditions. In other words, hospitals with better mortality scores and low inclusivity may be getting credit for things that are not their doing, while safety nets with high inclusivity but poorer mortality may be getting penalized for factors in patient health beyond their control.

Future measurements of patient outcomes should incorporate methods to account better for differences in community conditions among different hospital populations and to better distinguish between the contributions of those conditions and of hospital quality of care to mortality rates.

Covid-19 has shown that we all have a stake in the health of communities, because threats to health anywhere can rapidly become threats to our health everywhere. In that sense hospitals that take care of poor patients and people of color are providing leadership for America's health. In the face of daunting odds, safety nets are operating at the center of the social mission of health care.

In addition, the results of the Lown Index show that if the U.S. wants to get serious about tackling structural racism and population health, then both safety net hospitals and the communities they serve will need more support and resources to achieve health equity. Otherwise, their power to alter community life expectancy and other measures of health will continue to be severely limited. We need to face these facts head on if we want all hospitals to go from good to great in every community in the nation.