The centers for Medicare and

Transcription

1 MarketWatch Medicare s Policy Not To Pay For Treating Hospital-Acquired Conditions: The Impact The financial impact of the policy so far is small, but the public attention it has attracted may lead to improved quality. by Peter D. McNair, Harold S. Luft, and Andrew B. Bindman ABSTRACT: In 28 Medicare stopped reimbursing hospitals for treating eight avoidable hospital-acquired conditions. Using 26 California data, we modeled the financial impact of this policy on six such conditions. Hospital-acquired conditions were present in.11 percent of acute inpatient Medicare discharges; only 3 percent of these were affected by the policy. Payment reductions were negligible (.1 percent, or $.1 million equivalent to $1.1 million nationwide) and are unlikely to encourage providers to improve quality. Options to strengthen the incentives include further payment modifications for hospital-acquired conditions or expanding the hospital-acquired condition policy to exclude payment for consequences, additional procedures, and readmissions. [Health Aff (Millwood). 29;28(5): ; /hlthaff ] The centers for Medicare and Medicaid Services (CMS) pays for acute inpatient care for Medicare beneficiaries using the inpatient prospective payment system (PPS). 1 In 26 the inpatient PPS allocated $14 billion in payments for acute inpatient services about 2 percent of overall hospital revenues and 32 percent of Medicare spending nationwide. 2 Complications of care are estimated to account for up to 15 percent of inpatient costs, with approximately half of these considered preventable, yet Medicare payments have traditionally not been tied to quality of care. 3, 4 Hospital payments may actually increase if provider-caused complications result in new diagnoses triggering higher-cost diagnosisrelated group (DRG) classifications. The policy. It is difficult to justify continuing to pay providers for poor care, and Medicare has identified the opportunity to use funding policy to encourage higher quality and fewer unintended outcomes. 5 In 28 Medicare implemented its hospital-acquired conditions policy, 6 to penalize hospitals for poorquality care and encourage them to eliminate avoidable complications. The policy excludes from payment under the inpatient PPS eight avoidable complications that are considered hospital-acquired conditions: foreign object retained after surgery; air embolism; blood incompatibility; stages III and IV pressure ulcers; in-hospital falls and trauma; catheterassociated urinary tract infection (UTI); vas- Peter McNair (peter.mcnair@dhs.vic.gov.au) is a Harkness Fellow at the Philip R. Lee Institute for Health Policy Studies, University of California, San Francisco (UCSF). Harold Luft is director of the Palo Alto Medical Foundation Research Institute in California. Andrew Bindman is a professor of medicine, health policy, epidemiology, and biostatistics at UCSF. HEALTH AFFAIRS ~ Volume 28, Number DOI /hlthaff Project HOPE The People-to-People Health Foundation, Inc.

2 Health Tracking cular catheter associated infection; and surgical site infection mediastinitis (inflammation of the deep central chest tissues often due to bacterial infection) after coronary artery bypass graft (CABG). The codes defining an avoidable complication involve diagnoses, procedures, and condition present on admission (CPOA) that is, if the diagnosis was detected at admission. Hospital-acquired conditions are counted only if they arise within the same admission as the initial procedure. Other complications secondary to such a condition (for example, sepsis, or blood poisoning, that arises from catheter-associated UTI) are not considered for exclusions in the policy. This study used actual hospital discharge data from California to estimate the likely effect of the hospital-acquired condition policy on revenue for Medicare inpatient PPS funded acute inpatient discharges. It then proposes some steps that Medicare could take to provide additional financial incentives for hospitals to reduce complications in acute inpatient care. Setting cost weights and inpatient PPS payments. The implications of any wellintendedpolicydependonexactlyhowitis implemented. Two major steps in the inpatient PPS payment process involve (1) setting relative cost weights and (2) allocating payments based on the cost weights and adjustments to the hospital-specific price. Cost weights are set using an administrative data set with procedure and diagnosis codes and charges (from which hospital costs can be estimated using a cost-to-charge ratio). Diagnosis and procedure codes are used to form DRGs. Only the first nine diagnosis codes and the first six procedure codes from a discharge record are used by the CMS in determining the DRG. For each DRG, the average cost across all hospitals is calculated, then divided by the average cost across all DRGs to determine a relative weight. For example, the relative weight in 26 for DRG 55 Coronary bypass without cardiac catheterization, without major cardiovascular diagnosis is3.6151,aboutfourtimestherelative weight for DRG 167 Appendectomy without complicated principal diagnosis, without complications and co-morbidities (.8929). The average cost weight is then multiplied by a hospital-specific price to form the basis of a payment. The hospital-specific price reflects various factors, including regional wage rates, whether the hospital participates in teaching (indirect medical education, or IME), and the hospital s provision of care to a disproportionate share of low-income patients. Medicare furtheralterstheinpatientppsforspecific discharges where the cost of treating the patient exceeds the DRG-specific payment by a hospital-specific fixed amount and pays hospitals 8 percent of any additional costs above the fixed amount (outlier payments). Although not paying hospitals for potentially expensive avoidable complications appears to be an important disincentive for poorquality care, the degree to which the hospitalacquired condition policy places hospitals at financial risk has not previously been examined. The financial impact depends on (1) whether the deletion of a hospital-acquired condition code changes the DRG to which a discharge is allocated, and (2) how this changes the payment. This can be determined only by simulating the payment process. Study Data And Methods Data source. The University of California, San Francisco, Committee on Human Research, Institutional Review Board approved our use of the California Office of State Health Planning and Development (OSHPD) 26 PatientDischargeDataset(PDD)forthisstudy. 7 The PDD records, among other things, patient demographics, up to twenty-five diagnosis codes, and up to twenty-five procedure codes. These are the same data (albeit with more code fields) that Medicare uses in calculating hospital payments. The OSHPD applies several hundred audit rules to ensure the validity of the PDD. California is one of the few states routinely collecting CPOA information a field necessary for identifying Medicare s avoidable complications. Specific diagnosis codes defining two of Medicare s eight designated avoidable complications were introduced after 1486 September/October 29

3 26; the 26 PDD can provide only upper limits for the estimates of these two avoidable complications. Specifically, the 26 code covering vascular catheter associated infection also captures all instances of infection, sepsis, or septicemia associated with infusion, injection, transfusion, or vaccination as a complication of care; the 26 codes defining pressure ulcer stages III and IV capture all pressure ulcers as a complication of care, irrespective of their stage. We selected acute inpatient discharges where the Medicare inpatient PPS was the expected source of payment. Discharges without an associated charge were deleted, because their cost, and hence outlier payments, could not be estimated. The data were linked to the CMS Impact File, which includes hospitalspecific operating and capital cost-to-charge ratios, wage indices, and other factors necessary to estimate the cost and total Medicare payment for each discharge. 8 Payment estimates included IME, disproportionate-share hospital (DSH), remote hospitals, early transfer, and cost outlier calculations. 1, 9 Simulating the impact of the hospitalacquired condition policy. We simulated the impact of the policy by deleting the targeted complication codes from the discharge records and determining whether deletion affected the DRG classification of the hospitalization. If the DRG changed, we estimated the effect on hospital payments as the difference between payment under the initial DRG and the reallocated DRG, accounting for other aspects of inpatient PPS policy. Deleting complication codes changes DRG allocation only if the discharge was initially assigned to a DRG with complications and comorbidities and the deleted complication codeisthecriticaldifferencebetweenthetwo DRGs. 1 In cases involving multiple morbidities or complication codes, a hospital-acquired condition code is unlikely to be the critical code that triggers DRG (re)allocation. Hospitals commonly use software that legally reorders secondary diagnosis codes to maximize revenue. Following the introduction of Medicare s new policy, diagnosis codes reflecting avoidable complications are likely, whenever possible, to be reported beyond the ninth code, because they will not influence the DRG allocation. To simulate this, when an avoidable complication code was excluded, the remaining codes were restacked to ensure that, where possible, nine diagnosis codes and six procedure codes are available for DRG (re)allocation. Although the policy takes account of only those complications identified in the initial admission, we also explored selected acute Medicare inpatient PPS funded (re)admissions likely to be the direct consequence of a hospital-acquired condition. To assess the nationwide relevance of our study, we used the 25 National Inpatient Sample (NIS) to compare the prevalence of all hospital-acquired condition codes between Medicare-funded discharges in California and the thirty-six other states contributing to the 11, 12 NIS. Unlike the California data, the NIS data do not include the CPOA indicator necessary to define a hospital-acquired condition, so the comparison counted hospital-acquired condition codes regardless of CPOA status. 13 Study Findings The 26 PDD contains 86,147 acute care discharges for which Medicare was the expected source of payment. We excluded 9,675 records that lacked information on charges. 14 Medicare s hospital payment factors were matched to charges for 31 of 349 hospitals. 15 Medicare base and outlier payments were calculated for 95 percent of all discharges, for which the estimated total payments were $8.623 billion. This includes outlier payments for 14,74 high-cost discharges amounting to $334 million (3.87 percent of the total). Excluding avoidable complication codes. Of the 767,995 Medicare discharges wherethemedicarepaymentcouldbefully calculated, 828 (.11 percent) had codes meeting the definitions for at least one of the six definable hospital-acquired conditions (Exhibit 1). Payment for twenty-seven of these discharges would not be affected because the hospital-acquired condition was listed beyond the HEALTH AFFAIRS ~ Volume 28, Number

4 Health Tracking EXHIBIT 1 Impact Of The Medicare Hospital-Acquired Condition Policy, California Medicare Discharges, 26 Acute Medicare inpatient PPS funded discharges with an avoidable complication Acute Medicare inpatient PPS funded discharges where complication code exclusion alters the DRG assignment Avoidable complication Foreign objects retained after surgery a Air emboli Blood incompatibility Falls and traumas b Catheter-associated UTI c Mediastinitis after CABG d Subtotal Number Infusion-associated infection d Pressure ulcers all stages e 273 1,543 Total payment, thousands of dollars 1, ,12 3, ,778 4,45 69,947 Number Change in total payment, thousands of dollars Total 2,644 92, SOURCE: Office of Statewide Health Planning and Development Patient Discharge Dataset, 26. NOTES: PPS is prospective payment system. DRG is diagnosis-related group. UTI is urinary tract infection. CABG is coronary artery bypass graft. a Includes one discharge with concurrent stage III or IV decubitus ulcer. b Includes six discharges with concurrent stage III or IV decubitus ulcer. c Includes four discharges with concurrent stage III or IV decubitus ulcer. d Includes all cases of infection, sepsis, or septicemia associated with infusion, injection, transfusion, or vaccination as a complication of care. e Excludes cases with additional hospital-acquired conditions. first nine diagnoses used to determine the DRG; percent (93 of 828) of discharges were eligible for outlier payment. The estimated payment for the 828 discharges where a hospital-acquired condition was detected was $17.8 million; the estimated cost was $18.4 million. In-hospital falls and traumas. In-hospital falls and traumas along with catheter-associated UTI were the most commonly coded of the six definable hospital-acquired conditions. Ignoring the hospital-acquired condition diagnoses changed the DRG category in only 3.1 percent (26 of 828) of cases, resulting in a $92,17 (.1 percent) reduction in annual Medicare payments in California. The DRG category rarely changed, either because the discharge was already grouped to a simple DRG (rather than a DRG with complications and comorbidities ) or because the hospital-acquired condition was not the only condition or complication converting the discharge from a simple DRG to one with complications and comorbidities. There were no cases in which the DRG changed on the basis of excluding blood incompatibility, air embolism, or mediastinitis codes. Eleven percent (93 of 828) of the hospital-acquired condition discharges were eligible for outlier payments, which accounted for 12.7 percent of total payments for those ninety-three discharges. Pressure ulcers and infections from catheters. Stages III and IV pressure ulcers and vascular catheter associated infections are not accurately defined in the 26 PDD. When an estimate of the outer bound of these two hospital-acquired conditions was included in the estimate of the policy effect, the proportion of discharges eligible for the hospital-acquired condition policy was.34 percent (2,644 of 1488 September/October 29

5 767,995), with estimated costs of $12. million and payments of $88.4 million. The outerbound proportion of discharges affected by the policy was.7 percent (55 of 767,995), with a financial impact of.3 percent (Exhibit 1). More than a quarter of these discharges (668 of 2,644) incurred outlier payments, which accounted for 26.3 percent ($22.4 million of $88.4 million) of the total payments for those hospitalizations. Hospital-acquired condition associated readmissions. Of the 767,995 discharges, 73 included a code for mediastinitis in the absence of the CABG codes required by the policy. Etiology of the mediastinitis could be determined for 75 percent of these discharges. Seven discharges included codes for cardiothoracic procedures (for example, valve replacement) that involved sternotomy (an incision through the center of the chest for cardiac surgery); twenty-eight discharges included a postoperative infection (twenty-five) or septicemia code (three) and appeared to be readmissions that involved the treatment of cardiac surgery associated infective mediastinitis. An additional eighteen discharges were readmissions principally for the treatment of foreign objects retained after surgery, and two readmissions were principally for treatment of an air embolism as a complication of a medical procedure. Hospital-acquired conditions in the NIS data. The prevalence of these conditions among Medicare-funded hospitalizations in the NIS did not vary between California and all other states (Exhibit 2). Discussion The annual number of cases in California affected by the CMS s 28 hospital-acquired condition payment policy is likely to be small and the cost implications for hospitals minor. Using 26 hospital discharge data, we estimated that in California, the policy would reduce hospital payments by $92, $227,. If we extrapolate the California figures to the nation as a whole, this corresponds to $1.1 $2.7 million nationally. This is much lower than CMS estimates of anticipated cost savings. 6 Limitations of the study. This study was limited to data from California, representing only 8 percent of total Medicare acute inpatient PPS payments. The distribution of hospital-acquired conditions among Califor- EXHIBIT 2 Prevalence Of Hospital-Acquired Condition Codes In Medicare-Funded Discharges, 25 National Inpatient Sample, California And United States U.S. rate California rate ICD-9 diagnosis code Percent 95% CI Percent 95% CI All catheter-associated UTIs a Infusion-associated infection a Pressure ulcers all stages Foreign objects retained after surgery (2.41, 2.54) (.68,.75) (32.39, 32.8) (.9,.12) (2.19, 2.58) (.6,.82) (31.77, 32.99) (.8,.16) Air emboli Blood incompatibility Mediastinitis In-hospital falls and traumas (all) (.,.1) (.1,.1) (.1,.2) (67.1, 67.51) (.,.2) (.,.2) (.1,.5) (67.15, 68.37) SOURCE: California Office of Statewide Health Planning and Development Patient Discharge Dataset, 26. NOTES: Hospital-acquired condition codes are present in the first nine diagnosis codes, irrespective of condition present on admission flag. ICD-9 is International Classification of Diseases, Ninth Revision. CI is confidence interval. UTI is urinary tract infection. a Irrespective of organism. HEALTH AFFAIRS ~ Volume 28, Number

6 Health Tracking nia Medicare patients, however, was similar to that in other states. Our modeling of the financial impact of Medicare s hospital-acquired condition policy uses data collected before the policy was implemented.basedonotherpoliciesinwhich payments are linked to hospital coding, such as the introduction of DRGs for prospective payment, one might expect hospitals to change their coding practices in response to the policy. 17 If anything, however, we would expect that changes in coding would further diminish the policy s financial impact. The payment calculation does not include all factors contributing to the inpatient PPS (for example, new technology copayments, or a 2 percent reduction in payments for hospitals that fail to provide data on specified quality indicators), which could marginally alter payments. 1 Unlike the outlier payment provisions, these other factors merely adjust all payments to a hospital, instead of affecting specific cases. It is important to note that we merely sought to model Medicare policy and did not include payment for additional nonacute care required as a result of the complication (such as rehabilitation following in-hospital fracture). This underestimates the financial impact of preventing avoidable complications. Unintended effects. In response to the policy, hospitals may try to identify potential hospital-acquired conditions as being present on admission. This may involve thorough documentation of admission history and examination or additional tests. This study suggests that the revenue at risk does not warrant additional investment in testing. Of more concern, patients with comorbidities, and arguably at increased likelihood of hospital-acquired conditions, could experience barriers to care. This would also be unwarranted, because these conditions are unlikely to be critical diagnoses triggering DRG reallocation in patients with If reductions in payment draw attention to preventing hospitalacquired conditions, thecmsmaybewise to further strengthen the impact of the policy. serious comorbidities. Strengthening the impact of the policy. Despite (or perhaps because of) the limited financial impact of the hospitalacquired condition policy, the identified avoidable complications most of which are nursing care sensitive appear to have been accepted by hospitals as an appropriate target. The policy, however, is predicated on financial incentives raising the profile of hospitalacquired conditions and, at best, encouraging systematic measures that will minimize their likelihood. Such incentives must be large enough to attract the attention of both clinicians and hospital administrators. Reporting hospitals with high rates of these conditions could also influence changes in hospitals behavior. Chunliu Zhan and colleagues have demonstrated that admissions involving complications of care create losses for hospitals. 18 We estimate that in California Medicare, discharges in 26 involving a hospital-acquired condition were responsible for losses of $.7 $13.7 million that is, between seven and sixty times the payment modification arising from the policy. Although hospital-acquired conditions are not solely responsible for the losses in these cases, there are potential gains to be made in preventing these complications. If reductions in payment draw attention to preventing hospital-acquired conditions, the CMS may be wise to further strengthen the impact of the policy by the following means. Expanding the number of deleted codes per discharge. Increasing the number of complication codes targeted for deletion from discharges involving hospital-acquired conditions might increase the proportion of discharges where the policy affects payment. Such codes would include plausible outcomes arising from an identified hospital-acquired condition. For example, pulmonary embolism as a complication of care in the same discharge as in-hospital frac- 149 September/October 29

7 ture could be added to the list of diagnoses to be excluded. This would increase the likelihood that a case would be shifted to the simple DRG. Expanding the discharges affected by the policy. Paying only for conditions present on admission and using complication-averaged inpatient PPS payments are two much more robust financial incentives to encourage the prevention of hospital-acquired conditions. A 19, 2 more politically palatable approach might be for the CMS to iteratively expand the set of discharges affected by the policy. For example, mediastinitis qualifies as a hospital-acquired condition only if it is in the same episode as CABG. Nine discharges in our data involved mediastinitis and CABG. An additional seven discharges involved mediastinitis and a cardiac surgery procedure with sternotomy. If mediastinitis is preventable for CABG, then this hospital-acquired condition could be expanded to include all cardiac procedures involving sternotomy or, at least, to include mediastinitis in the same discharge as valve replacement. 21 Including readmissions associated with hospitalacquired conditions. The policy focuses only on complication codes occurring in the same discharge as the initial procedure. A patient with a hospital-acquired condition can be transferred or discharged, and the payments for any subsequent admission are not affected by the policy. In the case of mediastinitis, twelve discharges fell within the definition of hospitalacquired condition withoutanypaymentimpact on hospitals, yet an additional twenty-eight discharges had a postoperative infection code (25) or a septicemia code (3) and appeared to be readmissions for sternotomy-associated mediastinitis. For foreign objects retained after surgery, seventy-nine discharges were affected by the policy; eighteen more discharges were principally for the treatment of foreign objects retained after (previous) surgery. In the case of an air embolism, four discharges were potentially affected by the policy; two more discharges principally treated air embolism as a complication of a medical procedure. Nonpayment for such readmissions is problematic, because the hospital to which a person is readmitted might not be the hospital in which the hospital-acquired condition was sustained. One approach might be to pay for these readmissions with an equal sum deducted from the hospital in which the condition was sustained (at least up to the payment of the initial admission). For example, a hospital performing a CABG procedure resulting in a readmission to another hospital for mediastinitiscouldhaveitspaymentforthecabgreduced by the amount of the payment made to treat the mediastinitis. A less severe approach might be to reduce the initial (CABG) payment by a percentage of the readmission (mediastinitis) payment. Altering the payment modification. Components of the current payment method include the base payment, reflecting adjustments for IME, DSH, remote hospitals, early transfer, and cost outliers. IME, DSH, and remote hospital adjustments are inappropriate targets for payment modification, because they are not casespecific. The remaining inpatient PPS components, base and outlier payments, are appropriate policy targets. The current hospital-acquired condition policy focuses on modifying base payments through hospital-acquired condition code deletion and DRG (re)assignment, but, as shown, the reassignment policy has little overall impact; targeting outlier payments may be far more effective. The 26 policy pays 8 percent of outlier costs above a threshold loss of around $23,6. The CMS could withhold outlier payments for discharges involving hospital-acquired conditions. In our example, this would affect percent of total payments for hospital-acquired condition eligible discharges rather than the percent of payments withheld under the current policy. This policy option, however, creates much stronger incentives for hospitals not to include these codes among their listed diagnoses or to locate them beyond the ninth diagnosis code. The CMS could require (and enforce through audit) that avoidable complication codes be included in the first nine diagnosis codes. It is far easier to audit cases that are eligible for HEALTH AFFAIRS ~ Volume 28, Number

8 Health Tracking outlier payments than to audit all cases. Spillover to private insurers. Any strategies used by hospitals to reduce the incidence of avoidable complications (rather than just creative coding) in response to the CMS policy will probably benefit private health insurers through hospitalwide initiatives. The incentive for hospitals to reduce avoidable complications would be increased if private insurers also negotiated reduced payments for avoidable complications. Although this would seem to be an unlikely outcome, given that private insurers generally do not use DRG-based payments, the New York State Medicaid program and various private insurers have announced their intention not to pay for serious and preventable never events, many of which overlap with the policy. 22, 23 Press reports, however, do not make explicit the details of how the plans will be implemented. In itspresentform,the policy applied by the CMS to not pay for avoidable complications will have limited financial impact. The wide public attention it has attracted, however, may lead to some quality improvement efforts. The CMS has some options available to strengthen this policy to provide more substantial incentives to reduce complications of clinical care. Peter McNair was a Harkness Fellow supported by the Commonwealth Fund. The authors acknowledge the editorial assistance of Shane Reti, Peter Hockey, and Angelina Nicole. The opinions expressed are those of the authors and do not necessarily represent the views or the policy directions of the Commonwealth Fund. NOTES 1. Medicare Payment Advisory Committee. Hospital acute inpatient services payment system [Internet]. Washington (DC): MedPAC; 27 Oct [cited 29 Jun 24]. Available from: Payment_Basics_7_hospital.pdf 2. Ashby J, Stensland J. Hospitals: assessment of payment adequacy. MedPAC public meeting, 27 December 6 7; Washington, DC [Internet]. Washington (DC): MedPAC; 27 Dec 6 7 [cited 29 Jun 24]. Available from: 3. Ehsani JP, Jackson T, Duckett SJ. The incidence and cost of adverse events in Victorian hospitals Med J Aust. 26;184(11): Wilson RM, Runciman WB, Gibberd RW, HarrisonBT,NewbyL,HamiltonJD.TheQuality in Australian Health Care Study. Med J Aust. 1995; 163(9): Open letter. Paying for performance: should Medicare lead? Health Aff (Millwood). 23;22 (5): PartII,U.S.DepartmentofHealthandHuman Services, Centers for Medicare and Medicaid Services,42CFRParts412,413,415etal.,Medicare program; proposed changes to the hospital inpatient prospective payment systems for acute care hospitals and fiscal year 21 rates and to the long-term care hospital prospective payment system and rate, year 21 rates; proposed rule. Fed Regist. 29 May 22;74(98): In 29, after we had modeled the policy, Medicare added four hospital-acquired conditions: poor glycemic control, postoperative infection for orthopedic and bariatric surgery, and postoperative DVT. 7. California Office of Statewide Planning and Development. Healthcare Information Division Public Data Set [Internet]. Sacramento (CA): OSHPD; [cited 29 Jun 24]. Available from: DischargeData/PublicDataSet/index.html 8. DHHS. Impact file for IPPS FY 26 final rule [Internet]. Baltimore (MD): Centers for Medicare and Medicaid Services; [cited 29 Jun 24]. Available from: InpatientPPS/FFD/itemdetail.asp?filterType= none&filterbydid=&sortbydid=2&sortorder =descending&itemid=cms22536&intnumper Page=1 9. Medicare reduces payment for discharges that are transferred to another inpatient PPS acute hospital or, in 182 DRGs, transferred to another postacute care facility, and where the length-of September/October 29

9 stay for the discharge is less than the DRGspecific geometric mean length-of-stay. 1. For example, an adult patient with a diagnosis of 55.1 Unilateral inguinal hernia with obstruction, who undergoes a repair procedure, 53. Unilateral inguinal hernia repair, is allocated to DRG 162 Inguinal and femoral hernia procedures age 17 without complication nor comorbidity. If this case were complicated by a diagnosis of 561 Paralytic ileus, it is allocated to DRG 161 Inguinal and femoral hernia procedures age 17 with complication or comorbidity. However, if the case were complicated by 4939 Asthma without status asthmaticus, it is not reallocated (that is, paralytic ileus is a critical code that triggers reallocation; asthma without status asthmaticus alone would not). 11. Difference of two proportions (shrunken estimates) test (α=.5). 12. AltmanDG,MachinD,BryantTN,GardnerMJ. Statistics with confidence. 2d ed. London (U.K.): BMJ Books; The most recent version of the NIS available at time of publication was 25. Not all states contributing to the NIS collect condition present on admission (CPOA). 14. These discharges were predominantly undertaken by health maintenance organizations. 15. Although accounting for 13.8 percent of facilities, they had only 3.5 percent of Medicare patients. 16. In twenty-three cases, the hospital-acquired condition was catheter-associated UTI; in four, decubitus ulcer. 17. Simborg DW. DRG creep: a new hospitalacquired disease. N Engl J Med. 1981;34(26): Zhan C, Friedman B, Mosso A, Pronovost P. Medicare payment for selected adverse events: building the business case for investing in patient safety. Health Aff (Millwood). 26;25(5): ZhanC,ElixhauserA,FriedmanB,HouchensR, Chiang YP. Modifying DRG-PPS to include only diagnoses present on admission: financial implications and challenges. Med Care. 27;45(4): McNair P, Borovnicar D, Jackson T, Gillett S. Prospective payment to encourage system wide quality improvement. Med Care. 29;47(3): This may have to exclude replacement of valves damaged by Staphylococcus aureus infection. 22. Melnick GA, Fonkych K. Hospital pricing and the uninsured: do the uninsured pay higher prices? Health Aff (Millwood). 28;27(2):w New York State Department of Health. Medicaid to cease reimbursement to hospitals for never events and avoidable errors [Internet]. Albany (NY): Department of Health; 28 Jun 5 [cited 29 Jun 25]. Available from: medicaid_cease_paying_never_events.htm HEALTH AFFAIRS ~ Volume 28, Number