This review of the variations among countries was based on the results of the two methods in the various countries. Tentative explanations for these differences include:

• There is an obvious positive link between the performance of NDP and the economic development, although this does not apply for all the aspects of the policy.

• A number of technical components when implemented adequately make a great difference in terms of output: for instance a bad selection of drugs and an inefficient procurement system lead invariably to shortages of drugs; on the contrary, a good registration system has a positive impact on rational use of drugs (Sri Lanka), a local manufacturing industry seems to have a positive influence on prices of drugs as long as the country is big (India). It was agreed that this last aspect should be analysed more in detail and followed carefully in view of the GATT agreements which may reverse the picture. A well-thought drug financing system can influence prescribing patterns. However, in all cases, supervision and monitoring are necessary.

• Countries with no local industry (national or multinational) can implement the main aspects of NDP with less problems (Sri Lanka).

• Compulsory use of an EDL in the public sector is just a first step in improving availability and affordability. These objectives seem more difficult to achieve in countries where pharmaceutical market is mostly private and where there is no regulation of drug prices (the Philippines).

• The geographical situation of countries is also an important factor in the difference in performance, e.g.: to be an island is an advantage as it makes drug imports easier to control (Sri Lanka).

• A few motivated people can make a difference between success and failure mainly to get things started but to rely on only a few people is dangerous as it does not ensure sustainability (Zimbabwe).

3.4 Policy innovations which can improve implementation of NDP

Policy innovations identified by the participants were related to the technical components and to the policy process. Under technical components, it seemed for instance that the creation of a social security system led to improved outputs of the policy (Colombia), the presence of a reimbursement system seemed to improve rational use of drugs as did some legislative requirements (e.g. in Sri Lanka, the antidiarrhoeal drugs are forbidden and therefore none of the children in the survey received them).

In terms of policy process, it seemed that improvements of the NDP through radical changes are more likely to happen when there were political windows of opportunity (e.g. new regimes: the Philippines), changes in the global political/economic environment (GATT, global economic liberalization: India), etc. When such opportunities present themselves, it is possible to implement quick, radical and widespread changes (Philippines). When such opportunities are not available, it could be better to use low profile, step by step strategies (Sri Lanka).

Successful implementation of the policy is also influenced by the planning process: strategies used for one stage of policy-making affect the probability of success of other stages e.g. strategies in the formulation stage affect the probability of implementation; therefore planning for implementation is important even at the time of the formulation of policy.

It was stated by the participants that an important lesson of the research at this stage was that the technical soundness and/or economic rationality of the policy did not always imply that the policy was politically viable.