Contracting for Health Services with Unmonitored Quality

In both the NHS and Medicare, recent emphasis has been on contracts with payment based only on the ntumber of patients treated. It is shown that, withouit direct monitoring of quality or effort to reduce costs, stuch contracts are efficient only when it is efficient to treat all patients wanting treatment. It may not be when treatment costs are insured or stibsidised. Such contracts can then be improved by incltuding payments for the number of patients wanting treatment, as well as for the number actually treated. Even then, the otutcome will not generally be efficient if quality is multi-dimensional. In purchasing health services, government agencies, insurers and health maintenance organisations (HMOs) want to keep costs down without inducing providers of services to skimp on quality. By quality, we mean any aspect of service that benefits patients, whether during the process of treatment or in the health outcome after treatment. Typically, neither the provider's effort to reduce costs, nor the quality of service, is easily described in a way that makes specification in a contract straightforward, so the provider must be induced to supply both effort and quality. Contracting is thus what Holmstrom and Milgrom (1991) term a multitask agency problem. In both the British National Health Service (NHS) and US Medicare, recent emphasis has been on using contracts with payment based only on the number of patients treated (price-quantity schedules) for each specified medical condition and not on such additional information as the actual cost of treatment. Under Medicare, payment is (with certain exceptions) a fixed price per patient in each diagnosis related group (DRG). In the NHS, purchasers are encouraged to use fixed price contracts (known as cost per case contracts) or contracts with payment a non-linear function of the number of patients treated (cost and volume or sophisticated block contracts). National Audit Office (1995) summarises