May 14, 2002 Kenneth W. Kizer, M.D., M.P.H. President and CEO National Quality Forum 601 Thirteenth St, NW Suite 500 North Washington, DC 20005 Dear Dr. Kizer: On behalf of the American Health Quality Association (AHQA), a National Quality Forum (NQF) member representing the national network of Quality Improvement Organizations (QIOs, formerly known as PROs), thank you for this opportunity to comment on the NQF steering committee report, "Nursing Home Performance Measures." QIOs are the primary means by which the final NQF measures will be communicated and implemented in the nation’s long term care facilities. AHQA encourages the Forum to take action on the recommendations in this letter, as they come from the entities who are charged with the responsibility of persuading providers, practitioners, and the public to trust and rely upon the new quality indicators. Risk Adjustment. Validation of Risk Adjustment Methodology. In general, the risk adjustment methodology described in the steering committee report omits documentation of the validity of the risk adjustment methods employed. This raises a question as to whether the risk adjustment methods in the report have been validated. Recommendation 1: NQF should provide adequate evidence of validation prior to initiating a vote on a risk adjustment methodology for the patient outcomes measures in the draft report. Risk Adjustment and Outcomes Measures. Adequate risk adjustment for publicly reported core measures of patient "outcomes" are an essential element to accurately telling the story of care in a long-term care facility. We applaud the Steering Committee for recognizing the importance of providing a risk adjustment methodology. At the same time, we believe several measures in both the chronic care and post acute measure set require better risk adjustment. In a final report submitted to CMS in October 2001, titled Identification and Evaluation of Existing Quality Indicators that are Appropriate for Use in Long-Term Care Settings, the Abt Associates researchers concluded that "existing quality indicators were not adequately adjusted for differences in case mix and in assessment accuracy across facilities rendering them vulnerable to selection and ascertainment bias." The currently proposed measures do not yet address this shortcoming identified by Abt. Risk adjustment is important for several reasons: 1. Facility operators and health care professionals, who are being asked to work to improve care in the areas spotlighted by the measures, must have trust and confidence in their validity and fairness, or cynicism will result and improvement will be delayed. 2. Consumers and family members who are being urged to use the measures by federal officials must be able to rely on the measures to distinguish better from worse performers. 3. Facilities providing quality care to high-risk residents should be valued and their level of service extolled in our communities. At present, however, facilities that admit high-risk residents may appear to be poor performers in some of the quality measure data appearing on CMS’ website and in related news reports. Their reward for taking in more challenging residents may be a decline in consumer interest in their services. 4. In the absence of valid and reliable risk adjustment, publication of unadjusted quality measures will create a perverse incentive for providers to deny admission to the highest risk patients, in order to avoid adverse consequences in the marketplace that may follow from seemingly poor performance. Access barriers for high-risk residents would certainly be a very significant unintended effect of public reporting of these measures. It is difficult to imagine any outcome associated with even the most rudimentary risk adjustment methodology that would approach the risks of having no adjustment. NQF must take care to avoid inadvertently creating access barriers for these most vulnerable residents. One way to avoid the complexities of risk adjustment of outcomes measures is to invest in the development of well-constructed, evidence-based clinical quality process indicators. An example of the distinction is measuring the frequency of second and subsequent heart attacks (an outcome measure) or the percentage of heart attack patients for whom beta blocker drugs and aspirin are appropriate that actually receive them in a timely manner (a process indicator). This process measure is a scientifically well-founded predictor of good health outcomes (a 30% reduction in the risk of a subsequent heart attack), but offers several advantages over outcomes measures: 1. Even the best outcome indicator is at best a flag to trigger study and identification of the processes that produced that outcome, each of which must still be identified, evaluated, measured and changed. Process indicators measure the thing that must be changed, rather than an event far downstream in a complex set of caregiving processes. By focusing attention on the way the care is provided, process indicators simultaneously build understanding in the course of measurement. 2. A science-based clinical process indicator is less likely to lump quality problems together with factors and outcomes beyond the provider’s control. This is more fair and avoids inappropriate blame and punishment at the hands either of regulators or the marketplace. 3. The purpose of quality indicators, whether used by the public or by a QIO or a provider, is to identify problem areas for further analysis, and spur improvement in care processes that will benefit all patients. Clinical process indicators can be refined to focus in on the care of those patients for whom there is no controversy concerning the best diagnostic and treatment interventions. Good indicators show whether a set of ideal patients is not receiving clearly indicated care, because if the care is inadequate for ideal patients, the provider knows they are likely to have a problem with their clinical processes for all patients that deserves their attention. By initially focusing attention on ideal patients, clinical process indicators can help bypass long and unproductive arguments over whether this or that patient should have received the care in question, and stakeholders can rapidly concur in the need for active improvement efforts and get to work to bring it about. One drawback of clinical quality process indicators is that they require good scientific evidence that may be lacking in the long-term care field. For example, it may remain scientifically unproven which clinical practices are effective in accelerating the healing of pressure ulcers, or in preventing their occurrence. Under these circumstances, measuring the prevalence of bedsores at a given point in time is the best one can do until the science catches up with clinical practice. Recommendation 2: NQF, the Centers for Medicare and Medicaid Services (CMS), and private foundations should invest in identifying clinical areas where the science is sufficient to support a second generation of quality indicators that will focus in on specific clinical processes requiring scrutiny and improvement. The objective of such a project should be to produce clinical process indicators at least as sound as those successfully being used by QIOs to evaluate inpatient hospital care today. AHQA would be an enthusiastic supporter of such an effort, and will bring the expertise of its membership to resolve this task. In our discussion of individual measures requiring further work, below, we identify some problematic proposed outcome indicators that might be candidates for replacement by process indicators (if the scientific evidence is sufficient). Recommendations for Improved Specification of Individual Measures. Pressure Ulcer Prevalence. In measuring pressure ulcer prevalence, the measure set being used in pilot states excludes pressure ulcers identified in the admission assessment, but counts pressure ulcers which are present in the subsequent assessment. More severe pressure ulcers may not heal, particularly in certain high risk patients, in the time interval between admission and the next clinical assessment. Under this system, facility operators have an incentive to refuse admission to patients with severe pressure ulcers to avoid being inappropriately labeled as poor quality providers. The potential disadvantages for such patients far outweigh the practical disadvantages that may result from adjusting for the level of risk of such patients. This is an area that would benefit from a process measure that would evaluate whether proven practices to treat and heal bedsores have been employed by the nursing home. Weight Loss Prevalence. In the measures now being used in the pilot states, planned, desirable weight loss, such as the weight loss resulting from appropriate diuretic therapy to reduce the burden on the heart in congestive heart failure, is included in this publicly reported quality outcome measure. The measure should be refined by excluding heart failure patients from the prevalence measure, or if the scientific evidence is sufficient, by creating a process indicator that tracks whether the facility is doing what it should be to prevent unplanned weight loss. Improvement in Walking. In the currently proposed measure, residents with bilateral lower extremity amputation are not excluded from the reported value. If they were excluded, a facility would remain accountable for improving the ability of its residents to walk, but would not be scored as failing to improve walking ability in a person lacking legs. Inadequate Pain Management. The measure in use in the pilot states is labeled as though it reflects pain management, yet does not focus on residents’ pain after treatment. For example, if a resident reports "horrible" pain and is promptly and effectively treated, the measure records and reports this instance of reported pain to the public that the same as if it had been ignored. This is of course misleading. The title of the measure, "inadequate management of pain" is particularly inappropriate given the lack of specificity of this measure. In addition, a facility in which the staff is attuned to the problem of untreated pain may seek out and assiduously document resident reports of pain so they can initiate appropriate treatment. Under the current measure, such a facility will be publicly reported as inferior to a facility with an identical incidence of pain in which the staff is not as alert to this problem. A better measure would properly focus on problems requiring improvement, such as ongoing or unsuccessfully resolved pain. Recommendations for Additional Quality Measures. AHQA believes that NQF should add two important measures to the core measurement set. Immunization Measure. Many AHQA members are partners with the Centers for Disease Control and Prevention (CDC) in a multi-state project to measure and improve long-term care provider performance in reducing vaccine-preventable morbidity and mortality. Published reports and data available from these projects indicates that a measure can and will be effectively implemented in the field by QIOs working in collaboration with nursing facility staff. AHQA vigorously supports CDC’s comments calling for inclusion of immunization status measures in the core data set. Under the CDC’s proposed measure, facilities would be responsible for increasing the proportion of nursing home residents who are age 65 or older and are either: (1) screened upon admission and found ineligible for vaccination, or (2) screened upon admission, found eligible, and receive vaccine promptly following admission. Residents found "not eligible" would be those up-to-date for the vaccine according to the Advisory Committee on Immunization Practices (ACIP) recommendations, those who have medical contraindications for the vaccine, or those who refuses the vaccine. Pneumonia and influenza together comprise the 5th leading cause of death in persons older than 64. Invasive pneumococcal disease annually kills 6,000 to 7,000 persons. Data collected in the 1999 National Nursing Home Survey indicates that a majority of nursing homes has organized programs for delivering influenza, but many do not yet have programs to ensure appropriate use of pneumococcal polysaccharide vaccine (PPV). The reported 1999 average PPV rate for residents is 38%, less than half of the U.S. Public Health Service’s (USPHS) now-elapsed Healthy People 2000 goal of 80% vaccination of residents. The USPHS Healthy People 2010 goal for PPV is 90% for all persons older than 64 years, including those residing in nursing facilities. Recommendation 3: AHQA urges the NQF to include immunization indicators in the core measurement set. Fall Prevention Measure. Morbidity and mortality related to falls are among the most significant public health threats facing nursing homes residents. The MDS measurement set currently tracks the prevalence of falls, incidence of new fractures, accidents resulting in falls over 30 and 180 days, and accidents resulting in fractures in the last 180 days. We believe the MDS is well suited to developing a measure related to falls. QIOs have successfully implemented quality improvement programs to reduce the incidence of falls in long term care facilities, and could implement programs in response to NQF measures in this area. Recommendation 4: AHQA urges the NQF to include in the core measurement set one or more indicators of the effectiveness of providers’ fall prevention efforts. Please communicate to the Steering Committee our sincere appreciation for their dedication, hard work and good service to the NQF and to older Americans, and thank you for affording this opportunity to comment on the proposed nursing home quality measures. Please contact Dr. Mark Boesen or me at (202) 261-7571 with any questions regarding this letter. Sincerely, David G. Schulke Executive Vice President DGS:mdb