Which includes all outpatient procedures and services provided during one day to the same patient?

Diagnosis-Related Groups

DRGs, discussed extensively in Chapter 11, were developed in the 1960s as an alternative way of paying for hospital care in order to encourage shortened lengths of stay. Experience with payment by days of care (per diem) showed that it promoted unnecessary, lengthy, and potentially dangerous use of hospital care, an important factor in the rapid escalation of costs in the health system. DRGs were adopted for payment for Medicare beneficiaries in the USA in 1983 and later became the standard method of payment for all insurance systems.

In the DRG system the insurer pays the provider hospital for a procedure or diagnosis rather than the number of days of stay in hospital. This has led to a large reduction in hospital days of care and a remarkable growth in the number of surgical procedures done on an outpatient basis. Since the introduction of DRGs, outpatient surgical procedures have grown from less than one-fifth to more than half of inpatient surgical cases. Outpatient surgery is safer for the patient and less costly to the insurer. DRGs have gradually been adopted as a case payment system for reimbursing hospitals in most developed countries.

The DRG system is widely considered to promote quality of care as an active process focusing on quickly addressing the diagnosis and management of the patient with rapid mobilization of treatment and return home. Critics of this system allege that DRGs encourage inappropriate early discharge of patients before optimal patient education and follow-up care have been provided, but long length of hospital stay has not been shown to improve patient outcomes. Critics also suggest that this may promote altering diagnoses to higher cost units of service. Others think that DRGs, by reducing length of stay, have turned hospitals into intensive care units with ultra-sick patients. Despite these issues, the trend towards short hospital stays and newer approaches to active treatment seems to be compatible with better care and improved outcomes, according to some measures. The rapid decline in mortality rates from coronary heart disease is thought to be due in large part to the activist treatment approach, with lengths of stay of 1 week or less for acute myocardial infarction compared to 6 weeks on average up to the 1970s.

DRGs

The DRG payment system was developed in the 1960s at Yale University in the US due to concerns about high costs and the search for alternative methods of payment. The DRG system was officially adopted in 1983 by the US Health Care Financing Administration (HCFA) as the basis for payment for hospitalization of Medicare patients. The DRG system has been the basis for paying for hospital care in the US since 1999 by most health insurers, and has been adopted by other industrialized countries—e.g., the United Kingdom and Israel—and some low- and middle-income countries, including the Philippines, and countries in eastern Europe, including nine countries in transition from the Soviet system.

In the DRG system, hospitals are funded based on a predefined payment rate for diagnoses or procedures in 495 classifications. This incentivizes the appropriate use of services with a reduction in length-of-stay, efficient use of diagnostic and treatment procedures, and reduces overall bed capacity. Implementation of the DRG system in the 1990s in the US led to a rapid decrease in admissions and an increase of outpatient services while bed occupancy rates and per capita hospital bed supply declined steadily. The DRG system, however, encouraged the falsification of diagnoses or reported severity of case definitions in order to increase revenues, which became known as “DRG creep.”

Prospective payment systems, such as DRGs, support rational use of hospital care as an effective way to achieve a balanced health service system and must be associated with quality assurance mechanisms. A regional approach for hospital budgeting as part of a comprehensive network helps to achieve equity and provides incentives for increasing health in the population.

How to Pay for Unobservable Heterogeneity?

Fixed payments per DRG put pressure on hospitals to compete. Because payments levels are set at average cost, hospitals which are affected by exogenous factors that induce higher than average costs risk losses. If they are already operating at full efficiency, they cannot realize further savings through efficiency gains. Hence, careless implementation of a PPS is likely to create undesirable incentives for selecting patients and lowering care quality. A regulator who aims at maximizing social welfare must design a payment system that creates virtuous incentives for enhancing hospital efficiency, without providing deleterious incentives for patient selection and quality reduction.

To address this question, many theoretical papers have tried to improve the basic model by lifting assumptions relative to patient and hospital homogeneity, and by allowing for endogenous levels in the number of procedures and quality of treatment. Using various theoretical frameworks and hypotheses, these papers show that social welfare can be improved through a mixed payment system that combines a fixed price with partial reimbursement of the actual cost of treatment per stay. To deal with unobserved sources of heterogeneity in costs, the regulator can construct a menu of contracts that combine a lump-sum transfer with partial reimbursement of actual costs. When the hospital chooses a contract, it reveals its unobserved cost component. Currently, however, such a payment scheme is not implemented in any health system. In fact, the theoretical design of the contracts often relies on unobservable variables or functions, such as, for instance, the function describing the disutility of the hospital manager's cost reduction efforts. Hence, such theoretical designs are hardly used in reality.

Another strategy is to use econometrics to evaluate unobservable sources of cost heterogeneity. The sources of hospital cost heterogeneity are summarized in Table 1. A hospital's activity is more or less costly, depending on its infrastructure, the existence of economies of scale or of scope, the quality of care and the cost reduction effort provided by the hospital manager (moral hazard). Moral hazard can be split into two components: long-term moral hazard and transitory moral hazard. Long-term moral hazard is supposed to be time invariant: the hospital management can be permanently inefficient. An example of permanent inefficiency would be an obsolete elevator which is very slow and subject to frequent breakdowns and which is not replaced for several years. Transitory moral hazard is linked to the manager's transitory cost reduction efforts. For instance, the manager can be more or less rigorous, each year, when negotiating prices for supplies or for services provided to the hospital by outside firms. It would be optimal for social welfare to eliminate long-term moral hazard as well as transitory moral hazard. However, it is very difficult to separate long-term moral hazard from other sources of cost heterogeneity which are legitimate.

The use of a three-dimensional nested database makes it possible to identify transitory moral hazard. It is then possible to design a payment that allows for hospital heterogeneity in costs, while still providing incentives to increase efficiency because it does not reimburse costs due to transitory moral hazard (see the technical appendix).

A fully PPS reimburses each stay with a fixed price regardless of the actual cost of the stay. The payment systems currently implemented in most countries take some observable sources of cost heterogeneity, such as local input prices, into account. A preferable method of payment would be to allow for observable and some unobservable sources of cost heterogeneity, provided they are time invariant. With such a payment rule, the regulator reimburses each hospital for extra costs that might correspond to undesirable long-term moral hazard, but which can as well correspond to legitimate heterogeneity. Nevertheless, this method of payment creates incentives to increase efficiency because it does not reimburse extra costs that are a result of transitory moral hazard.

The general idea is that the regulator has no means to disentangle legitimate and illegitimate sources of time-invariant cost heterogeneity, i.e., to separate the wheat from the chaff. In this context, it might be preferable to accept to pay for long-term moral hazard in order not to penalize hospitals which have legitimate sources of cost heterogeneity. Is this view unreasonable? The question becomes an empirical one: if transitory moral hazard has a substantial impact on cost variability, it would be possible to achieve large gains in efficiency even while paying for permanent sources of hospital cost variability.

An empirical estimation has been carried out by Dormont and Milcent (2005) on a sample of stays for acute myocardial infarction in French public hospitals. It appears that the cost variability because of transitory moral hazard was quite sizeable. Simulations show that substantial budget savings – at least 20% – could be expected from implementation of a payment rule that takes all unobservable hospital heterogeneity into account, provided that it is time invariant. This payment rule is easy to implement if the regulator has information about costs of hospital stays. A drawback is that it gives higher reimbursements to hospitals which are costlier because of permanently inefficient management. However, it has the great advantage of reimbursing high quality care. Moreover, it can lead to substantial savings, because it provides incentives to reduce costs linked to transitory moral hazard, whose influence on cost variability can be sizeable

Unintentional lack of transparency and the effects on justice

The ecology of shareholders and stakeholders judged ethical and just must at least include transparency as necessary in a modern liberal democracy, particularly those relying on the “Invisible Hand” of laissez faire capitalism (see Case 3 and Chapter 4). Lack of transparency can be intentional, such as “gag” rules that were applied, even if only implicitly, by managed care organizations, or can be unintentional. One way of unintentional lack of transparency is ignorance, where important factors and principles are not recognized by the parties involved. Frequently, these opacities are unintended consequences of well-intentioned efforts.

It is important to emphasize the relationship of intention to ethics and justice, for example, the notion of a crime versus an unethical or unjust act. One component is the intention on the part of the perpetrator of the unethical or unjust act, mens rea in legal terminology. An intentional unethical act often can be readily recognized as unjust and perhaps even a crime. It becomes more problematic when the unethical act is unintentional. But if unintentional unethical acts are “forgiven” or “given a pass,” then it is the person who suffered the unethical act who “pays the bill.” Indeed, reducing personal responsibility on the part of the physician or healthcare professional may be one motivation to discuss medical errors as “systems” errors. However, the justification often given is to relieve some fear on the part of physicians and healthcare professionals, allowing them to more fully participate in remediation and prevention efforts. It is unclear how successful these efforts have been (Wears and Sutcliffe, 2020).

The diagnosis-related group (DRG) method or reimbursement of hospitals may be an example of unintentional lack of transparency that could be argued to limit patient care, which again may not be unethical or unjust. However, to come to some conclusion of the ethics and justice of DRGs, it is important to assess how DRGs affect patient care, particularly unintentionally limiting care in a manner that reasonable persons blinded to their self-interests would find unacceptable. Further, an ethical analysis of DRGs is relevant to any method of prospective payment, such as capitated care. The intent of the DRG program was “[create a] DRG prospective payment system, Medicare pays hospitals a flat rate per case for inpatient hospital care so that efficient hospitals are rewarded for their efficiency and inefficient hospitals have an incentive to become more efficient” (https://oig.hhs.gov/oei/reports/oei-09-00-00200.pdf). Typically, the reimbursement is based on the average length of stay (for inpatients) associated with specific diagnostic groups. As a government report stated,

The average standardized charge for each DRG is calculated by summing the charges for all cases in the DRG and dividing that amount by the number of cases classified in the DRG. Statistical outliers, those cases outside three standard deviations of the average charge for each DRG, are eliminated. The average charge for each DRG is re-computed and then divided by the national average standardized charge per case to determine the weighting factor.

(https://oig.hhs.gov/oei/reports/oei-09-00-00200.pdf)

The operating presumption is that basing reimbursement on average length-of-stay reimbursements should be fair, assuming that enough patients within a DRG are treated by a specific institution. A patient staying in the hospital longer than the average length of stay would cause a financial loss that could be made up by another patient whose stay is shorter. However, in this author’s experience, the DRG length of stay was a target that was to be met with for every patient. Whether intended or not, soon some physicians and healthcare professionals were modifying their practice more to comport with the DRG’s average length of stay than what might be in the patient’s best interest. Note that if the distribution of lengths of stay is normal (following a bell curve), 50% of patients will stay longer than the average and 50% will stay shorter. However, if every patient stays the average or less, then 50% of patient who should have stayed longer will be discharged prematurely. Again, such actions are not necessarily unethical or unjust, depending on which moral theory is to supervene.

Perhaps one measure of premature discharges, perhaps related to the undue influence of DRGs, is the readmission rate; however, this is problematic as it may represent only extreme cases. The assumption is that an excessive readmission rate is taken as evidence of premature discharge of a patient. However, it is unknown how many patients are discharged prior to optimal treatment but whose conditions are not sufficient to warrant readmission. How pervasive and significant the effects of physicians emphasizing length of stay, as possibly evidenced by the readmission rate, are difficult to assess and the results of various studies have been mixed. Not many studies have directly examined whether the length of stay in a DRG-based system of reimbursement increases the risk of premature discharge. In part, the complexity and confounds, such as comorbidities, demographics, and socioeconomic status, make such analyses very difficult. For example, just increasing the variance in any measure, as would occur in more complex situations, would make it more difficult to determine the effects of the length of stay on readmission rates. In a study of treatment of head and neck cancers, shorter lengths of stay increased the risk of readmission (odds ratio and [95% confidence interval] 1.34 [1.22–1.48]) (Puram and Bhattacharyya, 2018). However, a number of other studies failed to demonstrate any effect of shorter lengths of stay on readmission rates (Rich and Freedland, 1988; Morse et al., 2013; Weissenberger et al., 2013; Kim et al., 2016; Dexter et al., 2017).

It is interesting and perhaps valuable to better analyze the effects of DRG average length of stay, at least for the potential of misunderstanding, as the implications are relevant to any prospective payment system based on the benchmarks such as the average length of stay. Assume that the physician or healthcare professional takes the average length of stay as the target for the management of each and every patient and that the target has a significant effect on the physician’s or healthcare professional’s discharge from care. What are the implications from a statistical perspective? For example, if every patient used to calculate the DRG associated average length of stay was exactly the same, for example, 7 days, then the physician or healthcare professional discharging the patient after 7 days likely would be appropriate for each and every patient. What if the range of length of stay was from 6 to 8 days with a mean of 7 days (assuming a normal distribution within the data set)? Then, on average, some patients may be discharged 1 day earlier than the average, making the likely impact relatively small. If the range is from 1 to 13 days, a significant number of patients may be discharged many days earlier than would be optimal and likely would experience a greater rate of harm, a violation of the obligation to the ethical principle of nonmaleficence. Thus, more variance in the data set used to determine the DRG average length of stay creates a greater impact on the consequences of discharging every patient at the average length of stay, assuming earlier the discharge means greater harm. Perhaps the government excluding “outliers” in the DRG length of stay calculations, as described earlier, may be a problem.

This author could not find any studies that examined and reported the variance in the calculation of the DRG average length of stay with the exception that patients more than three standard deviations from the average were excluded as described previously. Note that this is not to say that there are no such studies but they would be relatively hard to find using traditional literature search engines. The exception was a study of the lengths of stay in patients with 17 dermatologic conditions in an area in Germany that reported mean and standard deviations in the lengths of stay (Wenke et al., 2009). Median and interquartile scores were also reported, but for the sake of discussion, means will be considered as they were not much different than the median for many conditions. In order to combine experiences across different diagnostic categories, the standard deviation was used to normalize the standard error by dividing the standard deviation by the square root of the sample size. The standard error was divided by the mean and then multiplied by 100 to generate the relative standard error of the mean.

As can be seen in the figure shown here, there is a broad range of relative standard errors of the mean. To appreciate the significance, in the case of lymphomas, consider that the relative standard error of the mean was 7.46 and the associated mean was 9.2 days with a standard deviation of 11.6 (the caveat is that the median was 5 days, suggesting a skewed deviation, and therefore the relative standard of the error must be taken with caution). In this case, the standard deviation is greater than the mean. Targeting solely the length of stay would result in many patients discharged prematurely. In the case of psoriasis, the mean length of stay was 15.6 days with a standard deviation of 7.2 days or half the mean. The relative standard error of the mean was 1.13. In the case of psoriasis, discharging all patients at 15.6 days would result in relatively fewer patients being discharged prematurely. To be sure, the treatment of psoriasis may be very different than the treatment of lymphoma where treatment of the former is more regimented compared to the latter. The point here is not to comment on dermatological practices but only to illustrate the problematic nature of using the average length of stay to guide treatment (see Montgomery, 2019) (Fig. C8.1).

The discussion just described makes no claim to the ethical status or justice of using the average length of stay or other such metric as a significant factor in healthcare decisions by any shareholder. There is, at least, the potential for such an approach to affect the consequences (clinical outcomes) and thus the ethical status can be judged in terms of the consequences in the context of moral theories. For example, from a Utilitarian moral theory, if the resources saved by early discharge are less than the resources expended for any subsequent remedial treatment, such as readmission (disregarding any pain or suffering from insufficient treatment), then targeting the average length of stay, while perhaps not fully fulfilling the obligations to the ethical principles of beneficence, nonmaleficence, and autonomy to the individual patient, would not be unethical and, therefore, not unjust. A Deontological moral theory would hold that the average length of stay should not have an effect if it is the right of every patient being hospitalized as long as necessary to achieve optimal treatment. From an Egalitarian moral theory, as long as every patient is treated the same, that is, all patients are discharged at or prior to the average length of stay and sufficient numbers of persons (e.g., citizens who presumably are shareholders) agree (directly or via their proxy), then the use of average length of stays would not be unethical or unjust. The Libertarian moral theory would settle on a contract with unethical and justice behaviors based solely on violation of the contract.

Ultimately, all moral theories resolve down to Contractualism in modern liberal democracies. This is not to say that such contracts cannot be regulated consistent with other moral theories, such as a Deontological moral theory. Indeed, in most pluralistic modern liberal democracies, there are constraints on contracts. The elements of most socially acceptable contracts include (1) an exchange of agreed upon goods, (2) each party is free to enter into the contract, (3) each party is capable of entering into a contract, (4) there is a good faith effort to fulfill the terms of the contract, and (5) lawful purpose. Justification of the elements of a contract cannot come from a Libertarian moral theory, as Contractualism is not part of a Libertarian moral theory but a necessity to avoid anarchy. Further, the elements do not appear to abide any qualification, thereby reducing the normative effects of Egalitarian or Utilitarian moral theories.

Contracts rely on a mutual exchange of goods with the parties entering into such contracts freely. However, there is a subtle difficulty in DRG-based systems. The “ICD-10-CM/PCS MS-DRG v34.0 Definitions Manual” wrote:

When clinicians use the notion of case mix complexity, they mean that the patients treated have a greater severity of illness, present greater treatment difficulty, have poorer prognoses and have a greater need for intervention. Thus, from a clinical perspective case mix complexity refers to the condition of the patients treated and the treatment difficulty associated with providing care. On the other hand, administrators and regulators usually use the concept of case mix complexity to indicate that the patients treated require more resources which results in a higher cost of providing care. Thus, from an administrative or regulatory perspective case mix complexity refers to the resource intensity demands that patients place on an institution. While the two interpretations of case mix complexity are often closely related, they can be very different for certain kinds of patients [italics added]. For example, while terminal cancer patients are very severely ill and have a poor prognosis, they require few hospital resources beyond basic nursing care.

In the past, there has sometimes been confusion regarding the use and interpretation of the DRGs because the aspect of case mix complexity measured by the DRGs has not been clearly understood. The purpose of the DRGs is to relate a hospital’s case mix to the resource demands and associated costs experienced by the hospital [italics added]. Therefore, a hospital having a more complex case mix from a DRG perspective means that the hospital treats patients who require more hospital resources but not necessarily that the hospital treats patients having a greater severity of illness, a greater treatment difficulty, a poorer prognosis or a greater need for intervention.

(https://www.cms.gov/ICD10Manual/version34-fullcode-cms/fullcode_cms/Design_and_development_of_the_Diagnosis_Related_Group_(DRGs)_PBL-038.pdf)

This quote suggests a difference in perspective and expectations between physicians and healthcare providers and the healthcare delivery systems. Such differences, if they are not resolved, raise the question of whether the terms of a good contract can be met, thus placing stakeholders at risk. Should the differences ultimately place the patient at risk, then where does the responsibility lie? Do physicians and healthcare professionals understand the statistical nuances of average length of stay or other similar benchmarks and, if not, should they? If they should, then how much responsibility falls on the institutions responsible for education, training, and supervision?

Obtaining New Procedural and Episode-of-Care–Based Reimbursement Codes That Adequately Describe and Reimburse High-Value Therapies Can Be Prolonged and Difficult to Achieve

While existing procedural and episode-of-care DRG-based reimbursement codes often do not accommodate novel transformative therapies, in some markets it may be a difficult and prolonged process to obtain a new code to support adequate reimbursement. In the US and other countries, obtaining a new procedural code (e.g., CPT in the US, Classification Commune des Actes Médicaux (CCAM) in France, Operationen-und Prozedurenschlüssel (OPS) in Germany) for clinician reimbursement for administering novel outpatient therapies may require demonstrating the need for a new code and evidence for the value of the new procedure, which includes showing sufficient existing procedure volume in practice or why the existing code set does not adequately describe the new procedure or therapy. (Busse et al., 2011; Remuzat et al., 2015; American Medical Association; Jorgensen and Kefalas, 2015; German Institute of Medical Documentation and Information (DIMDI)) This sets up a chicken-or-egg conundrum, in that obtaining a new procedural reimbursement code requires demonstrating volume for the procedure, but achieving such volume may require that clinicians are sufficiently reimbursed for administering the therapy, without needing to take on risk of procuring a high-cost therapy prior to being reimbursed by the payer (and one hurdle to provider adoption observed when Provenge launched in the US at $93,000) (Timmerman, 2011; Holcombe, 2012a; Wong, 2014). In addition, the process of obtaining a new procedure code and specific payment rate can be prolonged. In the US, for example, the process of issuing a new CPT code can take as long as 18–24 months, followed by up to another year for CMS to set the Medicare payment rate. After this entire multistep process, the payment rate set may not be any better aligned with the clinician effort to administer the new therapy than the previous best matched code.

For new inpatient therapies in the US, obtaining a new DRG code is very rare and not typically achievable. In other markets, it may be easier to obtain a new procedural or DRG-based reimbursement code for a novel inpatient therapy, but it could also be a multistep, multistakeholder process that takes substantial time and effort for the new code to be implemented, and then even more time for payment rates to approach desired payment range with annual updates. In some markets, such as Germany and France, the process of obtaining a new or revised DRG code to support hospital reimbursement for inpatient treatment is possible, albeit prolonged and complex, and typically must be initiated by an official health system entity such as a hospital or a relevant medical society (Busse et al., 2011; German Institute of Medical Documentation and Information (DIMDI)a,b). It is now clear-cut that the process is not easily amenable to smaller manufacturers with single assets that need appropriate reimbursement immediately at launch to achieve economic sustainability (Remuzat et al., 2015).

Nevertheless, when existing DRG-based codes with unacceptably low reimbursement rates must be used and cannot accommodate a new innovative inpatient therapy, additional options do exist but come with their own challenges. These options for DRG-based reimbursement to better accommodate costly new inpatient treatments include separate payments, supplementary payments and cost-outlier funding (Busse et al., 2011). In Germany, a hospital may apply for New Diagnostic and Treatment Methods Regulation (NUB) reimbursement that provides for a temporary separate payment arrangement outside of the DRG-based reimbursement (Busse et al., 2011). The application process involves multiple steps and stakeholders and depending on the specific scenario may require each individual hospital to apply separately for their own NUB reimbursement (Busse et al., 2011).

In the US, there are temporary options that would be classified as supplementary payments and cost-outlier funding. These options include New Technology Add-on Payment (NTAP) and Hospital Outlier Payment mechanisms. The NTAP payment mechanism is meant to accommodate additions of costly new technologies to DRG-based reimbursement based on novelty, substantial clinical improvement and cost-based inadequacy of current MS-DRG payment rates and allowing for reimbursement of innovative technologies during a temporary period (i.e., 3-year) prior to the next DRG update incorporating the new technology in the cost data (Center for Medicare and Medicaid Services, 2018c; Center for Medicare and Medicaid Services). The US CMS publishes cost thresholds for consideration of NTAP under the Inpatient Prospective Payment System (IPPS). Historically, however, most NTAP applications are not successful. For example, from 2001 through 2014, CMS approved only 35.8% of NTAP applications (Commonwealth and Young, 2015); in 2016, 62.5% of NTAP applications were denied (Littmann, 2015). However, as elaborated earlier in this section, even when an NTAP is granted by CMS, given that innovative technologies often target a small percentage of the patients and discharges for a DRG, the CMS payment update at the 3-year mark typically does not increase the payment sufficiently to accommodate the new technology (Werner, 2018).

The other mechanism in the US to accommodate inpatient stays more costly than typical payments allow, which could apply in the case of incorporating new innovative therapies, is the Hospital Supplemental Outlier Payment (Center for Medicare and Medicaid Services, 2013). In an example highlighted on the CMS website, the FY-2005 fixed-loss threshold for consideration of a DRG stay for outlier payment was $25,800 beyond the standard CMS-accepted DRG cost, which CMS derives not on the basis of hospital charges but based rather on their accepted cost-to-charge ratios (Center for Medicare and Medicaid Services, 2013). The Outlier Payment to the hospital is ultimately a percentage of the cost above the DRG cost (e.g., if the CMS-accepted cost for a specific FY-2005 DRG case were $20,000, then the cost of the outlier inpatient stay would need to exceed $45,800 to be eligible for an Outlier Payment, and CMS would pay a percentage of the costs exceeding $20,000). For most recent transformative therapies, Outlier Payments would still be insufficient to accommodate the cost of providing the new inpatient therapy. In addition, the US Office of the Inspector General (OIG) issued a report calling for greater CMS scrutiny of Outlier Payments to hospitals, based on their concentration in certain hospitals and overall prevalence (Department of Health and Human Services, 2013). The report may lead to greater stringency in qualifications for Outlier Payments, making it more difficult for providers and manufacturers to negotiate for outlier cost-based payment of innovative therapies in the current reimbursement environment. One potentially positive outcome may come from the OIG recommendation that CMS evaluate whether the MS-DRGs associated with the most frequent and substantial outlier payments might justify coding changes (Elko, 2013). Notably, between 2008 and 2011, 25.4% of MS-DRG for allogeneic BMT (014) claims submitted to Medicare were submitted for Outlier Payment (Department of Health and Human Services, 2013), which could indicate that the MS-DRG payment rate is now too low. The American Association of Blood Banks (AABB) and National Marrow Donor Program (NMDP) recently submitted comments to CMS on the proposed FY2018 IPPS highlighting the need for increased payment for allogeneic BMT in part to cover the cost to the provider of stem cell acquisition from a blood bank (American Association of Blood Banks, 2017; National Marrow Donor Program, 2017).

Hospitals

Over 90% of hospitals in Sweden are publicly owned, with the most advanced care carried out at seven public university hospitals. There are only six private hospitals in the country, and three are nonprofit. Sweden has the fewest number of acute inpatient beds per capita of any of the 10 countries in this book [14].

Hospital payment varies across the regions with a mix of global budgets, diagnosis-related groups (DRGs), and performance-based payment based on quality attainment. Global budgeting still dominates in most counties and tends to be based on historical costs, with DRGs were used in only 3 of the 21 regions in 2020.c Some regions tried DRGs but reverted to global budgets because it did not increase efficiency. This may have been because physicians, who are generally salaried, did not have an incentive to change their behavior [24].

Background on hospital finances and expected impact of health care reform

Since the 1980s most U.S. hospitals have been paid by Medicare through diagnosis-related groups (DRGs). The introduction of DRGs shifted payment from a “cost plus profit” structure to a fixed case rate structure. Under a case rate reimbursement, the hospital is not paid more for a patient with a longer length of stay, or with days in higher intensity units, or receiving more services. The diagnostic categorization of the patient determines the reimbursement rate. Thus, if all other things are equal, if a patient has a shorter stay in a lower intensity bed with fewer procedures and tests, the costs to the hospital will be lower and the revenue will be unchanged, thus the contribution margin (revenue minus variable costs) will be improved. Of course, these tradeoffs will need to also result in neutral or improved quality, or the improvement will be illusory.

Under this design, in place for more than 25 years, hospitals are not at risk for costs of care outside of their doors and in fact may profit from a cycle of hospitalizations followed by extensive use of specialists and outpatient facilities (particularly affiliated outpatient surgery centers, testing and diagnostic centers, and cancer care). Costs that drive up cost per admission such as entry through the emergency department (ED), use of ICU beds, and delays in care or delays in discharge have a direct negative impact to the hospital.

All hospitals are not paid on DRGs; a small but important subset such as Veterans Administration hospitals, critical access hospitals, or “DRG exempt” hospitals have different contractual arrangements that will affect their financial case for palliative care. It is therefore important to verify how a hospital is paid based on case rates before significant financial modeling. Even in the other cases, compelling approaches may exist and many of the suggestions included in this chapter will apply, but these will need to be crafted with clarity about hospital priorities and realities.

As of 2009, national data indicated that Medicare discharges represented 37.3% of total discharges, Medicaid represented 20.4%, private insurance 32.9%, uninsured 6.1%, and other 3.2%.4 When considering total days of care, Medicare represents a higher proportion of total care. When reviewing cases likely to have palliative care needs, Medicare often represents 60% or more of patients. Contracts with private insurance and with Medicaid often also reflect DRG case rate design, although the level of reimbursement may vary from Medicare rates. Therefore it is common for 80% or more of patients cared for by palliative care to be covered by case rate payments. However, checking payer mix and contract type through a general discussion with finance staff is advisable before building an elaborate business case. For example, a suburban hospital with a large obstetrical presence may well have needs different from those of an urban tertiary center.

In spite of the focus of hospitals on managing costs, U.S. acute-care hospital cost per case is more than double the median of 10 comparative developed countries, although median length of stay is below those of the comparison group.5 This situation of high costs per case versus low length of stay has implications for the potential of palliative care to affect financial performance of hospitals. The high costs per day reflect high capital costs (fixed costs per bed), high intensity of care (proportion of surgical cases and ICU cases), and higher procedural activity (laboratory, radiology, other interventions) and labor costs. During the past decade the proportion of U.S. patients seeing 10 or more physicians in their last 6 months of life has grown, hitting 36.1% in 2007,6 and the proportion of total U.S. patients being admitted through the ED has also grown from 33% to 43% as of 2010. A shortage of critical care beds is listed by hospital administrators as the top reason for ambulance diversions in 2010, accounting for 42% of diversions; 50% of urban and teaching hospitals reported ED capacity issues.7

The Affordable Care Act of 2010 has introduced significant change to the historic pattern of DRG payment for inpatient care. Changes targeted for 2012 to 2014 are expected to increase the financial risk borne by hospitals for care in and out of their doors (“bundling” for 30-day episodes of care), offer risk sharing for improved total costs of care (accountable care organizations), and introduce more penalties for gaps in quality, such as denial of payment for readmissions for certain conditions. This has accelerated the consolidation or expansion of health care delivery systems and has significantly increased interest from hospital leadership regarding continuum of care needs for patients with complex chronic illnesses.

In addition, the Hospital Consumer Assessment of Healthcare Providers and Systems (HCAHPS) survey of patient satisfaction will have a significant financial impact on the Center for Medicare and Medicaid Services (CMS) payment rates through a withholding and bonus program. The HCAHPS Survey is not specifically sensitive to palliative care activities, but includes pain management as an important component; the 2010 nationwide results show that only 64% of patients are very satisfied with pain management (this is the third lowest of all 10 satisfaction measures). Communication topics comprise five of the remaining nine measures. HCAHPS performance has the potential to affect millions of dollars of revenue per hospital and is tied to both relative performance and trended performance.8,9

These changes in the payment environment are collectively called value-based purchasing (VBP) and are considered very high visibility, which means that getting ready to succeed under the new rules is a top priority for most hospital senior leaders. The prospect of significant change in delivery system priorities and increased risk also increases the emphasis of hospital leadership on comprehensive change initiatives and reduces interest in small “one-off” innovations. Thus palliative care programs will need to be clearly aligned with the bigger initiatives in methodology, formal collaborations, and metrics. It is very important to note that the VBP activity is introduced by CMS for Medicare patients, with an emphasis on seniors. However, CMS payment initiatives will likely be adopted by commercial insurers, as was the trend when DRGs were introduced for case payments.

In summary, the majority of U.S. hospitals have been operating within a case rate payment system. Outpatient activities serve as independent profit centers or as feeder systems for inpatient admissions. Expected changes in 2012 and following will significantly shift the focus to costs throughout the health care delivery system and will further highlight comparative quality results and consumer satisfaction rankings through Hospital Compare and other sources.

Palliative care has a compelling case to make about its role in improving care and reducing avoidable costs in the current system, with even more opportunity within the expanded continuum of care focus if palliative care leaders can be active in the design, delivery, and measurement of expanded services and can work in collaboration with other services.

10.6 Summative ontologies: DRG

We conclude this tour of ontologies with what we call a “summative” ontology in the form of diagnosis-related groups (DRGs).11 We call this a “summative” ontology because the goal is to abstract patient encounters away from specific procedures and granular diagnoses and into a category that reflects the difficulty and general care pathway required for managing that patient encounter. As we discussed in Chapter 1, DRGs were born out of an effort to move from fee-for-service clinical reimbursement toward a “capitated” payment model wherein hospitals payments are given according to how challenging patient care is expected to be. Originating from work at Yale in the 1960s, the hope in doing this was initially to provide an accurate mechanism to monitor quality by facilitating patient grouping into broad, clinically meaningful buckets. The original DRGs consisted of 23 nonoverlapping groups called Major Diagnostic Categories (MDCs) that were partitioned in broad strokes such as major versus minor surgeries, neoplasms, trauma, etc. A logical extension of quality monitoring, the DRG system was first used to guide reimbursement in the late 1970s by the New Jersey State Department of Health and has since grown to be the principal means of abstracting patient care encounters for compensation by payers. The intention of DRGs is to incentivize hospitals to be more cost conscious by establishing a standard price for a given type of patient and patient encounter rather than paying hospitals according to an itemized bill for services rendered. The modern DRG groups patients according to diagnosis, treatment, and the expected length of stay and Groups are assigned based upon both principal and secondary diagnoses as represented by ICD codes and procedures as represented by HCPCS codes. This grouping is typically performed on patient encounters using a specific software application known as a “grouper,” which is a critical component of the hospital revenue cycle. There are actually multiple variations of the DRG system each with their own specific tweaks.

There are currently two main systems of DRG assignment, which have slightly different evolutionary origins and concerns. The Medicare Severity DRGs (MS-DRG) encompass approximately 750 groups and seek to capture the medical complexity or severity of a patient by allowing for group modifications in the form of optional “complication/comorbidity” (CC) and “major complication/comorbidity” (MCC) annotations. For example, the MS-DRG ID for “heart failure and shock without CC or MCC” is 293, while the ID for “heart failure and shock with MCC” is 291. Eligibility for CC or MCC designations is based upon certain secondary diagnoses and procedures as defined in the appendices to the MS-DRG tables published by CMS. The focus of MS-DRGs is, unsurprisingly, squarely on the Medicare population to the exclusion of other relevant groups such as children and pregnant women. Thus, an attempt to create a variant of DRGs, which was more broadly encompassing, resulted in the All Patient (AP-DRG) and subsequent All Patient Refined DRGs (APR-DRG) with APR-DRGs serving as an improvement over AP-DRGs through the incorporation of severity of illness subclasses. Thus, APR-DRGs have four basic subclasses for each DRG relating to severity of illness and risk of death: minor, moderate, major, or extreme. As with MS-DRGs these APR-DRGs are assigned first by identifying the principal DRG based upon principal diagnoses and then aggregating secondary diagnoses and their expected severity of illness levels, and modifying severity of illness based on age, primary diagnosis, and additional procedures according to specific APR-DRG grouping logic. For certain analytical questions, DRGs are powerful labels for summarizing patient records because they represent lots of embedded associations between conditions and difficulty of treatment. For example, if we wanted to consider how effective or ineffective our hospital is at managing generally related patient encounters, we would prefer to consider DRGs rather than ICD, HCPCS, or SNOMED systems because DRGs speak to resource expense and length of stay per DRG is a generally useful summary of the efficiency of care. If, however, we noticed that certain DRGs in our institution has widely varying length of stays, we may rather wish to extract patient features in terms of ICD and HCPCS codes in order to identify subgroups accountable for this variation. As with all things, we wish to have the right tool for the right job and the job varies.

Hospital Inpatient Drugs

Drugs that are dispensed as part of an inpatient episode are generally not reimbursed separately but are included in the bundled diagnosis-related group (DRG) payment for the hospital admission. Medicare updates its DRG payment rates over time, based on national average costs, by DRG, as reported in hospital cost reports. Private payers negotiate various forms of bundled payment for inpatient hospital care, with private rates generally above Medicare rates but also no separate reimbursement for inpatient drugs. Thus in the short run the cost of new inpatient drugs (or price increases for existing drugs) are borne by hospitals, with pass-through to payers with a lag, if/when the drug becomes standard of care and reflected in average cost for the DRG. In exceptional circumstances, a very high-priced new drug may be reimbursed separately from the DRG temporarily, until its cost is included in an increased DRG payment.

This system of bundled payment for inpatient admissions puts hospitals at risk for inpatient drug costs in the short run. Hospitals therefore have incentives to be price sensitive in designing their formularies and negotiate price discounts with manufacturers in return for preferred formulary placement. Larger hospital systems that negotiate on their own behalf and can enforce formularies have greater bargaining power and get larger discounts than smaller hospitals and those that bargain indirectly through group purchasing organizations (GPOs). However, as with PBMs/PDPs, hospitals have little or no leverage to negotiate discounts for drugs that have few or no close substitutes, which includes many specialty drugs.

Care recipient characteristics

There are multiple ways healthcare recipients may be described, by number, age, provisional or confirmed diagnosis, diagnosis related group (DRG), injury type, population type, socio-economic status, disability status, location, residential status etc. How are these characteristics quantified and measured? The number of patients accessing a health care service is fairly simple or is it? This needs a qualifier such as a number of patients attending a service per day, or treated/cared for per day or reviewed per service per day. Then we need to define the service type and patient day. Is someone who has spent 6 hr in the day surgery unit or an acute hospital bed classed as an inpatient day? Patient day is a commonly used denominator for many statistics, thus it is important to ensure consistency in interpretation and data use. To overcome these issues and ensure consistency in interpretation one needs to agree on metadata.

The Australian Institute of Health and Welfare (AIHW) has established a repository for national metadata standards for health, housing and community services statistics and information known as METeOR. Other countries have similar arrangements. The AIHW repository is based on the ISO/IEC 11179 Metadata registries standard [7]. Metadata is information about how data are defined, structured and represented. Once endorsed these metadata are referred to as data standards. These standards improve the quality, relevance, consistency and availability of health data. This provides meaning and context to data. It also describes how data is captured and the business rules for collecting data. Individual data elements may be used as indicators of relevance to agreed National Minimum Data Sets used to calculate and present national operational statistics.

Individual patients may be diagnosed with a condition which then becomes the reason for them to access the health system and receive a service. Diagnosis is a commonly used term. It is the foundation for the identification of health trends and statistics globally. It refers to any one of the diagnosis as listed in the International Classification of Diseases (ICD) [8]. Such classifications do change over time. They are published according to version number. Any classification system requires the application of rules and the use of several data elements to enable coders to make a judgment to arrive at a valid code. The interpretation and application of such rules need to be reliable to enable meaningful comparisons of findings to be made. These ICD codes plus other data may be used to form the basis for the application of a grouping algorithm to reflect estimated average resources used. The resulting groups are known by various names such as Diagnosis Related Groups (DRGs) or Health Resource Groups (HRGs).

In summary, metadata describing data used in accordance with minimum data set collections and used for statistical reporting are known as data standards that need to be governed. This same principle regarding data standards and governance applies to clinical specialties interested in collecting data about, for example, specific disease indicators, or, organizations wishing to collect any type of data describing care recipient characteristics.