See our detailed methodology for how we created each metric, grade, and ranking for the Lown Hospitals Index.Download
The Equity grade combines assessments of community benefit spending, pay equity, and inclusivity to evaluate hospitals’ commitment to community health and civic leadership. Pay Equity, Community Benefit, and Inclusivity were weighted at 20%, 40%, and 40% respectively, to create the overall Equity grade.
The pay equity component measures the difference in compensation of hospital executives compared to health care workers without advanced degrees.
We obtained data for Chief Executive Officer (CEO) compensation from three different sources: the Securities and Exchange Commission’s (SEC) database, public payroll data, and IRS 990 filings from 2019. When CEO pay was unavailable, this information was imputed (estimated) using known values in regression models. For hospitals within systems (two or more hospitals), we distributed the system CEO’s salary among the constituent hospitals equally.
We obtained average worker wages from the CMS Healthcare Cost Report Information System (HCRIS) . We included lower wage staff, such as janitorial and kitchen staff, and medical records personnel, and excluded professional staff such as physicians and nurse practitioners, whose jobs require specialized degrees. For hospitals that had incomplete wage index information in HCRIS, we substituted the average wage for hospitals in the same state and urban or rural status.. We then estimated hourly wages for CEOs based on the work hours listed in their IRS forms, defaulting to 46 when the hours were not listed (the average hours worked among hospital CEOs), and calculated a ratio of CEO pay to average worker pay.
The community benefit metric measures hospital spending on charity care and other community health initiatives, as well as their service of Medicaid patients. Community benefit is a composite of three details: Charity care, Medicaid revenue, and Community investment. For hospitals with data available for all three metrics, each metric was weighed equally in the composite at ⅓ of the total score. For hospitals with data for two of the metrics available, each metric was weighed equally in the composite as half of the total score.
Charity care, or financial assistance, is free or discounted care provided on the basis of the patient’s financial situation. We measured charity care as a share of total expenses using the Centers for Medicare and Medicaid’s Hospital Cost Reports (HCRIS). Medicaid patient revenue was measured as a proportion of total patient revenue using HCRIS data from 2019.
Hospital spending on certain other types of community benefits, as a share of total expenses, was calculated using 2019 IRS data. For hospitals that filed with multiple hospitals as one tax entity, each individual hospital’s community benefit spending was estimated by prorating based on each hospital’s share of system revenue.
This metric includes a subset of community benefit spending that we deemed to be meaningful: subsidized health services, such as free clinics, some emergency services, telehealth services, and other services provided at a loss to the hospital; community health improvement activities such as health fairs, community health education classes, immunizations, interpreter services; contributions to community organizations; and community building activities that help increase the capacity of the community to address health needs and often address the "upstream" factors, or social determinants, which impact health, such as education, air quality, and access to nutritious food.
The inclusivity component evaluates the extent to which a hospital’s patient population reflects the demographics of the community in which it is located, based on race, income, and education levels. To create this metric, we used Medicare claims from 2020 and the U.S. Census Bureau’s American Community Survey from 2020.
Each hospital’s inclusivity score shows how the demographics of the hospitals’ community area (who the hospital could serve) compare to their actual patient population (who the hospital does serve). The “community area” radius is defined by the distance from which about 90% of the hospital’s Medicare patients travel. We used OpenStreetMap to measure travel time for the community area radius.
To calculate community area demographics, we applied Census data for people over the age of 65 on race, income, and education levels within all zip codes that fell within the defined hospital community area. We exponentially reduced the contribution from zip codes beyond the point at which 50% of a hospital’s patients had come. These demographics were then compared to the zip code demographics of the hospital’s actual patient counts.
Hospitals received higher scores if they had higher patient counts from zip codes with greater proportions of non-white patients, lower incomes, and lower levels of education, compared to their community area. Inclusivity by race, income, and education were weighted equally to create the overall inclusivity score. Hospitals in racially homogenous areas (97% or more white) were only scored on inclusivity by income and education.
The value of care category reflects hospitals' avoidance of unnecessary care and ability to achieve good outcomes without overspending. The value of care category is based on a two components, Avoiding Overuse and Cost Efficiency, weighed at a 2:3 ratio to create the overall value grade.
Overuse (also known in the clinical community as low-value care) is the delivery of a health care service that is more likely to harm than benefit the patient. The component comprises rates of overuse of 12 commonly overused medical services, chosen based on their validation in previous research on measuring overuse. Some of these services have been shown in high-quality clinical trials to be ineffective and are always considered overuse. Others are only considered low value when prescribed to patients with certain conditions.
To create this metric, we used Medicare claims data from January 1, 2018 – December 31, 2020. We measured 12 low-value services, four tests and eight procedures (see below). Only hospitals with the capacity to do four or more services were ranked. Hospitals’ overuse score for each service is based on the rate of overuse as well as the volume of overuse. The overuse composite ranking is based on all services, with more weight placed on the services that make up the larger share of overuse.
Arthroscopic knee surgery – Surgery to remove damaged cartilage or bone in the knee using an arthroscope (tiny camera). Defined as overuse for patients with osteoarthritis or “runner’s knee” (damaged cartilage). Excluding patients with meniscal tear.
Carotid artery imaging for fainting – A test to screen for carotid (neck) artery disease. Considered overuse for patients where syncope (fainting) is the primary diagnosis, and there is no history of syncope in the past two years. Excluding patients with stroke or mini-stroke, retinal vascular occlusion/ischemia, or nervous and musculoskeletal symptoms.
Carotid endarterectomy – Procedure to remove plaque buildup from a carotid (neck) artery in a patient to prevent stroke. Considered overuse when performed on female patients without stroke symptoms or history of stroke.
Coronary artery stenting – Also known as percutaneous coronary intervention. Procedure to place a stent or balloon in a coronary artery using a catheter. Also includes procedures where the artery is opened but a stent is not inserted. Defined as overuse when performed on patients with stable heart disease (not having a heart attack or unstable angina). Excluding patients with past diagnosis of unstable angina.
EEG for fainting – A test of the electrical activity of the brain. Considered overuse for patients where syncope (fainting) is the primary diagnosis, and there is no history of syncope in the past two years.
EEG for headache – A test of the electrical activity of the brain. Defined as overuse for patients with headache as the primary diagnosis on the claim and no history of headache in the past two years. Excluding patients with epilepsy and recurrent seizures, convulsions, and abnormal involuntary movements.
Head imaging for fainting – Considered overuse for patients where syncope (fainting) is the primary diagnosis, and there is no history of syncope in the past two years. Excluding patients with epilepsy or convulsions, cerebrovascular diseases, head or face trauma, altered mental status, nervous and musculoskeletal system symptoms, and history of stroke.
Hysterectomy – Procedure to remove the uterus. Considered overuse for patients without a diagnosis of cancer or carcinoma in situ.
Inferior Vena Cava (IVC) filter – Procedure to place a filter (a medical device) in the large vein in the abdomen to prevent blood clots from moving to the lungs. Considered overuse for all patients except those with history of multiple pulmonary embolism.
Renal artery stenting – Procedure to place a stent or balloon in the renal (kidney) artery. Considered overuse for patients with high blood pressure or plaque buildup in the artery. Excluding patients that had diagnosis of fibromuscular dysplasia of the renal artery (abnormal twisting of the blood vessels).
Spinal fusion/laminectomy – Procedure to fuse vertebrae together (spinal fusion) or remove part of a vertebra (laminectomy). Defined as overuse for patients with low-back pain, excluding patients with radicular symptoms, trauma, herniated disc, discitis, spondylosis, myelopathy, radiculopathy, radicular pain or scoliosis.
Vertebroplasty – Procedure to inject cement into the vertebrae to relieve pain from spinal fractures. Considered overuse for patients with spinal fractures caused by osteoporosis. Excluding patients with bone cancer, myeloma, or hemangioma.
The cost efficiency metric is a ratio of hospitals’ mortality rates compared to their Medicare costs. Hospitals with the lowest mortality and the lowest costs received the best scores in cost efficiency.
For cost efficiency, hospitalizations with probable or confirmed COVID-19 were excluded.
For mortality, we used hospitals’ risk-standardized 30- and 90-day mortality for Medicare patients hospitalized between 2018 and 2020. For cost, we used 30- and 90-day total risk-standardized Medicare payments for patients hospitalized in 2018 to 2020. Payments were standardized for hospital patient risk (the conditions that hospital patients have as well as the procedures they received), so that hospitals with sicker patients were not punished for spending more to treat them. Payments were also adjusted for patient survival, so that hospitals with low patient survival did not have artificially lower costs. Lastly, payments were adjusted for Medicare's regional cost differences, so that hospitals that get paid more by Medicare because of regional differences did not appear to have higher costs.
For each hospitalization, we found the claim payment amount in all claims within 30 or 90 days from the admission date. These claims included: inpatient, outpatient, carrier, skilled nursing facility, home health agencies, durable medical equipment, and hospice claims. We excluded any claims where Medicare denied the payment. Transfers were excluded.
Medicare adjusts their payment amounts to hospitals and other providers based on various geographic factors. To account for this, we calculated standardized payments using the Virtual Research Data Center’s public use files of 2018 to 2020 Hospital Referral Regions (HRR) standardized ratio tables for patients over 65.
Our goal for the cost efficiency score was to reward hospitals with low mortality rates and low costs, and give the lowest scores to hospitals with high mortality rates and high costs. We also decided to bias our scores to give hospitals with high costs and low mortality a higher score than hospitals with low costs and high mortality. This is because we believe that if there is a trade-off between costs and mortality, we should favor better mortality rates compared to lower costs.
The patient outcomes category reflects a hospital's performance as it relates to their patients' health and experience of care. This category is calculated from three components: clinical outcomes, patient safety, and patient satisfaction, which were weighted at a ratio of 5:2:1 respectively.
The clinical outcomes component measures how well the hospital keeps patients alive and prevents return trips to the hospital, over various periods of time.
For mortality and readmissions metrics, hospitalizations with probable or confirmed COVID-19 were excluded.
Using Medicare inpatient claims data from 2018-2020, we evaluated hospital clinical outcomes based on mortality and readmissions. The clinical outcomes is composed of risk-standardized rates of mortality and readmission, weighted at 80% and 20% respectively.
Mortality included rates of in-hospital, 30-day, and 90-day mortality. We chose these mortality endpoints to cover measurements in CMS' inpatient quality reporting programs as well as more extended periods, when mortality is a function of both hospital and community. We also included risk-standardized rates of 7- and 30-day readmission, to include both a shorter interval that would reflect inpatient care quality, and longer follow-up that would reflect post-hospital community support.
Hospitalizations and readmissions rates were risk adjusted using the Risk Stratification Index (RSI), a Lown Institute-specific version of a machine-learning algorithm in the public domain that has been validated on multiple national, state-based and hospital-based datasets using billions of insurance claims. RSI has been shown to predict outcomes with greater discriminatory accuracy compared with other publicly available risk adjustment tools, by adjusting outcomes for numerous patient conditions and procedures. This allows us to better compare outcomes across patients that have the same medical needs.
In addition to the patient conditions in RSI, we accounted for differences in hospital volume and case mix. At both the patient and hospital level, we included model effects for patients dually eligible for Medicare and Medicaid as an adjustment for socioeconomic factors.
For this year, we included new adjustment for hospitals having disproportionately sicker or healthier patients by including patient risk mix within the model. This means that the proportion of high-risk patients at a hospital is taken into account when looking at clinical outcomes.
The patient safety component measures how well hospitals avoid preventable patient safety errors. We used well established indicators provided by CMS on its Hospital Compare website for hospitalizations--such as rates of pressure ulcers, accidental punctures, and central intravenous line infections--for hospital admissions in 2018.
We included the CMS composite measure (PSI-90), which comprises 10 separate indicators of patient safety, as well as five hospital acquired infection (HAI) measures. We included a reliability adjustment for the HAI measures using the reported numerator and denominator counts from Hospital Compare. For a patient safety overall score, hospitals had to have had at least three of the PSI-90 or HAI results.
The six detailed metrics were weighted equally to formulate the overall Patient Safety component score. Like CMS, we excluded critical access hospitals from this metric due to lack of available data.
The patient satisfaction component measures aspects of the hospital experience as reported by patients. For this component we used CMS's annual Hospital Consumer Assessment of Healthcare Providers and Systems (HCAHPS) survey from 2018 to give a rating of patient experience across 10 factors, including hospital quietness, cleanliness, health care staff responsiveness, and communication. We took the average of the 10 linear mean scores of these factors published on the 2018 Hospital Compare site.
Hospital Compare did not report scores for hospitals that had fewer than 100 responses. We chose to include hospitals with between 50 and 100 responses after data analysis indicated that imputation of these scores would be reasonable to account for CMS’s mean calculations and adjustment. We calculated scores for these hospitals by extrapolating to the nearest median score of hospitals with similar survey responses.
THESE RANKINGS PROVIDE A MUCH-NEEDED AND LONG-OVERDUE FRAME THROUGH WHICH TO TRULY JUDGE THE VALUE OF OUR HOSPITALS."
People's lives depend on hospitals are making the right decisions. Let’s make sure they are guided by the most relevant tools possible, tools that measure what matters.