Skip to main content

Measure Specification

Define the Denominator

The denominator statement describes the population evaluated by the individual measure. The population defined by the denominator can be the same as the target/initial population, or it can be a subset of the target/initial population to further constrain the population for the measure. The measure developer must describe the denominator statement sufficiently so the reader understands the eligible population or composition of the denominator, and should not use codes in lieu of words to express concepts in written descriptions. The measure developer should define the denominator precisely and include parameters such as

  • Age ranges
  • Setting
  • Diagnosis
  • Procedures
  • Time interval
  • Other qualifying events, e.g., look-back period

Format: Patients, age [age or age range], with [condition] in [setting] during [time frame]

Examples

  • All patients aged 18 and older with a diagnosis of Chronic Obstructive Pulmonary Disease (COPD) (CMIT Family ID 00131) (CMS CBE #0091)
  • All patients at least 18 years old as of the first day of the reporting month who are determined to be maintenance hemodialysis (HD) patients (in-center and home HD) for the complete reporting month at the same facility (CMIT Family ID 00313) (CMS CBE #2978)

Denominator Exclusion 

Denominator exclusions refer to criteria that result in removal of patients or cases from the denominator before calculating the numerator. A denominator exclusion means the numerator event is not applicable to those covered by the denominator exclusion; an example of a denominator exclusion is to “exclude women who have had bilateral mastectomy from the denominator for a measure of screening mammography.”

The goal of denominator exclusion criteria is to have a population or sample, all of whom share a similar profile in terms of their likelihood of meeting the numerator criteria. 

Format: denominator-eligible patients who [have some additional characteristic, condition, procedure]

The measure developer must not specify systematically missing data as a denominator exclusion. The CMS consensus-based entity (CBE) Consensus Standards Approval Committee Guidance on Quality Performance Measure Construction notes systematic missing data (e.g., when poor performance is selectively not reported) reduce validity of conclusions that can be made about quality.

To avoid complexity, the measure developer should limit denominator exclusions to just those that are absolutely necessary. Ensure the measure logic expressions (e.g., Clinical Quality Language) clearly identify denominator exclusions. The measure developer should support an allowable denominator exclusion with

  • Evidence the denominator exclusion condition occurs with such frequency that it will distort the measure results without the denominator exclusion, and
  • Evidence the denominator exclusion significantly improves measure validity, and/or
  • Evidence of both empiric and face validity.

Also consider

  • Conditions that are present on admission should not count as an outcome of interest.
  • An outcome of interest can be very difficult to prevent in a population of interest, and therefore not an indication of substandard care.
  • Some inclusion criteria identify populations who are at very low risk for the outcome of interest, but then the measure developer incorrectly made a denominator exclusion to prevent dilution of the quality improvement denominator.
  • Some inclusion criteria are for the purpose of enhancing face validity with clinicians.
  • Some inclusion criteria are an inherent part of the quality improvement definition.
  • The inclusion criteria may conflict with the patient’s goals of care (e.g., advanced illness, terminally ill). 

Determine Need for Denominator Exception

A denominator exception allows the measured entity to get credit when the measured entity performs the quality action, but not penalized if not done for an appropriate reason. The exception allows the exercise of clinical judgment and implies the measured entity considered or offered treatment to each potentially eligible patient in the denominator. Denominator exceptions are most appropriate when measures of contraindications to drugs or procedures are relative (Spertus et al., 2010). The measure developer should only use a denominator exception in proportion measures. It is not appropriate for ratio or continuous variable measures.

Although no single agreed-upon approach to denominator exceptions exists, there seems to be consensus that denominator exceptions provide valuable information for clinical decision-making. Measure developers who build denominator exceptions into measure logic should be aware that—once implemented— denominator exception rates may be subject to reporting, auditing, endorsement/maintenance review, and validation of appropriateness. The measure developer should account for these factors in measure design and development. 

Example

An example of a denominator exception allowing for clinical judgment in the case of a patient with two chronic conditions

Asthma is an allowable denominator exception for the quality measure of the use of beta blockers for patients with heart failure. Thus, physician judgment may determine there is greater benefit for the patient to receive beta blockers for heart failure than the risk of a problem occurring due to the patient’s coexisting condition of asthma. If the measured entity gives the medication, the measure implementer does not search for denominator exceptions, and the patient remains in the denominator. If the measured entity did not give the medication, the implementer looks for relevant denominator exceptions and removes the patient−in this example, a patient with asthma−from the denominator. If the measured entity did not give the medication and the patient does not have any denominator exceptions, the patient remains in the denominator and the measured entity fails the measure.

Reasons for Denominator Exceptions

A measure developer should specifically define a denominator exception when capturing the information in a structured manner fitting the clinical workflow. 

Reasons for Denominator Exceptions Description Example
Medical Reasons Medical reasons should be precisely defined and evidence-based. The events excepted should occur often enough to distort measure results if not accounted for. A broadly defined medical reason such as “any reason documented by physician” may create an uneven comparison if some physicians have reasons that may not be evidence-based.  Medication specified in the numerator is known to cause harm to fetuses, and there is documentation the patient’s pregnancy is the reason for not prescribing an indicated medication. If in the course of a measure’s use, the measure developer finds that medical reasons resulting in a denominator exception occur in a high enough volume and are of universal applicability, then the measure developer can consider the denominator exception for redefinition as a denominator exclusion. 
Patient Reasons Patient reasons for not receiving the service specified may be a denominator exception to allow for patient preferences.  The patient has a religious conviction that precludes the patient from receiving the specific treatment, the physician explained the benefits of the treatment, and documented the patient’s refusal in the record.
System Reasons System reasons are generally rare. The measure developer should limit these to identifiable situations that are known to occur. A vaccine shortage prevented administration of the vaccine.
 

The measure developer must capture the denominator exception with explicitly defined data elements that allow analysis of the denominator exception across measured entities to identify patterns of inappropriate denominator exception and gaming and to detect potential healthcare disparity issues. Analysis of rates without attention to denominator exception information has the potential to mask disparities in healthcare and differences in measured entity performance.

Examples

  • Inappropriate denominator exception: a notation in the patient’s medical record indicates a reason for not performing the specified care, and scientific evidence does not support the reason.
  • Gaming: patient refusal may be a denominator exception; however, it has the potential for overuse. For example, a measured entity does not actively encourage the service, explain its advantages, or attempt to persuade the patient, and then uses patient refusal as the reason for nonperformance.
  • Disparity issues: the use of a patient reason for denominator exception for mammograms is found to be high for a minority population, which may indicate a need for more targeted, culturally appropriate patient education or closer examination of patient access issues such as lack of transportation or lack of childcare.

Transparency and Evidence

To ensure transparency, the measure developer should capture an allowable denominator exception in a way the measured entity can report it separately, in addition to the overall measure rate. The measure developer should support an allowable denominator exception with evidence

  • Of sufficient frequency of occurrence such that distortion of the measure results occurs without the denominator exception
  • The denominator exception is clinically appropriate to the eligible population for the measure

Although no single agreed-upon approach to denominator exceptions exists, there seems to be consensus that denominator exceptions provide valuable information for clinical decision-making. Measure developers who build denominator exceptions into measure logic should take caution that -once implemented- denominator exception rates may be subject to reporting, auditing, endorsement/maintenance review, and validation of appropriateness. The measure developer should account for these factors in measure design and development.

Last Updated: Oct 2022