Reimbursement Models

• In the context of this paper, reimbursement methods involve analytically calculated payments. These payments are incentive-driven for delivering health value. (see IMPACT approach). They are not merely covering bills (Fee-for-Service, FFS) submitted by healthcare providers. However, different methods, including FFS, can be combined when appropriate.
• When detailed medical data is available, both claim-dependent and claim-independent reimbursement methods can be developed. This includes claims data or, preferably, original clinical data.
• Methods for healthcare remuneration must be addressed (i.e., calculated, planned, and priced) simultaneously and holistically. Otherwise, there is a risk of inadvertently driving cost increases rather than optimizing them.
• Clear agreements and realistic (yet ambitious) expectations are essential for treatment cases and individual stages of the care pathway. Depending on the funding method, they also apply to health outcomes of specific patient groups. These expectations make it possible to use them as incentives in funding models. Additionally, it becomes feasible for services along a single patient’s journey to be provided by different service providers, including private healthcare institutions. Furthermore, why not combine services within care pathways that are funded by different insurance providers?

This section does not provide an exhaustive review of reimbursement methods in healthcare. It briefly introduces key approaches relevant to the current paper’s objectives. For the purposes of this discussion, I classify reimbursement techniques—which are neither mutually exclusive nor entirely inclusive of one another—as follows:

Fee-for-Service (FFS):
This is the most prevalent reimbursement method and is utilized in both inpatient and outpatient care. Payments are made for each individual service rendered.

Diagnosis-Related Grouping (DRG):
Typically used in inpatient settings, this method classifies cases into groups based on diagnosis. DRGs are generally limited to a single case (e.g., one hospital visit or one medical claim) and are determined primarily by the patient’s diagnosis, with additional factors such as surgical procedures and hospital stay length playing a role.

Bundled Payment:
Also referred to as value-based payment, episode-based payment, integrated payment, or care package payment model, this approach involves reimbursement for a medical case. It may span across multiple specialties. It can encompass more than one claim. While it is often limited to a single case (e.g., an episode of care for a specific condition), in cases of chronic illness, it may be defined by a specified time period. Bundled payments could include result-based expectations but need to be accompanied by a single responsible party overseeing the entirety of the care bundle.

Payment Independent from Detailed Visit Data:
This category includes funding methods such as readiness fees for hospitals or departments. Capitation fees are also included in general practitioner systems. Various performance-based incentives are measured by health indicators of patients on a general practitioner’s list. These methods do not rely on visit data but clinical outcome.

A care pathway does not necessarily determine whether a bundled payment approach is utilized.

In the context of this paper, it is essential to differentiate bundled payment (a reimbursement model) from a care pathway (also referred to as a patient journey, treatment pathway, clinical pathway, or healthcare pathway).

A patient care pathway is essentially a concept, whether expressed as treatment guidelines or derived from the academic and practical medical knowledge of physicians. It defines the medical services, as well as their sequence and quantity, required to achieve the best health outcomes for a patient.

In the context of health insurance and healthcare statistics, care pathways can be coded using medical claims data (clinical data), which allows them to be tracked, studied, and evaluated. This evaluation can be based on patient case mix parameters (input) and the outcomes achieved through the treatment (output).

Finally, if all the necessary conditions are met, a reimbursement model that incentivizes better outcomes can be added. This implies that a patient care pathway could serve as the foundation for developing reimbursement models. See Figure 1 for further clarification.

Why all reimbursement methods need to be developed and managed simultaneously?

The implementation of bundled payment in scenarios where DRG and FFS payment models are also applied raises several challenges. These challenges often lead to hidden or masked (and therefore unevaluatable) costs, or even an increase in costs.

The reason is straightforward: it is impossible to foresee all individual cases. This can result in scenarios where a single patient visit is counted as part of multiple bundles. This leads to increased costs. There can also be situations where it is unclear to which bundle a claim belongs. Such ambiguities can impact statistical surveillance and lead to inaccurate treatment cost calculations. FFS and DRG payments are prioritized and processed first. As a result, bundled payment is treated as a secondary or supplementary reimbursement model. This makes tracking actual costs and associating them with outcomes extremely difficult or even impossible.

When defining bundles—essentially determining which medical claims or visits are included in or excluded from a bundle—certain challenges arise. These challenges can be categorized as within-diagnosis pathway issues and between-diagnosis pathways issues (see Figure 2). Let us consider a hypothetical patient with two parallel health cases. In case 1, breast cancer is diagnosed and treated. Case 3 involves an underlying mental health issue. The patient sees a psychiatrist for depression. At some point, case 1 (breast cancer) impacts the patient’s mental health negatively. This impact results in dual reasons for visiting the psychiatrist. From the perspective of medical claim statistics, it becomes necessary to determine which bundle this claim should belong to.

This would be straightforward if the services provided were solely related to depression. It would also be easier if there were similar cases in the database. In such a situation, we could simply divide the services equally between the bundles. However, if other patients have different combinations—such as depression, schizophrenia, and hypertension instead of breast cancer—the new combination might not be common. Patients might have a mix of case 1 (breast cancer) and case 3 (depression) that is not typical. This combination might not be common enough for clear categorization. Conversely, when mental health effects resulting from a cancer diagnosis (case 1) are grouped under general depression services (case 3), the specific mental health impact of cancer is at risk of being underestimated. As a result, it becomes difficult to allocate the claim to the correct bundle or create meaningful statistical groupings.

In addition to the between-diagnosis pathways issues, challenges also arise within-diagnosis pathways. For example, if a patient needs treatment for both breasts, this includes case 1 and case 2. Claims might also arise when different diseases affect the same organ and require treatment by the same physician. Such claims might appear as part of a single case but are, in fact, independent. Similarly, when medical claims for oncologist visits carry diagnoses for diseases affecting both breasts, distinguishing between the two cases becomes challenging, as they technically represent separate conditions (e.g., a separate breast or a separate cancer type).

The key solution here lies in adopting a simultaneous and hierarchical approach to reimbursement modeling (see next paragraph).

Figure 2. Within-Diagnosis and Between-Diagnosis Pathway Issues in Applying Bundled Payment. This figure provides an example of a single patient who may have two parallel diseases (Case 1 and Case 3) or develop consequent health issues coded under the same diagnosis (Case 1 and Case 2). In the case of parallel diseases, multiple conditions coexist (e.g., depression and breast cancer). In the case of consequent diseases, the subsequent conditions are still distinct cases, even if they fall under the same diagnosis. These distinctions can arise when the same organ is affected by different diseases or when parallel organs, such as the breasts, are involved. These complexities make it challenging to classify claims accurately according to cases (Visit numbers?). If claims are categorized based on predefined agreements, this may lead to biased statistical management and misinterpretation in clinical evaluations. In addition, using different reimbursement methods that are still temporally duplicated (such as initial funding based on FFS or DRG rules, followed by post-payment reimbursement based on bundled payment rules) reduces transparency between the costs incurred and the outcomes achieved. Additionally, it may inadvertently lead to an increase in overall healthcare expenditures. This is despite the fact that both DRG and bundled payment models were introduced with the aim of optimizing healthcare funding.

There is reimbursement models based on claims and those that are independent of claims.

Reimbursement independent of claims

This can be seen as result-oriented motivational rewards. These models can be monitored using KPIs or quality parameters, ensuring that the doctors involved (either directly or indirectly) are rewarded accordingly. A good example would be incentive payments to family doctors based on the overall coverage of childhood calendar vaccinations within their practice.

Claim-dependent methods

These methods need to be adopted while simultaneously considering a hierarchical approach. The basis of modeling involves collecting all claims within a specific period (e.g., a year). Essentially, this means that each medical claim must be labeled according to the appropriate payment method.

1 – Care Pathway-Based Bundled Payment

The bundled payment funding model is reasonable to apply to diseases that:

1. Have treatment outcomes that are worse than desired,
2. Present a significant logistical challenge for both patients and healthcare systems,
3. Allow for the application of outcome-based differentiation in funding,
4. Carry significant costs or burden on the population if either untreated or treated,
5. Fulfill the conditions of bundled payment (including identifying a responsible party for the care pathway).

An important distinction here is that parameters such as treatment stages and associated specialties are based on input from specialists. The duration of each stage and their recurrence are also based on specialist input. Naturally, this requires statistical analysis of the current situation and a realistic definition of the desired outcomes. That said, such care pathways enable healthcare services to be “nudged” in the desired direction.

Defining such care pathways is a labor-intensive and analytics-heavy task. This makes it essential to apply these funding models based on specific needs.

2 – Mini-Care Pathway or a DRG-like Approach

This category includes the costs of treatments (within a year) that remain after the tagging of claims at the previous (higher) funding level. Like a care pathway, the treatment invoices under a single medical case are combined. These invoices fall within the same diagnosis and are unified into a single funding stream. However, unlike today’s DRG-based funding, the diagnosis-related funding would not depend on the number of treatment claims. It would not require inpatient treatment invoices.

As a side note, this method recognizes that the number of claims is not always directly linked to the disease. The relationship is not always straightforward. Some service providers deliver one visit and one claim. Others need several visits for similar clinical cases. Whether treatment occurs in an inpatient or outpatient setting can vary regionally or due to other specific circumstances.

Unlike today’s DRG model and the first-level funding model (Care Pathway-Based Bundled Payment), this method creates diagnosis-based groups using historical data. Machine learning-based grouping methods are employed. This allows consideration of far more parameters for grouping than humans can handle. Humans are responsible for directing the model through input parameters.

The mini-care pathway funding model is reasonable to apply to diseases that:

1. Involve shorter care pathways (simpler in terms of duration and stages) but still require sufficient funding attention.
2. Correspond to pathways that, due to resource limitations, are not defined under the first funding method.
3. Exhibit regional variation in clinical practice or coding.
4. Have inexplicably high treatment costs.

3 – FFS – Fee-for-Service

Treatment invoices (cases) do not qualify for the two previous funding levels. They cannot be addressed within a given time period. These invoices would fall under FFS reimbursement.

Overall:

• Funding levels are dynamic. This implies that the proportions of different diseases reimbursed at each level can vary. These variations depend on changes in resources and healthcare expectations.
• On the other hand, a key factor as described in the article on the IMPACT model. It is the independent, multifaceted, and high-quality input required for the development of these models.