Paying for Mental Health Care in Private Health Insurance in the Netherlands: Some Lessons for the United States

As part of national policy under the Health Insurance Act (HIA), the Dutch transferred responsibility for much of mental health care to private health insurers, including adult short-term (2008) and long-term (2015) care (2). Presently, private insurers pay for all outpatient and inpatient care except for inpatient care beyond a duration of 3 years. After 3 consecutive years of hospitalization, responsibility for costs shifts to the government. All mental health care for children (age ≤17) is paid for by local governments for up to 3 years, after which responsibility shifts to the central government; this care is not part of private health insurance. In 2020, mental health care is expected to amount to about 10% of total health care costs of private insurers under the mandatory benefits package.

The Dutch use a risk adjustment system whereby insurers are paid more or less for individuals whose care is predicted to be high or low cost, respectively. Largely for historical reasons, the Dutch developed a separate risk-adjustment formula for mental health care, in addition to the formula for general medical care. The Dutch risk-adjustment formula for mental health care uses a wider range of risk adjusters than used in the United States. The current mental health risk adjustment formula includes age, gender, geography, socioeconomic status, household size, source of income, pharmacy-based cost groups for mental illnesses, and variables based on multiple-year high spending on mental health care. Psychiatric diagnoses in the model include eating disorders, bipolar disorders, schizophrenia, addiction, pervasive developmental disorders, and personality disorders, with multiple indicators for each diagnosis according to the past length and site of treatment (https://www.rijksoverheid.nl/documenten/rapporten/2018/11/30/aanpassen-dkgs-psychische-aandoeningen-voor-de-risicoverevening). The U.S. marketplace model includes 10 categories related to mental health and substance use disorders: drug psychosis, drug dependence, schizophrenia, major depressive and bipolar disorders, reactive and unspecified psychosis, delusional disorders, personality disorders, anorexia and bulimia nervosa, autistic disorder, and pervasive developmental disorders except autistic disorder (3).

The Dutch phased in the shift of financial responsibility for mental health care to private insurers. In the first 2 years after the transfer, because of uncertainty about the financial effects of the reform, the shift in responsibility was in name only, with insurers receiving full cost-based compensation for all mental health spending. As experience accumulated and the risk-adjustment model for mental health care improved, more financial risk was transferred to insurers. Starting in 2010, insurers assumed responsibility for individual-level spending up to a threshold that was gradually increased over time. In 2020, insurers are responsible for about 97% of total spending on mental health care. Risk sharing in the form of individual-level cost-based compensations applies to 75% of insurer spending above a threshold value of about €90,000. The main policy benefit of cost-based compensation is to reduce insurers’ exposure to cost risk and to mitigate incentives to discriminate against persons with severe mental disorders. The main policy cost of cost-based compensation is dilution of incentives for cost control. By focusing risk sharing on only severely ill patients with the highest costs, the Dutch hope to have struck a good balance of these policy benefits and costs.

First, and at a high level, the Dutch have shown that responsibility for most mental health care can be shifted to private insurers when supported by a payment system that uses extensive risk adjustment for diagnoses, social factors, and prior spending, as well as some cost-based compensation for severely ill patients with very high costs. A distinctive feature of Dutch insurer payment is the use of a separate formula for mental health care. Ideas from the Dutch could still be imported into U.S. payment models generally designed to pay for all areas of insurer spending.

Second, in their risk adjustment formula for mental health care, the Dutch make extensive use of social factors (such as poverty level and household composition) and prior-use indicators that are likely to be particularly important when paying for recurring costs of mental health care. There has been some interest in the United States in incorporating social factors into Medicaid managed-care payment models, but this effort could be stepped up and extended to other insurance markets. The United States has shied away from introducing prior-use indicators into payment formulas, partly on grounds of fear of diluting incentives to limit care and partly for the simple practical reason that, in contrast to the universal health insurance system in the Netherlands, individual health insurance markets in the United States are subject to considerable “churn.” The feasibility of prior-use variables (e.g., two or more hospitalizations in the prior year) is increasing, however, as data integration across payers improves. Incorporation of these simple and powerful indicators is worth considering.

Third, high-cost risk sharing is an effective complement to risk adjustment. Mental health spending is highly skewed in a population, with a relatively small share of the population incurring any expenses, and a very few very high-cost spenders accounting for a large share of the costs. Risk adjustment models oriented to predicting averages are bound to leave the very high spenders with chronic conditions undercompensated. The Dutch have employed high-cost risk sharing to direct more funds to these severely ill individuals. A form of high-cost risk sharing is used in individual health insurance markets in Switzerland and will soon be reintroduced in German health insurance markets (4). In the U.S. marketplaces, high-cost risk sharing pays 60% of costs above a threshold of $1 million, 10 times as high as the Dutch threshold and too high to have much effect. Medicare Advantage makes no use of risk sharing. Both of these individual health insurance markets might see improvements in insurer incentives from targeted risk sharing not just for expensive and chronically ill individuals with mental illness, but for anyone with an expensive and chronic illness.