It is widely accepted that family planning services are essential to fertility decline. The proximate determinant of ongoing fertility decline in the developing world has been the widespread adoption of contraception. Previous studies have shown that the availability and accessibility of family planning services is an important determinant of contraceptive use.1 In Vietnam, the ease of obtaining contraceptives has been shown to be an important factor in the success of family planning programs.2 The government's Population and Family Health Project, a seven-year project begun in 1996, is an attempt to strengthen the primary health care systems of 20 provinces. These provinces were selected because they had the lowest contraceptive prevalence and the highest fertility rates, and because they lacked a substantial family planning donor presence.
Overall, reported contraceptive use among currently married women in Vietnam was 75% in 1997. Despite a 22-percentage-point increase in overall contraceptive prevalence during the last decade, the IUD still predominates among modern methods used in the country: In 1997, 39% of currently married women were using the method, accounting for 51% of all use.3
Using data from the 1997 Vietnam Demographic and Health Survey (VNDHS), we scrutinize here the impact of women's access to family planning services on contraceptive use. This article addresses the following major questions: How has the accessibility of services affected contraceptive use? How widely accessible are the different sources of family planning services? Finally, what are the differences in the accessibility of services between urban and rural areas, between project and nonproject provinces, and between nearby and remote, hard-to-reach communities? These questions are particularly important for evaluating and strategically orienting Vietnam's reproductive health program in the upcoming years.
DATA AND METHODOLOGY
The major data for this article were taken from the 1997 VNDHS, which was conducted by the General Statistical Office under a subcontract with the Population and Family Health Project of the National Committee for Population and Family Planning (NCPFP). Macro International furnished limited technical assistance for the survey. The 1997 VNDHS, which was conducted from June to October 1997, was a follow-up to the 1988 Vietnam Demographic and Health Survey and the 1994 Vietnam Intercensal Demographic Survey. Its primary objective was to provide up-to-date information on fertility levels, fertility preferences, awareness and use of family planning methods, breastfeeding practices, early childhood mortality, child health and knowledge of AIDS.
The 1997 VNDHS was conducted with a subsample of the 1996 Multi-Round Demographic Survey (MRS). The MRS households were located in 1,590 enumeration districts spread throughout the 53 provinces of Vietnam. On average, an enumeration district included about 150 households. For the 1997 VNDHS, surveyors selected a subsample of 205 enumeration districts, then chose 26 households from each urban enumeration district and 39 from each rural enumeration district.*
Of the 7,150 households selected, 7,031 were occupied at the time of the interview, and interviews were successfully completed at 7,001. These interviews identified a total of 5,704 eligible women (ever-married women aged 15-49), 5,664 of whom were successfully interviewed. To evaluate the impact of the national Population and Family Health Project, the 1997 VNDHS was designed to provide separate estimates for the whole country, for urban and rural areas in the country, for the 18 project provinces and for the 35 nonproject provinces.
The 1997 VNDHS used three questionnaires: a household questionnaire, an individual questionnaire and a community/health facility questionnaire. The household questionnaire listed all usual members and visitors in a selected household and collected information on their age, sex, education, marital status and relationship to the head of the household. This questionnaire's main purpose was to identify women eligible for the individual interview. In addition, it collected information on the characteristics of the dwelling unit (such as the source of water, the type of toilet facilities and the material used for the floor and the roof) and on household members' ownership of various durable goods.
The individual questionnaire collected information from all ever-married women aged 15-49 in the surveyed household. Each woman was asked to answer questions on her background characteristics (such as her age, education and residential history), reproductive history, contraceptive knowledge and use, antenatal and delivery care, infant-feeding practices, child immunization and health, fertility preferences and attitudes about family planning, husband's background characteristics, work information and knowledge of AIDS.
The community/health facility questionnaire collected information on all communes in which the interviewed women lived and information on services offered at the nearest health facilities. This questionnaire consisted of four sections. The first two sections gathered information from community informants on such characteristics as the residents' major economic activity, the distance to civic services and the location of the nearest source of health care. For the last two sections, interviewers visited the nearest commune health center and the nearest intercommune health center (if those facilities were located within 30 kilometers of a sample cluster) and collected information on such factors as the type of services offered and the number of days per week they were offered; the types, number and training of staff assigned to the facility; and the equipment and medicines available at the time of the facility visit. In addition, interviewers specifically asked about the contraceptive methods available at the facility—the pill, the IUD, injectables, the diaphragm and "other methods." Although these data are imperfect, they offer a preliminary understanding of contraceptive availability.
We adopted an approach in which we linked the information on contraceptive methods and access from the community facility modules with the individual information collected in the women's questionnaires. These two sources of data allowed us to evaluate the levels of access to and use of contraceptive methods.
"Accessibility" of family planning services generally refers to the extent to which appropriate contraceptive methods are available and the extent to which those in a given location who are seeking contraceptives can obtain services.4 In a broad sense, however, accessibility is a multidimensional concept that not only includes physical proximity and travel time to services, but also involves economic, psychological and attitudinal costs, cognition and the perceptions of potential clients.
Within the framework of the Population and Family Health Project, accessibility refers to the potential user's ability to physically reach service providers. This can be measured directly, by distance or by travel time needed to reach the nearest facilities if one uses the most popular means of transportation available in the community. The project also employs a baseline indicator of accessibility—the number of women of reproductive age who have access to contraceptive methods in a given year, divided by the total number of such women living in the same year. According to this definition, for services to be considered accessible, potential users must have access to at least one fixed service point (which may be a commune health clinic, a mobile team, a drugstore, a hospital, an intercommune clinic or a private doctor) not more than one kilometer away or to a community health worker or a family planning worker who visits at least once every two weeks.
Using this definition, we calculated the baseline indicator of accessibility to services.5 The results show that women living within one kilometer of a contraceptive source were almost three times as likely to be current users of a modern contraceptive method as were women residing more than one kilometer away. This was particularly true for nonclinical methods (the condom and the pill).
In this article, we first present findings by selected characteristics and compare the levels of accessibility in urban and rural areas, in project provinces and nonproject provinces and in nearby and remote communities.† The dependent variables are the current use or nonuse of modern methods and the currrent use of traditional methods. As binary variables with values of zero and one, these dependent variables are appropriate for estimating logit models.
We defined the independent variable as accessibility, which was measured primarily as physical proximity to services. This analysis goes beyond earlier ones to explore accessibility on two different levels. The first is access to any source of services. The second is the source-mix index; for this measure, we constructed a seven-item independent variable corresponding to the number of possible sources of services defined in the facility modules of the 1997 VNDHS.‡
To gauge the net effect of accessibility on contraceptive outcomes, we needed to control statistically for the effects of other influencing factors. Thus, we entered the individual women's age, number of children and education as covariates in the regression model.
The joint analysis of the individual-level and service-environment data enhances the efficiency of this analysis. Such a multilevel framework allows us to study accessibility not only as it is influenced by the individual women's characteristics, but also as it is shaped by the service environment in which the women live. Because we have information only about current accessibility of services, we limit our analysis to current contraceptive use among women who were married at the time of the survey.
Accessibility of Family Planning Services
Overall, 73% of currently married women interviewed in the 1997 VNDHS lived within one kilometer of a facility providing family planning services (Table 1); this percentage was 27 percentage points higher in urban areas (95%) than in rural areas (68%). However, the proportion of women using modern methods in urban areas was similar to that in rural areas (54% and 56%, respectively), as was the proportion with unmet need (7% and 9%). Project provinces were similar to nonproject provinces on most measures; the only notable difference was in the proportion of women living within one kilometer of a facility-based provider (60% vs. 78%).
Although family planning services were available to the majority of women, not all contraceptive methods were equally available. Condoms, oral contraceptives and IUDs were generally easier to obtain than were other methods. Urban-rural differentials in access to family planning services were sizable for almost all contraceptive methods, with access being substantially greater among urban women than among rural women for the pill, the condom, the injectable, the IUD and female sterilization (Table 2).
Urban and rural areas differed little in levels of current use of the pill, the injectable and tubal sterilization. Urban women were more likely than rural women to use the condom, but were less likely than rural women to rely on the IUD (Table 2). Thus, neither the pill nor the condom is widely used in Vietnam, even in urban areas, where they can easily be obtained. In contrast, almost three-quarters of the rural women who have ready access to the IUD use it, and about two-thirds of those who have ready access to tubal sterilization rely on that method.
As Table 3 shows, 84% of currently married women had access to at least one source of family planning services in their community, including both facility-based providers and such sources as drugstores and community health and family planning workers. Services and methods could be obtained from an average of about two sources (out of a possible seven), but access to specific types of sources varied by where women lived. Urban women had access to about twice as many types of sources as did rural women (3.7 vs. 1.8), and those in nearby communities had access to about one more than did those in remote communities (2.4 vs. 1.3). Overall, women were most likely to have access to commune health clinics and drugstores (55% and 47%, respectively), followed by community health and family planning workers (40% and 27%). Only 17% had access to private doctors and 9% to mobile teams.
As noted earlier, urban-rural differentials in service delivery in Vietnam were large in 1997. The Population and Family Health Project provided all women in urban areas with access to services, compared with slightly more than 80% in rural areas. Service delivery points such as health posts, drugstores, private doctors, family planning centers and hospitals are less common in rural than in urban Vietnam, where only 23% of the population resides. The much lower access rate and source-mix index for the countryside suggests that the bulk of future project investment will need to center on rural areas.
Family planning service access was somewhat greater in project provinces (89%) than in nonproject provinces (82%). In project provinces, community health and family planning workers, commune health clinics, drug stores, and mobile teams were the main service providers. Reinforcing our earlier finding on the same issue,6 the results suggest that the project's current strategy for upgrading commune health clinics and establishing models of voluntary workers, if well-implemented, is an appropriate avenue to improve the accessibility and quality of family planning services in project provinces.
A comparison of the service accessibility of remote and nearby areas shows the advantages of communities that have a developed infrastructure and road systems. Both the access rate and the source-mix index of the nearby, easy-to-reach communities were considerably higher than those of remote communities, where access to services was limited for almost all sources. Private doctors and mobile teams were also less accessible in remote communities.
Multivariate Regression Results
The rich data collected through the facility questionnaire allowed us to investigate the impact of accessibility on contraceptive practices. We addressed the specific question: Would better access contribute to improved use of modern contraceptives? In other words, our regression analysis examines the extent to which appropriate interventions in the service environment can increase women's use of contraceptives.
Table 4 presents regression results estimated from reduced forms of logit models in which the effects of key individual background characteristics—women's age, number of children and education—on three measures of contraceptive use are statistically controlled. To ease interpretation of the results, we present the results as odds ratios.
Accessibility was positively associated with contraceptive use for several subgroups of women. The logit model shows that women aged 25-34, those with more education, and those with at least three children were more likely to use a modern method when services were accessible. Accessibility was negatively associated with nonuse of modern methods and with current use of traditional methods (about 0.6 for each). For all three measures, control variables such as age, education and number of children significantly affected contraceptive practices.
In the logit regression results shown in Table 5, the independent variable was the source-mix index, a continuous variable with a maximum value of seven. When the effects of other factors were accounted for, an increase in the source-mix index reduced the odds that women used a traditional method or did not use modern methods (0.9 for each outcome).
Our analysis confirms that physical distance from family planning services does not have an important effect on use of modern methods. The proportion of women using modern methods did not differ much between rural and urban settings and among different types of communities, despite significant differentials in contraceptive access.
The study indicated significant differentials in access to contraceptive services between rural and urban areas as well as different types of communities. Differentials in accessibility largely favor nearby communities; the service environment in remote, hard-to-reach communities is especially poor.
Unlike previously published findings,7 our analysis indicates that accessibility has a significant impact on nonusers of modern methods and on current users of traditional methods. This suggests that improved access could substantially reduce the proportion of the population in these two groups. A considerable proportion of Vietnam's unmet need for family planning could be satisfied if the accessibility of different sources of services were increased. From a programmatic perspective, ensuring maximum access to contraceptive methods is a desirable goal that will strengthen programs among targeted populations.
Several contraceptive methods, such as the pill and the condom, are not widely used even in urban areas, where they are easily obtainable. A study on oral contraceptives in Vietnam indicated that the major reason women who had used modern contraceptive methods had never used the pill was that they did not know about the method.8 If this lack of information is the result of the strong campaigns conducted in Vietnam in the 1980s to promote IUD use and in the mid-1990s to promote sterilization, contraceptive use depends not only on availability but also on the intensity of promotion. Thus, along with availability of contraceptives, access to information on individual methods could reduce the bias in Vietnam toward urban areas and toward IUD and traditional contraceptive use.
The service availability data collected by the 1997 VNDHS have some analytic and interpretative limitations. First, the data reflect perceived rather than actual availability of family planning services. Selected informants were asked to identify only the nearest facility, regardless of whether it was used or preferred by community residents. Thus, informants who had not used a facility probably knew less about the services it provided. Consequently, the information about contraceptive accessibility was likely to be less accurate in areas where the use of services was low than in those where it was high.
Second, the underlying assumption that the nearest facility is the one people used or preferred may not hold true in areas where women can choose from among a wide array of facilities with different levels of quality. The assumption that women will visit the nearest family planning facility is further weakened if women intentionally go to family planning services that are farther away to maintain their anonymity.
Finally, the 1997 VNDHS is tied to population rather than to facilities. As a result, we are unable to say anything about the characteristics of facilities, since the sample may be representative only of health centers that are located near to and are familiar to informants. In many cases, facilities in sampled clusters did not correspond to the administrative boundaries that defined the local community. However, it is important to note that the 1997 VNDHS data give equal weight to all types of facilities.
In addition, the facility questionnaires did not provide much information on the quality of services, such as length of waiting time, staff attitudes and behaviors, and prices of services. Despite these limitations, the facility data provide some important information on the health and family planning service environment that is difficult to obtain from earlier sources. Taking these limitations into account, we recommend that future surveys conducted as part of the project focus on service quality as an aspect of accessibility.