Missing and Invalid Data

Introduction

Missing data refers to data elements, required for calculating a national hospital quality measure, that have no values present for one or more episodes of care (EOC) or event records. Invalid data refers to data element values, required for calculating a national hospital quality measure, that fall outside of the range of allowable values defined by The Joint Commission for that data element.

Reducing missing and invalid data minimizes the bias to a measure rate, because episodes of care with missing or invalid data cannot be included in the calculation of the observed measure rate. A measure's observed rate may not accurately reflect the patient population, if the excluded EOC and event records differ significantly from the EOCs and events with no missing data that were included in the measure calculation.

Data Collection and the Unable to be Determined (UTD) Allowable Value

Abstractors must touch and provide an answer to every data element that is applicable per the combined skip logic of all of the measures in a topic. While there is an expectation that all data elements are collected, it is recognized that in certain situations information may not be available (dates, times, codes, etc.). If, after due diligence, the abstractor determines that a value is not documented or is not able to determine the answer value, the abstractor must select “Unable to Determine (UTD)” as the answer. The “UTD” allowable value is used as follows:

• Admission Date, Birthdate, Discharge Date, Event Date, Event Type, ICD-10-CM Principal and Other Diagnosis Codes, ICD-10-PCS Principal and Other Procedure Codes, Psychiatric Care Setting for inpatient measure sets; Outpatient Encounter Date and E/M Code for outpatient measure sets; and, Psychiatric Inpatient Days-Medicare Only, Psychiatric Inpatient Days-Non-Medicare Only, Total Leave Days-Medicare Only, and Total Leave Days-Non-Medicare Only for the HBIPS measures do not have an “UTD” allowable value. Therefore, a case with UTD for these elements would require re-evaluation to ensure the correct value is entered, if available.
• Date, time, and numeric data elements, other than those listed above have an “UTD” allowable value option.
  • Rate-based proportion algorithms evaluate EOC records to a Measure Category Assignment = “D” or "E" (failed) depending on the desired direction of improvement of the associated measure when a date, time, or numeric data element containing an allowable value of “UTD” is evaluated. When the direction of the improvement is an increase in rate, the algorithm will evaluate the EOC records to a Measure Category Assignment = "D". When the direction of improvement is a decrease in rate, the algorithm will evaluate the EOC record to a Measure Category Assignment = "E".
  • Continuous variable and rate-based ratio algorithms evaluate EOC records to a Measure Category Assignment = “Y” (UTD value exists) when a date, time, or numeric data element containing an allowable value of “UTD” is evaluated.
  • The method by which data collection software collects “UTD" information is determined by the software; except the software cannot automatically default an “UTD” answer. The decision to enter an “UTD” for each data element is up to the abstractor, not the software.
  • Yes/No data elements: The allowable value “No” incorporates “UTD” into the definition. Refer to the measure algorithms in which each Yes/No data element is used to determine how the EOC and event records are treated.
• Data elements containing two or more categorical values: The “UTD” value is either classified as a separate allowable value or included in the same category as “None of the above/Not documented”. Refer to the measure algorithms in which each categorical data element is used to determine how the EOC record is treated.

Missing and Invalid Episode of Care (EOC) and Event Data

Missing and invalid data must be corrected and reprocessed in order for it to be processed correctly against the measure algorithm for expected outcome. This is important as data aggregation uses the measure outcome for each patient included in the measure’s sample or 100% of patients if the measure is not sampled.

• The measures have been structured with the precondition that the majority of general data elements which are missing data cause the EOC and event records to be rejected before processing against the measure. These data elements for Discharge measure include but are not limited to Admission Date, Birthdate, Discharge Date, and ICD-10-CM Principal Diagnosis Codes. For Event measures such general data elements include but are not limited to event-type, event-date, Admission Date, and Birthdate. For outpatient measures such general data elements include but are not limited to Outpatient Encounter Date, E/M Code, ICD-10-CM Principal Diagnosis Codes, Birthdate and Arrival Time. Refer to the Introduction to the Data Dictionary in this manual for the complete list of general data elements.
  • Not all patients have an ICD-10-CM Other Diagnosis Code or an ICD-10-PCS Principal and Other Procedure Codes. Records will be accepted with missing data for these data elements.
• Measure-specific data elements that are missing data will cause the EOC and event records to be rejected if any measure algorithm results in a Measure Category Assignment = “X” (missing data). If no measure evaluates to a category assignment of “X”, the EOC record will be accepted.
• General and measure specific data elements that contain invalid data will cause the EOC and events record to be rejected.

Abstraction Software Skip Logic and Missing Data

Skip logic allows hospitals to minimize abstraction burden by using the software edit logic to bypass abstraction of data elements not utilized in the measure algorithm. However, these bypassed elements also negatively impact data quality when elements are incorrectly abstracted and subsequent data elements are bypassed and left blank.

The use of skip logic by hospitals is optional and not required by The Joint Commission. Hospitals should be aware the potential impact of skip logic on data quality and abstraction burden. Hospitals utilizing skip logic should closely monitor the accuracy rate of abstracted data elements, particularly data elements placed higher in the algorithm flow.

Note:
*A missing value occurs when the abstractor does not select an answer for a data element (leaves it blank) or the software incorrectly saves or processes the value as a “null” instead of the correct value for a data element. An “UTD” allowable value is not considered missing data.

Population and Sampling:

For more information concerning aggregate data required for accreditation reporting, including population and sampling information, refer to documentation available on the Joint Commission’s Direct Data Submission (DDS) Platform. This documentation includes details on the specific aggregate data required for each chart-based measure and information concerning how to calculate the data.

For certification purposes, hospitals do not submit their initial patient population or sampling data.

Missing, Invalid, UTD Data Summary:

Missing Data	Invalid Data	UTD
No data element value is present. (blank or “null”)	The data element value falls outside of the range of defined allowable values.	The allowable value of Unable To Determine, “UTD” is present for the data element.