Calculating Quality Measures

• Overview
• What Are Quality Measures?
• Interpreting Your Data
• Quality Measure Technical Specifications and Data Algorithms
• References and Resources

Overview

This section includes information on quality measures, technical specifications, and explanations of how to calculate and interpret quality measure scores.

What Are Quality Measures?

Quality measures assign quantities to attributes of care and can be used to drive performance improvement.  Repeating the same quality measures over time and comparing the scores can identify problems and be used to monitor the impact of quality improvement efforts.  It is difficult to interpret the result of a quality measure as good or poor unless there is a standard of comparison. Well established benchmarks for hospice and palliative care quality do not exist currently, but are in development.  However, quality measure results can be meaningful if compared to goals set by an agency or a national organization.

Quality measure data obtained prior to initiating quality improvement activities can establish baselines for future comparisons.  Three months of data are often sufficient to establish a baseline; however you may find that you need more or less time depending on the number of clinical records abstracted each month.

The AIM Quality Measures include both process and outcome measures.  A process measure looks at whether or not something was done, such as whether pain or nausea was assessed during the initial visit.  Outcome measures focus on the result of the processes.  For example, if a patient has dyspnea at rest at the initial assessment, did the dyspnea improve within twenty-four hours?

The AIM Quality Measures are calculated on data that are collected retrospectively from events that have already occurred.  Thus, the accuracy of the data and the ability to paint a complete picture are both grounded in how the data were first documented.

Most of the quality measures in this toolkit are calculated as proportions of patients who met the criteria definition for the measure.  All of the quality measures are reported over specific periods of time, such as months or quarters, so that you can monitor performance over time.  Examining trends can be a useful way of assessing if an intervention you have put into action is working or whether performance is being affected by an external factor, such as a major organizational change.

Once the AIM Process is fully implemented, the AIM Quality Measures should provide an accurate picture of how well your agency is performing in certain areas and should help you identify where you may need to improve performance.

Return to top

Interpreting Your Data

It is understandable that when faced with data that show underperformance, you may have difficulty accepting the results.  However, be assured you are not alone!  The Institute for Healthcare Improvement describes four stages of acceptance that many people experience when first evaluating their own performance.  For some measures you might start out at Stage 4 and for others at Stage 1.

First Stage:      "The data are wrong."
Second Stage:"The data are right, but it is not a problem."
Third Stage:     "The data are right. There is a problem, but it is not my problem."
Fourth Stage:     "The data are right. There is a problem, and it is my problem."

Stages of Data Acceptance
When Your Score Isn't What You Expected

However, it is also important to realize that data are not always correct, and careful consideration should be given to any results.  The following questions may be used as prompts to assessing the accuracy of your data:

• Are the data reflective of the care you provide?
• Do the data identify what caused the less than optimal performance?
• Are the data directly related to the ultimate goal of the care you provide?
• What does the staff think of the data?

Making Sense of the Data
Understanding Proportions, Rates, and Patient-Days (or Headaches from the Fourth Grade)
Figuring Numerators and Denominators

Quality Measure Technical Specifications and Data Algorithms

The measures recommended as part of this Toolkit are calculated as either proportions or rates per patient-days.  Many of the quality measures are composite measures, which means the measures are comprised of multiple sequential criteria that must be met in order to receive "credit".  By "drilling down" or looking at the individual components of the measure, agencies may find that they can better design their quality improvement activities to target the areas where they have the lowest performance.  Because the quality measure scores are automatically calculated for 9 of the twelve measures, you may need to access your "raw" data to perform these additional calculations.  The data algorithms are provided to help describe the components that comprise each measure.

Technical Specifications and Data Algorithms: Measure Population Inclusion Criteria
Measures:

• M1: Percent of patients who are assessed for physical symptoms and screened for psychological symptoms during or on the same day as the admission visit.
• M2: Percent of patients with comprehensive assessment completed within 5 days of admission.
• M3: For patients who assessed positive for pain, the percent whose pain was at a rating of none or mild at the second pain assessment.
• M4: For patients who assessed positive for dyspnea, the percent of patients who improved within 1 day of assessment.
• M5: For the patients who assessed positive for nausea, the percent who received treatment within 1 day of assessment.
• M6: Percent of patients on opioids who have a bowel regimen initiated within 1 day of opioid initiation.
• M7: For patients who screened positive for anxiety, the percent who receive treatment within two weeks of screening.
• M8: Percent of patients who had moderate to severe pain on a standard rating scale at any time in the last week of life.
• M9: Percent of patients with documentation in the clinical record of an advance directive or discussion that there is no advance directive.
• M10: Percent of families reporting the hospice attended to family needs for information about medication, treatment and symptoms.
• M11: Number of adverse events per 1000 patient-days: falls.
• M12: Number of adverse events per 1000 patient-days: medication errors.

Accessing Your Raw Data from the AIM Excel Data Entry Collection Tool: Excel 2003
Accessing Your Raw Data from the AIM Excel Data Entry Collection Tool: Excel 2007

References and Resources

Return to top

IPRO Staff Access