Comments on Draft Hospital Harm Measures

Download Letter

The Society of Hospital Medicine (SHM) Performance Measurement and Reporting Committee submitted comments to the Centers for Medicare and Medicaid Services (CMS) on draft hospital harm measures. The measures are meant to be electronically-specified clinical quality measures, meaning they would incorporate data abstracted directly from hospital’s electronic health records. As hospital safety is an important aspect of hospitalist practice, SHM believes hospitalists should be engaged in the development of these measures. Our comments focus on measures for hospital-acquired pressure injury, acute kidney injury, and opioid adverse events. These measures, if finalized, would be placed in federal reporting programs in the future and be used to assess the incidence of these events.

Full comments on each measure are below:

Hospital Harm – Hospital-Acquired Pressure Injury

The Society of Hospital Medicine (SHM), representing the nation’s hospitalists, welcomes the opportunity to offer comments on the Hospital Harm – Hospital-Acquired Pressure Injury (HAPI) measure. This measure, as designed, is meant to use electronic health record (EHR) data to assess a hospital’s proportion of hospitalized patients who develop a new stage 2-4 pressure injury, deep tissue injury or unstageable pressure injury, or have a worsening of an existing injury to stages 2-4.

Existing measures, such as Patient Safety Indicator 03 Pressure Ulcer Rate, assess a facility’s rate of pressure ulcers at stages 3-4. We request more information from the measure developers about why this new HAPI measure incorporates stage 2 pressure injuries, which has the potential to greatly increase both a facility’s rate of pressure injuries and the number of hospitals reporting the measure. The measure rationale and clinical recommendation statement do not clearly indicate a reason for this expansion.

Standardized data and staging assessments will be critical for the measure. We note that there can be some variability in how different clinicians may score a pressure injury and that this could have an impact on the reported rates of HAPI. The Centers for Medicare and Medicaid Services (CMS) should strive, both with this measure and through other efforts, to ensure clinicians are consistent in their reporting, to derive the most value and quality improvement from this measure.

Different environments within the hospital may have different rates of pressure injuries. The measure may therefore be more meaningful for clinicians if there are performance benchmarks for these intra-hospital settings. For example, a hospital system with a high proportion of intensive care unit (ICU) beds may see a higher rate of HAPI when compared against a system with a lower proportion of ICU beds. Separate benchmarks based on setting or type of patients (e.g., observation, general inpatient admits, post-operative, and ICU) could yield more actionable and meaningful feedback data.

CMS asks about the impact of this measure by differences in care practices for patients on hospice or receiving end-of-life care. We agree that patients receiving end of life care or hospice should be excluded from the measure, as the nature of their care is different from other patients. We do believe it would be possible to identify patients through EHR extraction, particularly for patients admitted under hospice. It is important for the measure to be thoughtful about patients who transition to comfort care during their hospitalization. For example, a Do Not Resuscitate (DNR) order is viewed as a proxy for comfort care in some facilities, but in others this is considered separate from an order for comfort measures, a palliative care admission, or inpatient hospice/contract hospice admission. EHR abstraction will need to capture these differences across facilities.

Given that this harm area is already assessed by other claims-based and chart-abstracted measures, we strongly urge the Centers for Medicare and Medicaid Services (CMS) to avoid double counting events in quality assessment and pay-for-performance programs. We agree with the trajectory of moving towards EHR-supported measures where possible, but caution that other measures assessing a similar cohort should be removed.

Finally, CMS needs to consider the effects of case-mix on performance under this measure and to prioritize effective risk adjustment. Risk adjustment methodologies should ensure that hospitals that see sicker patients and are more likely to see a larger rate of pressure ulcers do not face penalties for the makeup of their patient population.

Hospital Harm – Opioid-Related Adverse Events

The Society of Hospital Medicine (SHM), representing the nation’s hospitalists, welcomes the opportunity to comment on the Hospital Harm – Opioid-Related Adverse Events measure. The measure, as designed, is the rate of hospitalized patients who receive an opiate antagonist (naloxone), outside of the operating room. In the first 24 hours of the hospitalization, an opioid must have been administered prior to receiving naloxone to be considered a harm. We agree with the Centers for Medicare and Medicaid Services (CMS) that hospital-administered opioid adverse events are an important area for attention, particularly given the broad concerns with opioid use today.

We strongly believe this measure needs to target hospital harms during hospital stays and are concerned the current specifications may be overbroad. There are instances, for example, where hospitalized patients surreptitiously obtain and use opioids, resulting in overdose and requiring narcotic antagonist administration. CMS seems to recognize community-based opioid use in patients receiving naloxone in the first 24 hours of hospitalization by requiring evidence of the hospital administering opioids. This requirement for evidence of administration of opioids by the hospital could be expanded. Excluding patients suspected or confirmed to have self-administered street drugs or prescription opioids obtained outside of the control of the care team would sharpen the focus of the measure onto preventable hospital harms.

We also have concerns over the wide variation in prescribing rates across hospitals. The discrepancy between prescribing rates of 5% to 72% cited in the measure rationale portends challenges in comparing hospitals fairly. Some of these concerns may be addressed by risk adjustment based on the patient population, but without more information, we are concerned that this measure may not accurately account for differences between facilities with high and low volumes of opioid use. 

Hospital Harm – Acute Kidney Injury

The Society of Hospital Medicine (SHM), representing the nation’s hospitalists, welcomes the opportunity to comment on the Hospital Harm – Acute Kidney Injury (AKI) measure. The measure, as designed, would measure the proportion of patients who suffer the harm of a substantial increase in serum creatinine or the initiation of renal dialysis.

While we concur with CMS that hospital-acquired AKI should be prevented and avoided where possible, we do have concerns with this measure construction and its ability to assess preventable instances of harm. For example, contrast CT scans with intravenous contrast as well as angiograms may be necessary during a hospitalization, but may be associated with renal failure. In addition, it may be difficult to differentiate preventable hospital-acquired AKI from AKI driven by the primary disease process, particularly in cases of sepsis and shock. Most of the relevant interventions (fluids, vasopressors and protocol-based hemodynamic management) to prevent AKI would be captured under sepsis quality measures. In effect, treating the underlying precipitating cause of the AKI (in this case, sepsis) is the only potential prevention for the AKI.

We also have concerns with the numerator of the measure being too wide to capture only preventable hospital harm events. We believe the measure should have a rolling window of 24-48 hours to ensure that illnesses present or developing on admission are not inadvertently coded as hospital-acquired.

The measure could have an additional exclusion for patients who are hospitalized for fewer than 2 days, as they will not have an established trend of stability that would distinguish hospital-acquired AKI from AKI that was developing or present on admission. This could be done by excluding patients who have not demonstrated stable renal function on several consecutive measurements prior to decline in renal function.

Given that this harm area is already assessed by other claims-based and chart-abstracted measures, we strongly urge the Centers for Medicare and Medicaid Services (CMS) to avoid double counting events in quality assessment and pay-for-performance programs. We agree with the trajectory of moving towards EHR-supported measures where possible, but caution that other measures assessing a similar cohort should be removed.