Methodology

Northwestern Memorial Hospital includes measures from a variety of national sources that meet the following criteria:

    • Patient-Centered: relevant to patient/consumer decision making and evaluation of care
    • Valid: using professionally accepted validation and reliability evaluation, measures are shown to be valid reflections of underlying quality and safety, are replicable and reliable
    • Transparent: methodologies are publicly available
    • Professionally Accepted: national and/or specialty leaders acknowledge that the measures are meaningful for evaluation and improvement of care
    • Benchmarked: with rare exceptions, we include only measures with comparative benchmarks

View our complete list of agencies and measures sources

Data

Our data gathering is comprehensive: 

  • The data in our Quality website is primarily from patients’ medical records.
  • Professional staff review every medical record and assign special codes to each one based on the physicians’ notes and other data describing the patient’s illnesses, complications, surgeries and other procedures  
  • For some quality measures, specially-trained nurses review the records to learn whether specific care processes were provided and whether complications arose
  • All data is subject to audit and review by numerous independent agencies, including:
    • Centers for Medicare and Medicaid Services
    • Joint Commission (Hospital Accreditation Agency)
    • Insurance companies
    • Others

We update measures at least once a year - more often when we are able to collect sufficient meaningful data.  We have included all the most relevant and useful measures adopted by leading independent agencies.  

Data Validation

Measures are drawn from two principal sources. Each is validated and audited.

  1. Patient medical records, abstracted by trained professionals according to current data element definitions: Many of the quality measures in this rating are based on careful review of patient medical records to abstract specific items of clinically relevant information. The abstractors are trained, and the information can be audited both by internal hospital staff and by external auditors, such as insurance companies, Centers for Medicare and Medicaid Services, and their agents. Examples of measures which use clinical abstraction include the CMS Core Measures, Society for Thoracic Surgeons, and many more.
  2. Information on patient bills: It is a routine process for each patient’s medical record to be reviewed by trained medical records coders who assign well-defined codes to the patient’s diagnoses and surgical and other procedures. These codes appear on the patient’s bill and enable the payor to make payment determinations. There are well-established industry guidelines for coding, and there are extensive audits. Coding data are readily available and inexpensive, since every bill is already coded and no additional work is needed to gather data. However, coding is generally viewed as less reliable information because it does not provide the level of detail and specificity that clinical abstraction does (see above). Examples of measures which use billing codes include the AHRQ quality and patient safety indicators.

Other sources of data in this Quality Rating include patient surveys (HCAHPS and Press Ganey); hospital self-assessment using externally-driven criteria (some Leapfrog measures); and other information such as average length of stay derived from hospital records.

Data Display

Decimals

Numbers are rounded using standard rounding conventions.  Additionally, for most measures, if the value is less than 1.0, round to 2 decimal places.  If the value is greater than 1.0 but less than 10, round to 1 decimal place.  If the value is greater than 10, show only whole numbers. For measures that show observed to expected ratio, odds ratio, standardized infection ratio, and hours per patient day, two decimal places are used.  

When the data comes from a public website, the number of decimal places matches those displayed on the public website.  

Y-axis

Data on charts is displayed using a consistent approach to the y-axis values. The goal is to display the data consistently and realistically.  The y-axis always ranges from zero to a maximum selected according to this logic:

  • If the highest data value on the chart is less than 1, the y-axis maximum is 1 and the increment is 0.5.
  • If the highest data value equals 1, the y-axis maximum is 5 and the increment is 1.
  • If the highest data value on the chart is greater than 1 but less than 30, the y-axis maximum is determined by doubling the highest value and round to the nearest 10.  The y-axis increment is one-fourth or one-fifth of the y-axis maximum.
  • If the highest data value on the chart is greater than 30 but less than 100, the y-axis maximum is 100.  The y-axis increment is 20.
  • If the highest data value on the chart is greater than 100,  the y-axis maximum is determined by doubling the highest value and round to the nearest 100.  The y-axis increment is one-fourth or one-fifth of the y-axis maximum.
  • If the highest data value on the chart is greater than 1000, the y-axis maximum is determined by doubling the highest value and round to the nearest 1000.  The y-axis increment is one-fourth or one-fifth of the y-axis maximum.

To learn more about public reporting of hospital performance measures, please see the release of "Guiding Principles for Public Reporting of Provider Performance" by the Association of American Medical Colleges.