Data Sources

Hospital Compare Data

WhyNotTheBest.org includes performance data reported publicly on the Centers for Medicare and Medicaid Services (CMS) Web site, Hospital Compare. The site includes performance data for nearly every U.S hospital in the nation, over 4,500 institutions that report data to Hospital Compare. Specifically, it includes:

• 27 Hospital Quality Alliance (HQA) measures that report how often hospitals delivered recommended care processes in the following four areas: heart attack, heart failure, pneumonia, and surgical care improvement. These process-of-care, or "core," measures are reported for all patients (i.e., not just Medicare patients). Only measures for which there are four quarters' worth of data are reported on the site. This includes three "legacy" measures, which CMS has retired and for which hospitals are no longer required to report data.
• 10 measures from the Hospital Consumer Assessment of Healthcare Providers and Systems (HCAHPS). This survey asks a random sample of recently discharged patients about important aspects of their hospital experience. HCAHPS is a relatively new survey, and hospitals across the U.S. are not yet achieving very high scores across all of the questions. Nevertheless, some hospitals are scoring significantly better than others. HCAHPS measures are reported for all patients. Patients rate each question on a scale of 0 to 10, where 0 is the worst and 10 is the best. Responses to each question are grouped into the following categories: Rating 9 or 10 (always); Rating 7 or 8 (usually or sometimes) and Rating 6 or lower (never). For example, one survey question asks how often their nurses communicated well, and respondents reported their nurses ("Always," "Usually," "Sometimes," or "Never") communicated well. For more on HCAHPS information, visit the HCAHPS website: www.hcahpsonline.org.
• Readmission rates. This includes patients readmitted to a hospital within 30 days of discharge from a previous hospital stay for heart attack, heart failure, or pneumonia. Readmission rates are reported for Medicare patients only.  Readmissions rates displayed on this site reflect 3 years of data. For more information, visit the CMS Hospital Compare Web site.
• Mortality rates. The rates take into account deaths within 30 days from all causes after an initial hospitalization with a principal diagnosis of heart attack, heart failure, or pneumonia. Mortality rates are reported for Medicare patients only.  Mortality rates displayed on this site reflect 3 years of data.  For more information, visit the CMS Hospital Compare Web site.
• Medicare reimbursement rates for patients whose primary diagnosis was heart attack, heart failure, pneumonia. The rates are reported at the hospital level and for Medicare patients only. Medicare payments for the same diagnosis-related group may vary. According to CMS, a hospital may get a higher payment for any or all of the following reasons:
  • It is classified as a teaching hospital
  • It treats a high percentage of low-income patients (disproportionate share)
  • It may treat unusually expensive cases (outlier payments)
  • It pays its employees more compared to the national average because the hospital is in a high-cost area (wage index). Note: The hospital's wage index is calculated using the hospital's payroll records, contracts and other wage related documentation.

  Standardized Infection Ratios: CLABSI DATA

  In addition, the site includes data on the incidence of central line-associated bloodstream infections (CLABSIs) , a type of infection introduced when a central-line catheter or tube is placed in a large vein in the neck, chest, or arm to enable the rapid administration of fluids, blood, or medications to critically ill patients.   CLABSIs can be prevented through proper insertion and care of the central line. 

  The data come from ten states - Colorado, Delaware, Maine, Missouri, New York, Rhode Island, South Carolina, Tennessee, Vermont and Virginia - in which hospitals report data as a result of state laws as well as from 34 other states in which some hospitals have voluntarily reported infection data to The Leapfrog Group (these states are AK, AL, AZ, CA, CT, DC, FL, GA, IA, ID, IL, KS, KY, LA, MA, MD, MN, MI, MS, NC, NE, NH, NJ, NM, NV, OH, OK, OR, PA, TX, UT, VT, WA, WI). In all, the site includes data from some 937 hospitals. The data were made available by special arrangement with Consumer Reports Health, who integrated data from each of the states and from the Leapfrog Group to produce the results displayed here. To locate hospitals reporting CLABSI data, search via hospitals' Location/Characteristics using the  "Hospitals Reporting…" filter, choosing the "Patient Safety" option. Or, follow the shortcuts to view performance data for hospitals in a particular state listed above. 

  The risk of infection varies across different types of intensive care units (ICUs).  Therefore, the CLABSI data are reported as standardized infection ratios (SIRs), a measure developed by the Centers for Disease Control and Prevention to summarize comparisons of data from an individual hospital's ICU to national infection rates for that particular type of ICU.

  This analysis adjusts for the fact that Leapfrog and the states have data from varying  ICU mixes, requiring comparisons to different average infection rates.  For instance, the average infection rate for cardiac ICUs nationwide is two per 1,000 central line days, so a particular cardiac ICU with a rate of three infections per 1,000 days has 50 percent more infections than average. For surgical ICUs, the national average rate is 2.3 infections per 1,000 central line days, so a surgical ICU reporting a rate of 4.6 infections per 1,000 central line days has 100 percent more infections than average.  The SIR pools these comparisons across all ICUs for which a hospital reports CLABSI data.

  Thus a SIR = 1 means that the hospital's ICUs produce CLABSIs at the same rate overall as would be predicted from national rates.  A SIR > 1 indicates the hospital had more infections than predicted from national rates, and a SIR < 1 implies it had fewer infections than predicted. So, for example, a hospital with a SIR of 1.50 reported 50% more infections than would be predicted from national rates, and one with a SIR of 0.70 reported 30% fewer infections than national rates for its mix of ICUs.

  We publish reported infections for all hospitals that meet either of the following sample size requirements:

  • At least 1,000 central line days (CLDs) OR
  • At least three infections, regardless of the number of central line days.  A hospital that reports three or more infections, even in fewer than 1,000 CLDs, could not achieve better than a rate of three infections/1,000 CLDs; this is 50 percent higher than the national infection rate for medical ICUs.

  In summary, the Standardized Infection Ratio is calculated as follows:

  1. For each reporting ICU, multiply the number of central line days by the published national infection rate for the ICU type, divided by 1,000, to estimate the number of infections predicted for that ICU if it were to report CLABSIs at the same frequency as the national rate (CLD x national rate / 1,000). 
  2. Within each hospital, calculate the sum of predicted infections and the sum of reported infections across all reporting ICUs.
  3. Calculate SIR = (total reported infections/total predicted infections).   For CLABSI data, a SIR of 1.0 indicates that a hospital is performing just as would be predicted from national rates.

  For more information on the methodology and data limitations, please see the Consumer Reports Health Web Site. 

  
  
  
   

  Data Not Available

  Note that not all hospitals report data for all measures; the site only publishes hospital data when there are four quarters' worth of data available for that particular measure.  Not all hospitals report all data for all measures.

  There is never data available in the following cases:

  • measures for which there are no state averages available (readmissions and mortality measures)
  • benchmark data (e.g., top 10%) for reimbursement rates
  • benchmark data (top 25%) for mortality rates and readmission rates
  • benchmark data for HCAHPS ratings of 7 or 8; HCAHPS rating 6 or lower (except for state and national average)
  • CLABSI measure for which there is no state or national average available

   

  Creating Summary Performance Measures

  In addition to reporting performance for each process-of-care measure, WhyNotTheBest.org presents summary performance scores for each hospital for each of the following four conditions:

  • Heart Attack Summary Rate (composite of seven process-of-care measures for this condition)
  • Heart Failure Summary Rate (composite of four process-of-care measures for this condition)
  • Pneumonia Summary Rate (composite of six process-of-care measures for this condition)
  • Surgical Care Improvement Summary Rate (composite of seven care processes used to prevent surgical infections)

  To create summary scores for each condition, the site uses a methodology prescribed by the Joint Commission. This approach suggests that the summary score be the number of times a hospital performed the appropriate action across all measures for that condition, divided by the number of opportunities the hospital had to provide appropriate care for that condition. Summary scores will not be displayed if all measures in that condition were less than 30 cases.

  Scores are not weighed, except that measures with larger denominators do contribute more weight to the calculation of the mean for that measure. None of the measures is risk adjusted.

  Creating an Overall Quality Measure

  In addition to creating performance scores for each clinical condition, we create an overall quality measure (Overall Recommended Care) that takes into account 24 Hospital Quality Alliance process-of-care measures (excluding three measures that are no longer being collected).

  It is possible for a hospital to submit “0” on a particular measure, indicating that it had no patients whose treatment was relevant to that measure. For example, a hospital reporting seven measures of heart attack care could submit denominators of 0, 6, 2, 12, 30, 29, and 14 and still receive an overall quality measure.

  In all instances we calculate a weighted average, wherein we add all the numerators and divide by the total of the denominators.

  We also rank hospitals on these overall measures, but to be ranked hospitals must satisfy the following additional criteria:

  • Measures must contain all available quarters worth of data.
  • Must report all indicators within the topics (i.e. fully reported). For example, HF composite score will be ranked only if all 4 HF indicators were reported.
  • At least one indicator in the topic had denominator >= 30.

  Some hospitals report enough data to be considered eligible for inclusion in the top performers listing. These criteria are detailed below.

  Identifying the Top Performers

  For each measure included on the site (excluding Reimbursement rates and CLABSI data), the site identifies the top 1 percent of performers—the “top performers.” It also includes other benchmarks: the top 10 percent and the top 25 percent, as well as top performers by hospital type (e.g., safety net, teaching, etc.).

  Top Performers on Process-of-Care Measures
  To appear among the top performers on the process-of-care measures, a hospital must have reported data for every available measure and recorded data on 30 or more patients for each of the four conditions (heart failure, heart attack, pneumonia, and surgical care improvement).

  No explicit weighting was incorporated, but higher-occurring cases give weight to that measure in the average. Since these are process measures (versus outcome measures), no risk adjustment was applied. 

  Top Performers on HCAHPS
  To appear among the top performers on the HCAHPS data, hospitals are ranked according to the percentage of survey respondents giving a 9 or 10 rating of overall hospital care. The site uses the results of the following question as a measure of patients’ overall experiences:

  “Using any number from 0 to 10, where 0 is the worst hospital possible and 10 is the best hospital possible, what number would you use to rate this hospital during your stay?”

  Note that the site does not apply exclusion criteria to create these performance rankings. All hospitals are included in the calculation of the percentile scores.

  Top Performers on Readmission and Mortality Rates
  The site identifies hospitals whose performance is statistically better than the national rate, as reported on Hospital Compare.

  Calculating Benchmarks

  For the purposes of calculating benchmarks, we identified hospitals in the following way:

  • Safety Net Hospitals: membership in the National Association of Public Hospitals and Health Systems and/or serving 30% or more Medicaid patients
  • Rural Hospitals: Hospitals designated by CMS as Critical Access Hospitals
  • Teaching Hospitals: membership in the Council of Teaching Hospitals and Health Systems
  • Academic Health Centers: membership in the American Academy of Medical Colleges
  • For-Profit: based on ownership, from the American Hospital Association database
  • Not-for-Profit/Private: based on ownership, from the American Hospital Association database
  • Government – Non-Federal: based on ownership, from the American Hospital Association database
  • Government – Federal: based on ownership, from the American Hospital Association database

  Top 1%, Top 10%, Top 25%: Top n% is the lowest score achieved by a hospital in the top n% (i.e. the minimum threshold to be in the top n%).

  Note: State average rates are not available for readmissions and mortality measures. 

  For further information about the methodology, please contact wntb@cmwf.org.