The US Medicare policy of not reimbursing hospital-acquired conditions: what impact would such a policy have in Victorian hospitals?

Activity-based inpatient funding systems pay for the amount of treatment provided to patients rather than for the expected outcomes of clinical treatment for a given condition. These systems include payments for preventable complications of care, which funders are increasingly looking to avoid.

The United States Medicare system, which reimburses hospitals for inpatient care using a diagnosis-related group (DRG)-based casemix payment system, has recently ceased paying for eight specific hospital-acquired conditions (HACs) that are considered preventable. The US Centers for Medicare and Medicaid Services (CMS), having identified funding policy as a way to encourage higher quality care and fewer unintended clinical outcomes,1 implemented a non-payment for HACs policy in 2008.2 The policy attempts to penalise hospitals for poor-quality care and to encourage them to implement strategies for eliminating avoidable complications. The eight avoidable complications excluded from casemix payments are: foreign object retained after surgery; air embolism; blood incompatibility; pressure ulcer stage III or IV; in-hospital falls and trauma; catheter-associated urinary tract infections; vascular catheter-associated infections; and surgical site infection — mediastinitis after coronary artery bypass graft. (Since the drafting of this article, the CMS have added four HACs to the list: poor glucose control, infection following specific orthopaedic procedures, infection following specific bariatric procedures, and deep vein thrombosis following specific orthopaedic procedures.) The codes that define the HACs take into account International classification of diseases — clinical modification (9th revision) (ICD-9-CM) diagnosis and procedure codes as well as condition-present-on-admission indicators. The policy removes diagnosis codes from the payment calculation process, in the expectation that this will result in a lower payment.

The HACs policy has attracted much media interest in the US.3 Although a recent study has shown that this policy will have an impact of as little as 0.001% on US Medicare’s $106 billion budget for acute hospital funding,4 it is of interest because it may be the first instance of a jurisdiction refusing to pay for complications of care.

In the state of Victoria, complications of care are estimated to account for up to 15% of inpatient costs.5 The landmark “Quality in Australian health care study” found that about half of all complications were considered preventable.6 DRG-based payments for inpatient care are not tied to the quality of the outcome,7 and provider-caused complications may actually result in increased payments.

In this study, adapting methods outlined in a previous study carried out in California,4 we translated the eight HAC diagnosis codes from ICD-9-CM to the International classification of diseases — Australian modification (10th revision) (ICD-10-AM) (5th edition) codes and used acute inpatient admission data from Victorian hospitals to estimate the likely effect the HACs policy would have on hospital payments in Victoria.

There were 2 047 133 acute admissions to Victorian public or private hospitals between 1 July 2007 and 30 June 2008, incurring total estimated payments of $4902 million. There were 14 700 cases (0.72%) in which the length of stay was higher than expected for the specific AR-DRG (high outliers). These cases attracted $253 million (5.17%) in additional payments.

Of the 2 047 133 acute cases, 994 (0.05%) had one or more diagnoses meeting the code definition for one of the seven definable HACs (Box). The total estimated payment for the 994 cases in which an HAC was detected was $24.1 million; 310 of these cases (31.2%) were eligible for outlier payments, which represented 17.3% ($4.2/$24.1 million) of their total payments.

In-hospital falls and trauma, pressure ulcers, catheter-associated urinary tract infections, and foreign objects retained after surgery were the most commonly coded of the seven definable HACs we studied. Removal of diagnosis codes for these seven HACs changed the DRG category in 134 cases (13.5%). There were no cases in which the DRG changed on the basis of excluding codes for air embolism or mediastinitis. We estimate that these 134 cases would account for a $448 630 reduction in total payments, representing only 0.01% of annual estimated payments for all cases ($0.4/$4902 million). Considering only public hospital cases affected by the policy, the proportion of cases would be 0.01% (87/1 301 898), with a financial impact of 0.01% ($302 914/$3117 million) (data not shown).

Our analysis shows that the annual number of cases likely to be affected by adopting the CMS policy of non-payment for HACs in Victorian hospitals would be small, and that the revenue implications for hospitals would also be small. Based on 2007–08 hospital admission data, we estimate that the HACs policy would reduce total hospital payments by between $448 630 (0.01%) and $1 690 724 (0.03%). The financial impact for public hospitals would be between $302 914 (0.01%) and $1 329 049 (0.04%).

In our analysis, long length-of-stay (outlier) cases accounted for 0.72% of all cases but constituted 31.2% of cases involving one of the eight HACs. This over-representation may be the result of long lengths of stay induced by these HACs (eg, for mediastinitis after coronary artery bypass graft) or may reflect an increased likelihood of complications (eg, pressure ulcers) for patients with long lengths of stay.

Impact of deleting the eight HAC codes: comparison between California and Victoria. Although small, the potential impact of the HACs policy in Victoria was 10 times the estimated proportion of payments affected by the CMS policy in the Californian population eligible for Medicare payments (0.001%).4 The small variations in payment can be explained by the robustness of DRG-based payment models. Admissions that group to a DRG + CC often involve not just one, but a suite of complications and co-morbidities, which reduces the likelihood that a single HAC code will make the critical difference in DRG assignment.

In Victoria, 14.2% (108/759) of affected cases regrouped to another DRG after deletion of one or more of the six HAC codes (excluding pressure ulcers). This rate is higher than the 3.1% (26/831) of California Medicare cases that were found to regroup after deletion of the six HAC codes.4 The difference in proportions between Victoria and California may be due to differences in the DRG grouper used in each jurisdiction (AR-DRGs versus Medicare-severity DRGs [MS-DRGs]). As there are more AR-DRGs than MS-DRGs (667 v 559), the former are more finely granulated. Also, California Medicare beneficiaries are predominantly 65 years or over, in contrast to Victorian beneficiaries, who include all residents of the state. As a result, California Medicare beneficiaries are more likely, on average, to have comorbidities. Hence, admissions involving Medicare beneficiaries in California are less likely to have one of the seven HACs as the critical diagnosis code that, on deletion, would change the case from DRG + CC to DRG – CC.

Study limitations. US hospitals have responded to the introduction of previous DRG-based payment policies with changes to their coding practices.9 While the simple code deletion strategy, as proposed by CMS, provides limited capacity to “game the system”, there may be little incentive for hospitals to include the eight HAC codes if their incidence is reported with no prospect of additional payment.

We based our study on Victorian data because Victoria was the only Australian jurisdiction with both DRG-based funding and a history of collecting a condition-present-on-admission indicator. However, the results of our study would be equally applicable to other states that use a DRG-based inpatient payment system and now collect condition-present-on-admission data.

Potential policy modifications. The HACs policy is predicated on the theory that financial incentives will at least raise the profile of the eight HACs and, at best, encourage systematic measures for minimising the likelihood of complications of care. Although financial incentives should be only one component of any reduction strategy for complications of care, they must be large enough to attract the attention of both clinicians and hospital administrators without jeopardising the immediate financial viability of health services, particularly public services in rural and remote areas that struggle to remain financially viable. (Small, remote hospitals in Victoria often receive block funding rather than casemix-based funding, reflecting the expectation of service provision rather than efficient throughput.)

Implementing the HACs policy would have a range of intended and unintended consequences. Non-payment for the eight HACs if they arose after admission might encourage careful detection and documentation of potential HACs at admission. The presence of most of the eight HACs can be detected during admission by a basic physical examination (by nursing or medical staff), and most can be prevented by standard nursing care practices (eg, postoperative surgical pack counts; use of pressure ulcer and falls risk assessment tools that prescribe preventive measures; aseptic vascular catheter insertion and routine insertion site care; and care and maintenance of indwelling urinary catheters). Victoria has a range of quality improvement initiatives in place to address some of the eight HACs, with particular emphasis on pressure ulcer and falls prevention.10,11 Although gains have been made in these areas, attaching a financial incentive might help to further focus attention on the issue and justify investment in prevention.

The CMS policy of not paying for HACs affects only a limited suite of complications of care that are arguably preventable. Another that has been suggested is to delete all codes for conditions arising after admission (irrespective of preventability) and to pay for them at a “complication-averaged rate”.12 This approach would have many advantages over simple output funding, but could be too great a change for hospitals to manage in a single step. Two incremental approaches to strengthening payment policies should be considered: increasing the proportion of payments affected by the policy, or starting with a low-impact policy (such as the HACs policy) and strengthening it over time.

There may be an opportunity to strengthen the policy by formalising existing data collection conventions. Specifically, all diagnosis and procedure codes in routine Victorian hospital data are sequenced chronologically, with some specific standardised exceptions. (This contrasts with the US system, in which only eight secondary diagnoses and five secondary procedure codes are used for MS-DRG assignment and can be resequenced by the hospital to maximise payment.) If these conventions were adopted as a matter of policy, sequelae of HACs (eg, sepsis following vascular catheter infection) could be identified and potentially used as part of the funding modification.

One of the aims of such a funding policy would be to draw attention to areas in which there is potential for improving patient outcomes. The impact of each policy would be largely dependent on the amount of attention it received, irrespective of the financial impact. While the capacity for the HACs policy to drive reductions in complications would be limited, as in the US, it would seem to be a prudent first step towards outcome funding. If introducing a HACs policy failed to draw adequate attention, jurisdictions could further strengthen the financial impact of the HACs policy in several ways — for example, by: (i) expanding the range of diagnosis codes for deletion per discharge by including codes that are frequent sequelae of the eight HACs; (ii) expanding the discharges affected by the HACs policy to include transfers to acute or subacute facilities; (iii) excluding payments for readmission for any of the eight HACs; and/or (iv) altering the general approach to payment for complications. These options have been discussed elsewhere, albeit in the US context.4

Alignment of funding incentives. Introducing any policy of non-payment for HACs is unlikely to result in immediate savings for funders, given the very small potential reduction in payments. Significant savings from reduced payment for the eight HACs, either through reduced length of stay or readmission, are likely to be offset to some extent by investment in prevention activities at the hospital level that will eventually be reflected in subsequent AR-DRG cost weights. Although the HACs payment policy was initially introduced in the US in response to the Deficit Reduction Act,13 the main benefit of the policy may be from reduced patient morbidity rather than large reductions in costs.

Components of a broader complication prevention program. Financial incentives are one part of a broader complication prevention program in which the dollars at risk need to be large enough to engage hospital administrators so that resources are provided for risk reduction/quality improvement. Feedback to hospitals on rates of HACs may, in and of itself, create an incentive for clinicians and hospitals to address problems, and such feedback could readily be incorporated into other current outcome-monitoring processes.14-16 Whether these rates should be published is a matter for debate. Some jurisdictions are choosing not to publish comparative outcomes in an effort to reduce gaming around outcome reporting and thus encourage health services to acknowledge and address, rather than deny, the need for risk-reduction strategies.14 Other jurisdictions are attempting to publicly report the results of remedial actions rather than the initial data.17 As a minimum, each hospital should receive benchmarked reports of their performance under qualified privilege.

There may also be an opportunity to achieve further HAC reductions by consolidating the data reported back to health services and developing virtual learning communities.