Pay-for-performance, also known as value-based purchasing, is a payment model that offers a financial incentive if certain criteria are met by a behavioral healthcare provider. The goal is to motivate providers to improve the care they provide, and proponents of pay-for-performance predict that improved care will reduce the need for future services, thus reducing both costs and demand for services.
Published studies have been unable to find any direct effect of pay-for-performance incentives on the quality metrics being measured. Some studies have found an improvement in the metric being measured; however, in each case the researchers were unable to prove the improvement in the metric was a direct and specific result of the pay-for-performance incentives. For example, in these studies, the provider in question began a variety of quality improvement efforts near the same time as the implementation of the pay-for-performance model, such as public reporting of quality data and other types of continuous improvement plans.
There are a number of factors preventing the pay-for-performance models from being more effective in improving the quality of care in behavioral healthcare. The most important of these factors are the lack of clear metrics for behavioral health, suboptimization, factors beyond the provider’s control, and mismatched incentives.
No Clear Metrics for Behavioral Healthcare
Pay-for-performance models vary greatly depending on the funding source. There are more than 150 different types of pay-for-performance models being used by different companies. The possible variations includes the total number of metrics measured, types of metrics, the goals for each of the metrics, and how both the baseline and improvement are measured for each metric.
In a medical healthcare setting, the positive outcome of treatment is the elimination or proper maintenance of the medical condition. The successful outcome of treatment for a broken arm is a healed arm with the same level of use prior to the injury. It is easy to determine how long successful treatment of a broken arm should last and how many visits the patient needs for that successful treatment.
Unlike medical healthcare, behavior healthcare has no standardized agreed-upon metrics, and the positive outcome of treatment may not be the removal of the mental health condition. (The successful outcome of treatment for schizophrenia, for example, is not the absence of schizophrenia at the end of treatment.) Not having a clear definition of successful treatment makes creating universal outcome metrics extremely difficult. In order to have more effective pay-for-performance models, patients, providers, and funding sources must all agree on reasonable definitions of successful outcome metrics for different mental health conditions.
Suboptimization is the concept of focusing on one part of a system rather than improving the whole system. One point that is often overlooked in pay-for-performance models is that the focus of improvement is on the few metrics being measured rather than improving the total quality of care provided or the patient’s quality of life. Like a car’s speedometer, a metric provides very useful—yet at the same time limited—information. The speedometer provides the driver with the information of how fast a car is moving. But knowing a car is traveling at 45 miles per hour without knowing the context of the speed is limited information. Is the speed limit 35 miles per hour? If that is case, the car is speeding. Is there a heavy snowstorm outside of the car? In that case, the car is going too fast for the weather conditions, regardless of the speed limit. In the same way, knowing a pay-for-performance metric without knowing the context of the metric’s impact on a patient’s quality of care or quality of life is not very useful.
One often-used metric in pay-for-performance models is how many patients discharged from a psychiatric hospital were able to have an office visit with a prescriber within seven days. The goal of seeing a prescriber within seven days to continue care is helpful to improving the patient’s quality of life. But when the seven-day follow-up is the onlymetric being measured, or when the metric is given extra weight in the pay- for-performance model, it can create problems. A health system might feel pressure to prioritize an appointment for a recently discharged patient from its psychiatric unit over one for an existing patient receiving outpatient services, or to make an appointment with an outpatient provider that has an opening within seven days rather than an outpatient provider which best fits the patient’s needs. By only focusing on the goal of making an appointment, other factors that may improve the quality of that patient’s care may be given lower priority. In order for pay-for-performance models to be more effective, any metric needs to be viewed in the totality of how the metric improves the quality of life and quality of care for the patient.
Factors Beyond the Provider’s Control
A number of pay-for-performance models have metrics that are not fully within the provider’s control to successfully achieve. A metric may have tasks which need to be completed by the patient or other providers in order to meet the goal. A significant factor that a provider cannot override is a patient’s right to refuse treatment, despite what the provider may believe is best for the patient. For example, the majority of pay-for-performance plans assume a person will want to have a follow-up appointment, but every patient has the right to choose if they want aftercare services after leaving a psychiatric hospital. A patient discharged from a psychiatric hospital may choose to follow up with his or her primary care physician for either medication monitoring or for a referral to a psychiatrist. Depending upon the pay-for-performance model, that appointment may not count towards fulfilling the metric. If a patient chooses not to continue treatment in an outpatient setting, a provider should not be penalized under pay-for-performance.
There are other logistical and financial factors which are beyond a provider’s control:
- A follow-up appointment could be changed by the outpatient provider after the patient is discharged from the hospital.
- A patient may not have the transportation resources to make a scheduled appointment.
- Weather conditions such as a snowstorm may cause an appointment to be rescheduled outside the seven-day window.
- A patient may need to choose between paying for shelter or food rather than medications or an appointment.
In order for the pay-for-performance model to be more effective, a provider must believe the metric or goal is within the provider’s ability to meet or exceed.
Incentives Are Too Small
A problem exists when the pay-for-performance incentives are not large enough to help providers make major changes. The majority of incentives range from 0% to 2% of the amount of services billed, and these incentives are only paid if the goal for the metric is reached. The cost to a provider to improve a metric comes from three key factors, including the costs associated with collecting and analyzing data, hiring an expert to help improve the metric, and the cost of any new equipment or additional staff to perform any new processes or tasks. If the total cost to improve is greater than any possible incentive, the result is a financial loss for the provider, which acts as a barrier to quality improvement. In order for the pay-for-performance models to be more effective, incentives need to either equal or exceed the costs to improve the metrics.
Incentives Are Too Large
One study found that patients presenting with higher risks were more likely to have difficulty accessing treatment, thus excluding them from pay-for-performance programs.A second study found clinicians in a pay-for-performance program would avoid treating more acute patients as a way to improve the clinicians’ performance ranking. The worst-case scenario is one where a provider excludes all the high-risk or hard-to-treat patients in order to make the most possible profit.
In order for pay-for-performance programs to be more effective, the incentives have to cover the cost of improvement but not at a level where the incentive motivates a provider to exclude a patient in need of services.
Pay-for-performance models in behavioral healthcare have great potential to improve the quality of life for people. Making changes to how current metrics are measured and linked to incentives, along with changes to the amount of possible incentives, will only improve the effect of pay-for-performance on the quality of care in behavioral healthcare.