«DEPARTMENT OF HEALTH AND HUMAN SERVICES Centers for Medicare & Medicaid Services 42 CFR Parts 409, 424, and 484 [CMS-1560-F] RIN 0938-AP55 Medicare ...»
Comment: A commenter disagreed with our choice of data used for the creep analysis, saying that he was not convinced that data from the final year of IPS could serve as a base period from which to measure nominal growth in case-mix. The commenter questioned whether these data were representative of post-PPS, and noted that there was a learning curve with OASIS.
The commenter wrote that until we made the “derived base period” information available to the public, we should defer further creep adjustments and roll back the first two stages. He also questioned Abt’s use of just 313,447 IPS OASIS assessments, and was concerned that 18 percent of the episodes could not be evaluated since the OASIS could not be reliably linked to claims. He also noted that much has been made of improvements in OASIS coding over time, which suggests that the OASIS was not properly coded at the time of IPS. He questions the validity of this sample since many HHAs were not filing OASIS at the time,
be reliable bases for measuring creep. He also suggested we make public the data showing actual use of S2 and S3, and the IPS data used as a proxy for S2 and S3 cases. He noted that there was no M0825 data in OASIS for the final IPS period;
therefore one could argue that the final IPS data understates case-mix.
Response: We disagree that OASIS data collected during the last year of IPS were so poor as to be unusable to measure the case mix during that period. Agencies were not supposed to be unfamiliar with OASIS in the fall of 1999. Medicare first proposed making OASIS mandatory in March 1997. The development of OASIS had been supported and publicized by a large industry group over the years (transcript of June 24, 1997, meeting of National Committee on Vital and Health Statistics, accessed at http://www.ncvhs.hhs.gov/970624b1.htm#oasis). OASIS was discussed in professional and research journals (for example, see Home Healthcare Nurse, May 1997, Vol. 15/5: 340-342). OASIS version B-1 was released in October 1998, one year before our observation period for the IPS baseline began. After first publishing a final regulation in January 1999 whose effective date was delayed on April 27, 1999, Medicare re-finalized the
begin OASIS data collection for Medicare, Medicaid, and all other skilled services patients by July 19, 1999. This was 2.5 months before the beginning of our IPS baseline observation period, though they did not have to transmit data (other than for testing purposes) until August 25, 1999. The Health Care Financing Administration (HCFA), CMS’s predecessor agency, issued a comprehensive OASIS Implementation manual in July 1999 containing item-by-item instructions about how to complete the OASIS assessment. It was for the use of HHA agency staff who would be implementing OASIS as a uniform core data set. HCFA conducted a national meeting of state OASIS coordinators in midSeptember 1999 to train them in responding to agency requests for information. Four million assessments were submitted by HHAs to state agencies from July 1999 to January 2000 (CMS–3006– F, Dec. 23, 2005). This is an indication that agencies were actively working with OASIS from the start of the OASIS effective date. Our inability to match all simulated episodes to an OASIS stems mainly from the fact that time points of data collection for OASIS before HH PPS did not necessarily match the starting points of simulated episodes. During that period, OASIS was collected for outcomes purposes, not payment purposes.
nominal case-mix growth should be expected. However, we based our case-mix change adjustment on the evidence that patient health status did not change substantially, notwithstanding that improved understanding of and application of OASIS occurred.
Contrary to the commenter’s implication that the IPS sample was small, our sample size of hundreds of thousands is extremely large. Scientifically, sample size adequacy does not hinge on the ratio of the sample to the total population, but does depend on the actual absolute numbers of observations. Regarding the 18 percent of IPS episodes without a matched OASIS, we appreciate the commenter’s concern, but we have good reason to believe that the sample we used is representative. Based on our understanding of the main cause of the OASIS shortfall (described above), we do not have reason to infer a bias in the assessments that we do have. We also note that the sample’s average is consistent with an average from an initial episode sample. Initial episodes are more likely to have a matched OASIS (89 percent for initial episodes vs. 75 percent for subsequent episodes) so using data based on initial episodes should reduce concerns about sample representativeness. The estimate of average case-mix weight that we get from the sample
estimate we get from the initial episodes sample in the direction we expect (1.096 vs. 1.125). That is, the estimate from total (initial and subsequent) episodes is lower because health conditions measured in OASIS and used in the case-mix system tend to be more severe around the time of admission.
Furthermore -- and most important in terms of the basis for our policy decision to adjust payment to compensate for nominal case-mix change -- using an initial episode sample would produce the same percentage growth in case-mix as using a combined initial and subsequent episode sample. As we stated in the CY 2008 final rule (72 FR 49833): “We used all episodes rather than just initial episodes. This change in our sample selection approach does not materially change the estimate of case-mix change, whether comparing the baseline to HH PPS 2003 or HH PPS 2005.” Finally, modeling case-mix on an IPS sample that could possibly deviate in some respects from a fully representative sample would not necessarily produce distortions in the relationships found by the modeling procedure. Our conclusions about real case-mix change depend upon those relationships.
As we have noted elsewhere in our responses to comments, we believe we have made available highly detailed information about
2008 and August 2009) and in our regulations. For years, claims and OASIS data have been routinely available for purchase from CMS for researchers who wish to analyze it and can guarantee the security of the data. We published data on the rates of use of S2 and S3 under the IPS baseline period and 2003 in Tables 8 and 9 in the May 4, 2007 proposed rule(72 FR 25396-25399). The table in this section, in a response to a comment, provides detailed annual therapy visit distributions and thereby reflects S2 and S3 rates year by year. We did not use M0825 in determining S2 and S3; instead, we used the therapy visits reported by providers on the matched paid claims.
Comment: A commenter asked that we re-examine the case-mix weights for congestive heart failure (CHF), chronic obstructive pulmonary disease (COPD), and similar chronic conditions. She wrote that we claim HHAs are seeing fewer of such patients, and that she believes this is either due to coding practices or to agencies not accepting these patients. The commenter believes that the current method for accounting for patients with these conditions results in a very low case-mix weight. This low case-mix weight, coupled with high nursing needs, causes these patients to exceed available reimbursement, leading to a loss
these diagnoses, refine how shortness of breath is assessed and points calculated, and consider the speed at which such patients can perform Activities of Daily Living (ADLs), and not just whether the patient can do the ADL independently.
Response: The case-mix model we finalized in the CY 2008 final rule recognizes more diagnoses than the original (FY2000) HH PPS model, and it includes the specific diagnoses mentioned by the commenter, CHF and COPD. Also, the CY 2008 case-mix model recognizes resource-intensive interactions (that is, combinations of conditions within the same episode). The model specifically recognizes the interaction of pulmonary conditions and ambulation: the cost of serving pulmonary patients with a limitation in ambulation is more during an initial episode, and this combination increases the case-mix score. We believe this interaction case-mix item does capture the burden of COPD on ADLs. Shortness of breath, as measured by OASIS item M0490, provides additional points for initial episodes. Providers receive points for these and other conditions identified from statistical modeling of the relationship between diagnoses and OASIS measures on the one hand, and resource costs on the other.
Agencies also receive points for secondary diagnoses, thereby
Furthermore, we implemented a case-mix adjusted payment for non-routine supplies, such as those related to ulcers or wounds.
All of the point values in the case-mix model represent the average addition to the resource cost of the 60-day episode when a patient has the condition associated with the points. The fact that agencies may encounter some cases more costly than the case-mix-adjusted payment is a result of the variability in patient needs inherent in the population. We believe that, on average, this model aligns payment and agency costs with acceptable accuracy. As shown in Table 1 of the CY 2010 proposed rule (74 FR 40958), the proportion of episodes (initial episodes and all subsequent episodes) where the patient was discharged from the hospital prior to entering home health and had a hospital principal diagnosis of CHF has decreased by more than one-third since FY 2000. We did not publish a similar statistic for COPD. The statistics in Table 1 do not reflect coding practices in home health agencies; the conditions in Table 1 come from the hospital principal diagnosis preceding the episode (where the discharge occurred within the 14 days before the first day of the episode). As for refining the dyspnea and ADL measures in OASIS, we have reviewed all items in the course
a need for improvement was apparent. This review did not result in significant changes along the lines suggested by the commenter. Furthermore, it is unclear how the speed of ADL performance affects the resource costs for nursing care, beyond the added costs already accounted for in the point-bearing items mentioned earlier in this response. Finally, all changes to the OASIS instrument have to be balanced against the added burden imposed on the agency to measure performance reliably and accurately.
To summarize, we are moving forward with our existing policy, as implemented in the August 22, 2007 CY 2008 final rule with comment, of imposing a 2.75 percent reduction to the national standardized 60-day episode rates and the NRS conversion factor for CY 2010. We will continue to monitor any future changes in case-mix as more current data become available. We will also continue to look at ways to enhance the Abt model, and depending on the availability of newer and additional data, look to take into account factors that might yet be unmeasured in the current model. Given the continued growth in nominal case-mix, we expect to revise, upward the 2.71 percent reduction to the national standardized 60-day episode
rule. Analysis in next year’s rule will update the measure of the nominal increase in case-mix and compute the appropriate percent reduction to the national standardized 60-day episode rates and the NRS conversion factor to account for that increase.
C. Proposed CY 2010 Rate Update
1. The Home Health Market Basket Update We proposed a HH market basket update of 2.2 percent for CY
2010. This update was based on IHS Global Insight Inc.’s first quarter 2009 forecast, utilizing historical data through the fourth quarter 2008. Since publication of the proposed rule, we have a revised market basket update based on IHS Global Insight Inc.’s third quarter 2009 forecast, utilizing historical data through the second quarter of 2009. The final HH market basket update for CY 2010 is 2.0 percent. A detailed description of how we derive the HHA market basket is available in the CY 2008 Home Health PPS proposed rule (72 FR 25356, 25435).
Comment: One commenter stated the market basket increase of 2.2 percent would not be sufficient to cover the increased costs of implementing OASIS-C, CAHPS, as well as increases in staffing costs. The ongoing phase-in of the case-mix “creep”
market basket increase which is lower than the previous year’s
2.9 percent. According to MedPAC, 25 percent of HHAs have negative profit margins. The increase in costs of operation will have a negative impact on the financial viability of these agencies.
The commenter noted that not-for-profit HHAs are investing more of their revenue in attracting and retaining qualified HH staff. The shortages of nursing and physical therapy personnel are a major challenge. HHAs compete with other providers to attract these professionals.
Response: We disagree with the commenter that the 2010 market basket update is not sufficient. The home health (HH) market basket is not designed to account for changes in total costs (such as those associated with the implementation of OASIS-C or other initiatives), but rather it is intended to measure the input price pressures that the average home health provider is expected to face in the coming year. The composition of the market basket itself is made up of a set of mutually exclusive and exhaustive cost categories that reflect the cost structure of the industry (in a given base year). The HH index’s cost shares (or weights) are based on data reported
health agencies. Each cost category is assigned an appropriate price proxy whose projected movements are weighted by their respective cost shares resulting in the actual market basket update.