How absolute is zero? An evaluation of historical and current definitions of malaria elimination

Abstract

Decisions to eliminate malaria from all or part of a country involve a complex set of factors, and this complexity is compounded by ambiguity surrounding some of the key terminology, most notably "control" and "elimination." It is impossible to forecast resource and operational requirements accurately if endpoints have not been defined clearly, yet even during the Global Malaria Eradication Program, debate raged over the precise definition of "eradication." Analogous deliberations regarding the meaning of "elimination" and "control" are basically nonexistent today despite these terms' core importance to programme planning. To advance the contemporary debate about these issues, this paper presents a historical review of commonly used terms, including control, elimination, and eradication, to help contextualize current understanding of these concepts. The review has been supported by analysis of the underlying mathematical concepts on which these definitions are based through simple branching process models that describe the proliferation of malaria cases following importation. Through this analysis, the importance of pragmatic definitions that are useful for providing malaria control and elimination programmes with a practical set of strategic milestones is emphasized, and it is argued that current conceptions of elimination in particular fail to achieve these requirements. To provide all countries with precise targets, new conceptual definitions are suggested to more precisely describe the old goals of "control" - here more exactly named "controlled low-endemic malaria" - and "elimination." Additionally, it is argued that a third state, called "controlled non-endemic malaria," is required to describe the epidemiological condition in which endemic transmission has been interrupted, but malaria resulting from onwards transmission from imported infections continues to occur at a sufficiently high level that elimination has not been achieved. Finally, guidelines are discussed for deriving the separate operational definitions and metrics that will be required to make these concepts relevant, measurable, and achievable for a particular environment.

Figure 1

Branching process diagrams for three definitions of elimination. The Poisson probability of each branch permitted under three elimination definitions and the total probability of all other unacceptable outcomes are depicted for three R_C values.

Figure 2

Maximum R_C at which the probability of failing to meet elimination criteria of fewer than three epidemiologically-linked local cases in three consecutive years is kept below risk thresholds. Thresholds of 1% (blue), 5% (red), 25% (green), or 50% (yellow) risk of failure are depicted, assuming the number of cases resulting from each case follows a Poisson distribution with mean = variance = R_C.

Figure 3

Approximate number of locally-acquired cases (both introduced and indigenous) expected to result from a given number of imported cases under a particular level of R_C.