In developed countries, the epidemiological shift in disease burden from acute to chronic diseases over the past 50 years has rendered acute care models of health service delivery inadequate to address the health needs of the population. In developing countries this shift is occurring at a much faster rate.

The health care delivery system has the potential to affect patients' adherence behaviour. Health care systems control access to care. For example, health systems control providers' schedules, length of appointments, allocation of resources, fee structures, communication and information systems, and organizational priorities. The following are examples of the ways in which systems influence patients' behaviour:

• Systems direct appointment length, and providers report that their schedules do not allow time to adequately address adherence behaviour (66).

• Systems determine fee structures, and in many systems (e.g. fee-for-service) the lack of financial reimbursement for patient counselling and education seriously threatens adherence-focused interventions.

• Systems allocate resources in a way that may result in high stress and increased demands upon providers which, in turn, have been associated with decreased adherence in their patients (67).

• Systems determine continuity of care. Patients demonstrate better adherence behaviour when they receive care from the same provider over time (68).

• Systems direct information sharing. The ability of clinics and pharmacies to share information on patients' behaviour regarding prescription refills has the potential to improve adherence.

• Systems determine the level of communication with patients. Ongoing communication efforts (e.g. telephone contacts) that keep the patient engaged in health care may be the simplest and most cost-effective strategy for improving adherence (69).

Few studies have evaluated programmes that have used such interventions, and this is a serious gap in the applied knowledge base. For an intervention to be truly multi-level, systemic barriers must be included. Unless variables such as these are addressed, it would be expected that the impact of the efforts of providers and patients would be limited by the external constraints.

The changing nature of disease prevalence also influences activities at the system level. Continued reliance on acute models has delayed the reforms necessary to address longer-term interventions for chronic conditions. In developing countries this shift is occurring at a much faster rate at a time when the battle against communicable diseases is still being fought.

In some countries, the attention of the policy-makers may remain focused on communicable diseases, for example HIV/AIDS and tuberculosis. However, these diseases are not effectively addressed by the acute care model. Even if it were to provide full and unrestricted access to appropriate drugs, the acute care model would lack impact because it does not address the broad determinants of adherence.