Health service coverage and its evaluation

Bulletin of the World Health Organization, 56 (2): 295-303 (1978) Health service coverage and its evaluation T. TANAHASHI 1 Health service coverage is considered as a concept expressing the extent of interaction between the service and the peoplefor whom it is intended, this interaction not being limited to a particular aspect ofservice provision but ranging over the whole processfrom resource allocation to achievement ofthe desired objective. For the measurement ofcoverage, several key stages are first identified, each of them involving the realization of an important condition for providing the service; a coverage measure is then defined for each stage, namely the ratio between the number ofpeople for whom the condition is met and the target population, so that a set ofthese measures represents the interaction between the service and the target population. This definition of coverage allows for variations, which are called "specific coverage ", by limiting the target population to specific subgroups differentiated by certain conditions related to service provision or by demographic or socioeconomic factors. The evaluation of coverage on the basis of these concepts enables management to identify bottlenecks in the operation of the service, to analyse the constraining factors responsible for such bottlenecks, and to select effective measures for service development. In many developing countries high priority has been given to the development of basic health ser- vices or primary health care. This involves two basic approaches: one is the development of new resources and technologies; the other is the effective use of available resources and technologies. In the context of the latter approach, a great deal of attention has recently been given to the management of health services. The fundamental issues in the management of a basic health service are: (a) How should resources be allocated and the service organized in order to serve as many people as possible? (b) Is the service reaching the people it should serve ? (c) Has the service been effective in meeting the people's needs? Although knowledge of the service's coverage is essential for answering these questions, the evalua- tion of coverage has not been common practice. There seem to be three major reasons for this. First, there has been some confusion about the concept of coverage, which has been measured and inter- 1 Scientist, Division of Strengthening of Health Services, World Health Organization, 1211 Geneva, 27, Switzerland. preted differently for various purposes and occa- sions. Secondly, the measurement of coverage invari- ably requires an assessment of " population ", with which few service personnel have been concerned. Thirdly, the use of information on coverage has rarely been considered in the planning or manage- ment of health services. The purpose of this paper is to: (i) re-examine and clarify the concept of health service coverage; (ii) propose an approach to the evaluation of coverage; (iii) illustrate some uses of information on cover- age in service management. A RE-EXAMINATION OF HEALTH SERVICE COVERAGE Comprehensive view ofhealth service coverage Health service coverage depends on the ability of a health service to interact with the people who should benefit from it (the target population), i.e., the abil- ity to transform the intention to serve people into a successful intervention for their health. This trans- formation process involves a variety of factors, such as availability of resources and manpower, distribu- tion of facilities, supply logistics, and people's atti- 3691 - 295- T. TANAHASHI r-- - ",____- n I I TARGET POPULATION ( SERVICE TARGET ) I I I POPULATION I I I a-.~ ~ ~~~~ - _ J Fig. 1. Schematic model of health service coverage and utilization. tudes to health and health care, to name just a few. It is impossible to observe the whole of such a many- sided process and evaluate it in every detail, but it is possible to observe the number of people for whom the service has satisfied certain criteria relating to its intended health intervention, and to compare that number with the target population. This has given rise to the concept of coverage and its evaluation. Coverage is normally expressed by the proportion of the target population who can receive, or have received, the service. The number of people for whom the service can be provided expresses the service capacity and indicates the potential of the service. On the other hand, the number of people who have received the service expresses the service output and indicates the actual performance of the service. We may, therefore, define the coverage related to service capacity as potential coverage and that related to service output as actual coverage. The relationship between service capacity and out- put is another important aspect of health service, which is called utilization or service utilization. It is normally expressed as the ratio between output and capacity, or as the rate of output, assuming the capacity of the service to be known. There is occa- sionally some confusion between the terms " cover- age " and " utilization ", and a clear distinction should be made. Utilization refers only to the ser- vice, and its measurement is only indirectly related to the size of the target population; on the other hand, coverage expresses a relationship between the service and the target population. For example, a high utilization of service facilities does not neces- sarily imply satisfactory coverage and could in fact imply the contrary. The above definitions of coverage and utilization are illustrated in Fig. 1, where each arrow represents the ratio of the component at its end to the compo- nent at its root. As the relationships between the three arrows indicate, actual coverage can be derived from the combination of potential coverage and utili- zation. The section with a broken outline represents the total population, the target population being nor- mally a subgroup within the total population. Classification ofmeasurements of coverage There are a number of ways of describing the capacity and output of a service; hence there are a number of ways of measuring coverage. It is unlikely that a single measurement of coverage could satisfac- torily reflect the complex interaction between the health service and the target population. In order to identify the measurements of coverage that reflect essential requirements for service provi- sion, let us imagine the process whereby a person in need of a certain kind of health care obtains the appropriate service. First, he looks for a service in his area that is relevant to his problem; when he has found there is one, he can use it only if he has the means of reaching it; whether he can afford it and decides to use it is another matter; if he does, then he receives the service; but the service may or may not solve his problem, depending on the quality of the service as well as the nature of his condition. Looking at this process from the point of view of service provision, it is possible to identify five impor- tant stages that successively lead to a desired health intervention and to define measurements of coverage appropriate to these stages. 1. First of all, some resources-manpower, facili- ties, drugs, etc.-are always required in order to provide a service; the availability of such resources limits the maximum capacity of the service, which in turn decides the amount of the service that can be made available to the target population. The ratio between this capacity and the size of the target population gives the measurement of coverage for this stage, and it may be called availability coverage. 2. Even if all the necessary resources are available, the service must be located within reasonable reach 296 EVALUATION OF HEALTH SERVICE COVERAGE of the people who should benefit from it. Meeting this condition can be considered as the next stage in the process of service provision; here, the capacity of the service is limited by the number of people who can reach and use it. The measurement of coverage based on this capacity may be called accessibility coverage. 3. Once the service is accessible, it still needs to be acceptable to the population, otherwise people may not come for it and may even seek alternative care. This " acceptability " may be influenced by such factors as the cost of the service to the user, the form of religion he follows, etc. If the service is accepted by the potential user, this is another step forward in the process of service provision. Here, service capa- city is limited by the number of people who are willing to use the accessible service, and the measure- ment of coverage based on this capacity is defined as acceptability coverage. 4. The next stage in the process of service provi- sion is the actual contact between the service pro- vider and the user. The number of people who have contacted the service is a measurement of service output; the ratio between this and the size of the target population gives a measurement of coverage that may be called contact coverage. 5. The contact between the service provider and the user does not always guarantee a successful intervention related to the user's health problem or an effective service. We can, therefore, consider another stage in the process of service provision where a service performance that is appraised as satis- factory by specific criteria is achieved. The number of people who have received satisfactory service is thus another measurement of service output, and the measurement of coverage based on this output is called effectiveness coverage. The above concept of coverage related to the process of service provision and the five measure- ments of coverage are illustrated in Fig. 2, while the classification of the measurements of coverage is summarized in Fig. 3 (A). It is worth noting from Fig. 2 that progress in the provision of a service means that the service must meet more requirements, hence the service satisfies the requirements of fewer people. The measurements of coverage correspond- ing to the five intermediate stages of the process follow the same trend (Fig. 3 B). The relationships between the different measurements of coverage are the key factors in the evaluation of coverage, which will be discussed later. GOAL OF SERVICE ACHIEVEMENT (5) EFFECTIVENESS COV. "People who receive effec- tive care" OPERATION CURVE (( CONTACT COVERAGE_____ / People who use service" / (3) ACCEPTABILITY COVERAGE "People who are willing to use service" (2) ACCESSIBILITY COVERAGE . "People who can use service" (1) AVAILABILITY COVERAGE People fOr whom service is aVailable" TARGET POPULATION NUMBER OF PEOPLE Fig. 2. Coverage diagram-illustrating relationships between the process of service provision and coverage measurements. AVAILABILITY COVERAGE ACCESSIBILITY COVERAGE ACCEPTABI LITY COVERAGE CONTACT COVERAGE EFFECTIVENESS COVERAGE IAVAILABILITY) IACCESSIBILIT'Y JACCEPTABIUTY) I CONTACT ? IEFFECTIVENESS)COVERAGE J j COVERAGE J j COVERAGE J > TCOVERAGE) _ ( COVERAGE RAG,.0 7780. Fig. 3. Classification and interrelationship of coverage measures. (A) Classification of coverage measurements; (B) Relationships between coverage measurements. Variations in the measurement of coverage The various measurements of coverage have been defined above for the different numerators in the coverage ratio, while keeping the denominator or target population the same. However, a particular subgroup of the target population can be chosen for the denominator, and the resulting measurement of COVERAGE POTENTIAL COVERAGE ACTUAL COVERAGE 297 CD .2 wC) A . uj0- ulw U- V,C) ui u CL. T. TANAHASHI (5) _ (4) _EF (5) ECTIVENES A (3) CNTACT ACCEPTABILITY TARGET POPULATION (3)-| WITH ACCESSIBILITY (2) ACCESSIBILITY _ PROVISION-SPECIFIC COVERAGE SUBJECT TO ACCESSIBILITY AVAILABILITY (1) TARGET POP. , TARGET POP. IN IN URBAN AREA RURAL AREA _ ---- \ (53 c 1 (5) D (4) _ _ _ 4 - (3)- (3)._ (2).------------\ I (2) -.-I-_--_ (1 ) --\( -\ TARGET POPULATION TARGET POPULATION IN URBAN AREA IN RURAL AREA POPULATION-SPECIFIC COVERAGES BASED ON URBNRuRAL SUBDIVISION Fig. 4. Illustrations of specific coverages. (A) Overall coverage; (B) Provision-specific coverage subject to accessibility; (C) Population-specific coverage in an urban area; (D) Population-specific coverage in a rural area. coverage will represent a different aspect of the interaction between the service and the population. In order to differentiate the coverage related to such a subgroup from that related to the whole target population, it may be called specific coverage. Two types of variation are important in coverage evaluation. One type arises from taking as the de- nominator a part of the target population for whom certain criteria related to service provision have been met. The actual coverage among people with ade- quate accessibility to the service is an example. Specific coverage of this type may be called provi- sion-specific. Provision-specific coverage is useful in estimating coverage or changes in coverage when the extent or methods of service provision have changed; its application will be discussed later. An example of provision-specific coverage is illus- trated in Fig. 4 (B), which concerns specific coverage subject to accessibility. A comparison between the coverage diagram (B) and the original diagram (A) shows that the former is a partial representation of the latter; this is because the denominator for provi- sion-specific coverage represents the part of the original target population that enjoys accessibility to the service. Another type of variation arises from the subdivi- sion of the target population by a factor unrelated to service provision. The differentiating factor can be chosen to suit specific purposes; subdivision by age or by urban and rural areas is a typical example. Specific coverage of this type may be called popula- tion-specific and it is useful in the analysis of inter- actions between service provision and factors affect- ing the target population. An illustration of population-specific cover- age is shown in Fig. 4 (C) and (D). The target population represented in the original coverage dia- gram (A) consists of the people from the urban area and those from the rural area in roughly equal proportions, and the portions of the coverage (or service) relating to these two groups are separated by the broken line. The population-specific coverage for each group (Fig. 4 C and D) is derived by expressing its share in proportion to the target population in the area concerned. The differences between the operation curves imply significant differences in ser- vice provision between the two populations. There are other specific coverages, but they all belong to one or other of the above two types or to a combination of them. An example of the combina- tion type is specific coverage subject to accessibility for rural populations. MEASUREMENT OF COVERAGE Description of services Fundamental to the measurement of coverage is the description, in observable or measurable terms, of the service whose coverage is to be measured. The description must at least answer the following ques- tions: (a) What is the aim of providing the service? (b) What does the service do? (c) For whom or what is the service intended? (d) What are the essential resources required for the service? (e) How much of the essential resources would a unit of service require? 298 EVALUATION OF HEALTH SERVICE COVERAGE (f) How can the performance of the service be measured? (g) What are the criteria for satisfactory perfor- mance of the service? We shall see in the following sections how the measurement of coverage depends on the answers to these questions. Unit ofmeasurement So far it has been assumed that coverage is expres- sed in terms of the service capacity or output as a proportion of the target population, and that their common unit of measurement is the person. This has been done simply to facilitate the discussion of the concept of coverage; the actual measurement demands more rigorous consideration. Here it is necessary to go back to the target population, which was defined earlier as the people for whom the service is intended. This definition is based on two simplifications, namely that the service is provided to individuals and that their number indicates need or demand of a population for the service. A more appropriate and general definition of the denominator of the coverage ratio is the amount of need as conceived by the service provider, which may be called the service target. The unit used to measure this amount can also be applied to the measurement of service capacity and output and should therefore be the unit of measurement for coverage. Most health services are intended for and pro- vided to people, and the person is still the unit most widely used for the measurement of coverage. There are, however, services whose targets are better expressed in other units. For example, sanitation activities are often directed to households; for ser- vices of this kind, the household may be a more appropriate unit of measurement. From the standpoint of needs, two types of health service can be identified; one is provided to meet the intrinsic need of the people, and the other type is provided in response to the incidental need or demand of the individual. An example of the former type is vaccination against a certain disease, say measles, the need for which does not depend on the incidental health condition of individuals. Most forms of medical care belong to the latter type, hospital care being a characteristic example. To differentiate between these two types of service, they may be called prevention-oriented and care-oriented, respectively. The service target of a prevention-oriented service is determined by the demographic characteristics of a population, and the unit of measurement is also demographic, such as the person or the household. The service target of a care-oriented service is deter- mined not only by demographic characteristics but also by epidemiological characteristics that affect the " prevalence rate " for service need. The appropriate unit of measurement for the care-oriented service is thus the episode or the case, and the service target may be several times greater than, or only a fraction of, the service target, depending on the episode frequency. Measurement in practice The measurement of the service target requires information on the population characteristics and the answers to the first three questions on page 298 will serve to indicate the information required. This may be available from existing sources such as census data, or obtainable from ad hoc surveys. The answers to the next two questions form a basis for determining the availability of service. It is easy to recognize key manpower and facilities as essential resources, but sometimes running costs or the supply of expendable items hamper the operation of a service. These should also be taken into con- sideration; otherwise, one may end up installing X-ray facilities without the electricity to run them. Records of service delivery constitute the most important source of information for assessing acces- sibility and acceptability, and even the records of similar services are often useful in this respect. If no relevant data are available a pilot study or research may be undertaken to obtain them. The answers to the last two questions-(f) and (g) above-are pertinent to the measurement of service output. In measuring service output it is necessary to use the same unit of measurement as for service target: for example, if the target of a health service is expressed in terms of households, then its service output for the measurement of contact cover- age must be expressed as the number of households to which the service has been rendered. The service output for the measurement of effectiveness coverage must similarly be expressed in terms of households, but in this case it is the households for which the service rendered meets the criteria of effectiveness. The criteria for effectiveness are provided from the answer to question (g). Simple criteria facilitate the measurement, and the more appropriate they are the more information can be gained on coverage. 299 T. TANAHASHI COVERAGE EVALUATION AND MANAGEMENT Identification of service bottlenecks Suppose that the five measurements of coverage are obtained. The stages in service provision have been defined in such a way that certain inequalities exist between them (Fig. 3 B). A large difference between an adjacent pair of the coverage measure- ment implies that, for a significant proportion of the target population, the service is failing to meet the requirements for progress in service provision. In other words, a large difference implies the exis- tence of a problem or a bottleneck in the service provision. In the coverage diagram, the bottleneck appears as a sharp shift of the operation curve to the left. Fig. 5 illustrates three examples of service bottle- necks, indicated by the shading on the operation curve. Example (A) shows bottlenecks in the avail- ability and accessibility of the service, which imply poor allocation and deployment of resources and facilities. Example (B) shows bottlenecks in accept- ability and contact. Low acceptability implies poor appreciation of service by the public; while the reason for inadequate coverage by contact can be a lack of public demand for the service, or failure to provide the service to some people because of faulty operation. Example (C) shows a bottleneck in effec- tiveness, which implies poor quality service. Analysis of constraints A bottleneck shows where the difficulty in service provision lies, but it does not pinpoint the factor responsible for the poor coverage. For example, poor quality service may be due to the use of ineffective drugs or incorrect administration of effec- tive drugs. A good knowledge of the health service and of the situation of the target population is thus required in order to analyse the constraining factors. Gathering the relevant information for the analy- sis is an integral part of coverage evaluation. Health personnel as well as clients and potential users are important sources of information, and consultation with them may very well be sufficient to permit the causes of bottlenecks to be identified. Supposing that existing knowledge or available information about the service is insufficient to iden- tify any particular cause for the bottleneck, other approaches are available. One approach is to compare the coverages of different areas. The chosen areas would differ in situation and processes of service provision with respect to the factor suspected to be responsible for the bottleneck. If a high corre- lation is found between the degree of coverage and circumstances involving the factor, the suspicion is likely to be justified. Another approach is to compare population-speci- fic coverages by subdividing the target population. In the subdivision, the factor that appears most likely to be responsible for the bottleneck is selected as the differentiating factor. If the right factor is chosen, there will be significant variation in the target-specific coverages. Yet another approach is to change conditions associated with the suspected factor experimentally, and compare coverage before and after the change. If a significant difference is observed, the suspicion may very well be justified. This does not exhaust the possible approaches to the analysis of contraints. Nor can any one approach be suggested as being better than others. The selec- tion of the best approach has to be made when and where such analysis is required, i.e., in the actual situation. -,1_________ (5) --- A I (4) --- I (3) -_--- I (1)~~~~~~~~~~~~~~~~~~~~~~~(2) - (1 ) - -----o PI (5) ____ (4) ---- I (3) - - - I (2) - . (1) __ I (5) =n I_ .1 (4) ----- I (2) -I (1 ) .__- Fig. 5. Examples of service operation and bottlenecks. (A) Bottlenecks in availability and accessibility of service because of poor allocation and deployment of resources and facilities; (B) Bottlenecks in acceptability of service because of poor appreciation of the service by the people; (C) Bottlenecks in effectiveness of service because of poor quality of the service. 300 EVALUATION OF HEALTH SERVICE COVERAGE Cost/effectiveness analysis for service development Once the factors constraining service development become known, proper remedial action needs to be taken. An important concern at this stage is the selection of an effective form of action, and cost/ effectiveness analysis provides a basic approach for this purpose. For such analysis it is essential to know the service output or actual coverage, particularly the effectiveness coverage, that is expected from the implementation of the remedial action. As shown in Fig. 2, the effective service is the consequence of successful service provision and the effectiveness coverage can only be assessed in conjunction with the other measurements of coverage. Therefore the cost/effectiveness analysis of an action for service development requires the assessment of its probable impact on the entire process of service provision or on coverage in general, which may not be an easy task. However, a remedial action normally changes only certain aspects of service provision, and it is possible to take advantage of this in assessing the prospective coverage. This is another application of the concept of specific coverage. To illustrate this, let us consider an expansion of service. The expansion increases the availability and accessibility of the service, but it may not change its acceptability or effectiveness. We can estimate the increases from the details of the resources and facili- ties involved in the expansion. If we assume that the people who gain access to the service by this expan- sion have similar health conditions and attitudes towards the service, we can expect the same coverage for them as for the people who already have access to the service (i.e., the same provision-specific cover- age subject to accessibility). Combining these two results we can calculate the expected coverage under the expansion. Fig. 6 (B) shows the new coverage diagram de- rived in this way. The unshaded part shows the estimated availability coverage and accessibility cov- erage (1' and 2'), and the shaded part the provision- specific coverage which includes the other measures (3), (4), and (5). As we assume the provision-specific coverage to remain the same after the expansion, the corresponding part (shaded area) of the operational curve in the new diagram is homologous in shape to the original diagram. To illustrate another application of provision-spe- cific coverage, let us consider the adaptation of a new service technology-another common action for service development. Unlike service expansion, this A B -- --------]I EFFECT. i I .~~~~~~ II!~ ~~~~~~~~~~~ i I ACCEPTABI ITY I I t/ACCESSIBILITY/ BLABI YI S g C <~~~~~~~~~~~~~~~~~~~ Fig. 6. Application of the concept of provision- specific coverage in assessing aspects of service de- velopment. (A) Coverage of existing service; (B) Expansion of the service-availability coverage and accessibility coverage change but provision-specific coverage suject to accessibility remains the same; (C) Adaptation of new technology-potential cover- age remains the same but actual coverage changes. action is likely to affect service personnel as well as the target population, bringing changes in the actual coverage and probably even in the acceptability of the service. These changes are difficult to assess without a trial estimate. To facilitate discussion, let us assume that the service with the new technology is provided through the same delivery channel as the existing service, and that between the two services there is no significant difference in resource requirements and availability or even in acceptability because of their apparent similarity in the eyes of the user. In brief, these conditions lead to the same service provision but only up to the stage of acceptability, and the same coverage may be expected with respect to the corre- sponding measures. Contact coverage and effective- ness coverage still need to be known for the assess- ment of coverage for the new service to be com- pleted. Hence, the experimental method of obtaining the two measurements of actual coverage is based on 301 T. TANAHASHI the people who have been users of the existing service. The method is efficient because the sample population is easier to identify and smaller in size than the target population as a whole. This trial estimate gives the provision-specific coverage subject to acceptability. By substituting this estimated cover- age for the corresponding part of the existing cover- age, we can calculate the expected coverage for the new service. Fig. 6 (C) is a diagram of the coverage in this case. As in Fig. 6 (B), the unshaded part shows the esti- mated coverage-in this case the coverage measure- ments (4') and (5')-and the shaded part shows the coverage by the existing service delivery system. From the diagram the number of people who reap the benefit of the new service can be estimated, and this permits a comparison of the benefits and costs and an evaluation of the merits of the action taken. These are examples of two typical applications of provision-specific coverage in the assessment of two typical forms of action for service development. For the assessment of other forms, other applications of specific coverage can be made. For example, if some actions involve changes in the target population, the application of population-specific coverage will prove useful. One such example may be service expansion specifically aimed at rural populations. As discussed above, the changes in coverage due to the expansion can be estimated by applying the concept of provision-specific coverage. We can improve the accurary of such estimates by applying it only to the population-specific coverage for the rural population concerned. TOWARDS THE DEVELOPMENT OF A COVERAGE EVALUATION SCHEME So far only a conceptual framework of cover- age evaluation has been discussed; some thoughts on the development of a coverage evaluation scheme may be in order. In brief, this requires three things: (a) information-demographic, epidemiological, and socioeconomic-on the population with which the service is concerned; (b) knowledge of the health problem that the service is intended to deal with and of the activities of the service; (c) ability to gather information on the operation of the service. Obviously, not every country or service system can meet these requirements: there may not be adequate census data for the information; there may not be enough experts with the knowledge, or there may not be an adequate service infrastructure with the ability. Under such-probably common-circum- stances, the evaluation scheme may first be devel- oped for a limited pilot area and later expanded as practical methods become established. This approach has many advantages. For cover- age evaluation, demographic and epidemiological information on the population is essential, but it is rarely readily available. The approach facilitates the gathering of such information by focusing attention on a population of manageable size. Sometimes, knowledge of the health problem and also of the ways in which the service intervenes in it needs to be gained from experience; a pilot operation gives an opportunity for this, thus facilitating selection of the appropriate target and coverage measures. If the coverage evaluation is meant for the service manage- ment, continuous gathering of information on the operation of the service is necessary; hence it is important to make this activity as simple and practi- cal as possible. For this, simple and appropriate indicators of the service operation must be identified and practical methods developed for gathering them. Again, a pilot operation will give an opportunity for doing so before the full implementation of the eval- uation scheme. ACKNOWLEDGEMENTS I should like to acknowledge stimulating discussions with several of my colleagues in WHO and useful comments, particularly from A. Rossi-Espagnet, R. Pibouleau, Y. Watanabe, G. Lavoipierre, S. Brogger, and E. Liisberg, on earlier drafts of this paper. 302 EVALUATION OF HEALTH SERVICE COVERAGE RtSUMJ2 LA COUVERTURE DE LA POPULATION PAR LES SERVICES DE SANTE ET SON EVALUATION La couverture de la population se definit, pour un ser- vice de sante, par 1'etendue de l'interaction constatee entre ce service et la population qu'il entend servir; cette interaction n'est pas seulement un aspect particulier de la prestation de services: elle englobe tout le processus allant de l'allocation de ressources A la realisation de l'objectif vise. Pour mesurer la couverture, il faut au prealable identifier les principales etapes aboutissant A cet objectif, etapes dont chacune correspond A la realisa- tion d'une condition importante de fourniture du service; en mesurant la couverture pour chaque etape - c'est-A- dire la proportion entre le nombre de personnes pour lesquelles la condition en cause est remplie et la popula- tion cible - on obtient un ensemble de donnees qui determinent l'interaction entre le service et la population cible. La couverture peut etre definie d'une maniere plus specifique lorsque ce critere est, par exemple, applique A un groupe particulier de population requerant des pres- tations d'une certaine nature ou presentant des caract& ristiques speciales sur le plan demographique ou socio- economique. La fourniture des prestations commence avec l'alloca- tion de ressources, laquelle conditionne la quantite de prestations qui pourront etre offertes A la population cible. Selon la maniere dont le service sera ensuite deploye, c'est-a-dire selon son accessibilite, les candidats aux prestations pourront ou non en faire usage. Mais les gens ne recourront au service offert que s'ils le jugent acceptable, condition qui est liee A de nombreux facteurs parmi lesquels figurent le prix des prestations et les cou- tumes locales. Ces trois premieres conditions etant reunies, le contact entre le service et les usagers potentiels ne pourra s'etablir que si les besoins de ceux-ci sont connus ou pergus. Finalement on pourra mesurer l'effica- cite du service en fonction de sa capacite de couvrir ces besoins en fournissant les prestations appropriees. La couverture etant mesuree A ces cinq niveaux suc- cessifs de distribution des services, on identifiera aise- ment tout goulot d'etranglement dans le fonctionnement, puisqu'on constatera en ce cas un ecart significatif entre les taux de couverture qui s'etabliront pour deux niveaux successifs. Si l'on soupconne certaines contraintes d'etre responsables de ce goulot d'etranglement, l'analyse de ces facteurs sera elle-meme facilitee par la comparaison du taux de couverture dans des situations qui se differen- cient par l'existence ou l'absence desdits facteurs. Enfin, cette fragmentation de la mesure de la couverture est egalement utile pour l'analyse de la relation coOt/effica- cite et le choix de strategies appropriees pour le d6velop- pement d'un service. 303