The United Nations Population Fund (UNFPA) estimates that the costs of family planning and other reproductive health services are increasing and that donors and countries will find it difficult to meet them.1 However, the UNFPA estimate is not based on detailed country studies; even if it were, simply extrapolating the findings of such studies may overstate future costs. For example, in recent research calculating the costs of adding various reproductive health services to family planning programs in Mexico and Zimbabwe, the authors argued that "many institutions delivering reproductive health services operate significantly below their physical capacity to see clients, and...much of the equipment required for expanding reproductive health services may already be available for use in family planning and other health services. Thus, the marginal capital costs of adding most types of...services should be small."2 The authors argued that "many institutions delivering reproductive health services operate significantly below their physical capacity to see clients, and that much of the equipment required for expanding reproductive health services may already be available for use...."3 The same argument may apply to staff time.
The costs of providing reproductive health care could be lower than those implied either by UNFPA estimates or by estimates obtained from combining average costs from microlevel studies conducted at service-delivery points. For example, with studies such as that described above, aggregating data would create total cost estimates that are likely to be too high, as these assume that expanding service provision requires hiring additional staff. But just as there may be excess physical capacity, there may also be excess human capacity. Thus, if workers spend much of their time waiting for clients or performing unnecessary paperwork, this time can be reallocated to serving clients. As a result, the incremental labor costs of adding reproductive health services would be low, as additional staff would not be needed.
If new tasks turn out to take up a large amount of time, clinics may need to hire additional workers; if no new workers are hired, providers may need to reduce the time they spend on the provision of existing services, with a resulting decline in the number of visits and the quality of services provided in these visits. If clinics were to add many new reproductive health services, they would likely need new equipment and more space as well as additional staff, and costs would increase.
While providers are apt to say that they do not have enough time to take on new tasks, data on time use show that providers have substantial unoccupied time. For example, in a survey of public-sector providers in Jamaica, about half reported that they did not have sufficient time to do counseling.4 Yet a study of three Ministry of Health clinics in Jamaica indicated that staff spent almost two hours not occupied, and just three and one-half hours with clients.5Similarly, the proposal for an intervention to expand counseling for antenatal clients at clinics in Durban, South Africa, called for additional staff to meet the need for services,6 but results from a client-flow analysis showed that providers were busy in the morning but saw few clients in the afternoon.7 Other studies of time use also have indicated that providers have considerable amounts of free time.8
Information on providers' use of time can be obtained in a variety of ways. The authors of the study conducted in Zimbabwe (a different approach was used in Mexico) used standard times (in minutes) for every activity performed by each category of staff, and multiplied this number by dollars per minute, using information on compensation packages and work time.9 The investigators did not discuss how these standard times were determined, however.
In a study of clinics in Côte d'Ivoire, facility managers were asked how staff, space and equipment were allocated to family planning relative to the other activities at the clinic.10 Yet a recently conducted comparison at four clinics in Jamaica of results from a time-motion study (one that uses trained observers to record the activities of workers) with those from manager interviews revealed that the managers considerably overestimated the time providers spent in contact with clients and in administrative work and underestimated the time they spent unproductively.11 The authors of a study of clinics in Paraguay in which information on time use was obtained from self-administered time sheets argued that these data did not differ substantially from those provided by a time-motion study.12 On the other hand, research conducted in Ecuadoran clinics showed poor agreement between results from a self-administered time sheet and those from a time-motion study with respect to nonproductive time and the time used to support client contact activities.13 Research conducted in Jamaica found a similar lack of agreement.14
Although time-motion studies are generally the "gold standard," even these may produce flawed results if providers alter the way they spend their time while they are being observed. For example, to appear busy to the observer, staff may increase the time they spend on administrative tasks. Although the amount of administrative time should rise with client load, data from clinics in Bangladesh, Jamaica and Zimbabwe indicated that administrative time was highest at clinics where providers spent the least amount of time with clients.15
In this article, we examine the cost implications of expanding the provision of syndromic management services for detecting and treating reproductive tract infections (RTIs), including sexually transmitted infections (STIs), in family planning clinics in Zimbabwe. Syndromic management of STIs allows health care workers to diagnose an infection without the need for sophisticated laboratory tests, because it is based on clinical syndromes and, in some cases, on assessments of an individual patient's risk for STI infection.16 Thus, it can be used where sophisticated laboratory tests are not readily available. The national algorithms in Zimbabwe provide guidelines as to how to treat women with different signs and symptoms, including lower abdominal pain and vaginal discharge.*
Many family planning programs are trying to reach a larger proportion of the population at risk for RTIs—sexually active women of reproductive age.17 In particular, Zimbabwe's National AIDS Coordination Program has noted that "ways are being tested to reach this population through integrating RTI and HIV management services into existing [maternal and child health and family planning] programs."18 According to the Zimbabwe Demographic and Health Survey, 31% of sexually active women were using a modern method of contraception in 1994, and 82% had obtained their method from a clinic.19 Thus, family planning clinics would seem to provide an ideal setting for reaching women in the general population who might have an STI.
As part of a study to provide the Zimbabwe National Family Planning Council with information to improve its management of RTIs, we obtained information allowing us to determine the costs of adding RTI services to the organization's clinics.20 Here, we show how the costs of family planning visits are affected by adding syndromic management services to existing family planning service provision. We pay particular attention to how visit costs are affected by the ways in which providers spend their time, and we model several scenarios in which we vary assumptions about how providers spend their time. We also consider both the timing of downtime during the day and overestimates of administrative time resulting from providers' desire to appear busy while being observed.
DATA AND METHODS
As it is applied in Zimbabwe, syndromic management incorporates a risk assessment that includes questions on behavior, evaluation of symptoms and a pelvic exam to check for signs. All women who adopt a family planning method, all clients coming in for a one-year check-up and some clients coming in for "other" reasons are supposed to be targeted to receive the syndromic approach. Providers had been trained in syndromic management, but because of concerns that they had not implemented the approach, they were retrained.
The research described in this article was part of a larger examination of the implementation of syndromic management in Zimbabwe. As part of this effort, specimens were collected from clients for laboratory testing to determine the accuracy of the syndromic approach (in terms of sensitivity, specificity and positive predictive value). These laboratory tests were not considered a routine service for clients, but were used only to validate the service that was being expanded—the syndromic approach.21
To determine if expansion of syndromic management would increase labor costs, we asked two questions: First, did the length of visits rise after clinicians were retrained in the syndromic approach? Second, could providers draw on unoccupied time to provide these services, or would they have to reduce the time they spent on provision of other services?
When providers are fully occupied, only the first question needs to be addressed, either by obtaining information on how visit lengths change as a result of adding syndromic management or by comparing services received by clients, some of whom receive syndromic management and some of whom do not. We follow both approaches.
The second question requires information on how the time that providers allocate to various services changes. If providers spend considerable amounts of time on administrative tasks or have substantial unoccupied time even after the new service has been added, it can be inferred that providers had time available in which to expand services. We examine the second question using cross-sectional data obtained after the retraining; unfortunately, we do not have valid pretest-posttest estimates of time use.
Finally, we obtained information on direct costs other than provider time. We then used scenarios that varied the percentage of contact time to determine how an increase in the number of visits affected average costs.
A mini-situation analysis conducted in January and February 1998 provided information on the percentage of family planning clients at two family planning clinics (Spilhaus and Mpilo) who received various services. A second mini-situation analysis (the posttest) was conducted in May and June 1998. In addition, during May and June 1998, a mini-situation analysis was conducted at a third clinic (Lister). Data from these analyses provided information on activities conducted and on the duration of contact time.
In each mini-situation analysis, investigators completed a precoded form for the client visits that they observed. The form contained information on whether a risk assessment was carried out, and if so, how many items of the risk assessment were completed,† whether a pelvic exam was done and whether specimens were collected for the lab tests to validate the syndromic approach. Data on family planning services provided, other clinical exams carried out and other discussions held were also gathered.
We used two techniques to determine whether retraining had led to a change in how providers allocated their time. First, for a week in January 1998 and a week in June 1998, providers were questioned at the end of each day about how much time they had spent with clients, including the number and duration of different types of visits, and how they had spent their time when they were not with clients. In addition, to check the validity of the data collected in these interviews and to obtain objective data on how providers spent their time during a full day, we conducted a time-motion study after retraining was completed. At Mpilo and Lister, providers were observed for a week (a total of 30 person-days). In Spilhaus, observations were conducted for 12 person-days. (The clinic was closed one day.) Observers used a stopwatch that beeped every five minutes. At the beep, the activity being performed was noted.
Several months after these data collection efforts, we gathered additional data to estimate the average provider costs of different types of visits for reproductive health services, including family planning. We included the costs of provider and support labor, as well as the costs of contraceptives and supplies. The Zimbabwe National Family Planning Council provided information on the salaries and benefits of clinic providers and support staff and on the costs of gloves (the main materials cost for visits). We also obtained information on the use of other supplies, but did not use these data because they varied so much among clinics. Finally, we obtained information on the costs of contraceptives from a recent study on the costs of commodities and materials for providing reproductive health services.22
To determine the impact of the various services received on visit length, we compared the mean length of visits for combinations of services (risk assessment, pelvic exam and lab tests) for clients seen in June 1998. Because of the small number of cases in some subgroups, we measured the significance of the differences in means using Mann-Whitney and Wilcoxon tests.
In addition, we determined the impact on staff workload of expanding the syndromic approach to targeted women, under the assumption that clients who were targeted by the clinics to receive syndromic management would actually receive it. We then calculated the time necessary to carry out these additional activities and determined whether the staff had the time to do so.
For our calculation of the provider cost of a visit, we allocated labor costs according to the time spent in providing services to new acceptors or continuing users. We were unable to determine what method a new user received, so we used the same time estimate for all acceptor visits. We divided other labor costs by the number of visits. Supply costs were charged to the visit in which particular supplies or contraceptives were used. All inputs were multiplied by their respective unit costs. We also calculated the labor cost per visit using actual data on how staff spent their time and under various scenarios in which we varied the percentage of time that staff spent with clients.
Visit Length and Services Received
Using data from the mini-situation analyses, we found that the median visit length for new acceptors was longer in the posttest period (27 minutes) than in the pretest period (20 minutes).‡ Visit length may have increased at posttest because at that time a higher proportion of women received any of the three services that we examine here: a risk assessment, a pelvic exam, or collection of lab specimens. For example, the proportion of new users who received a risk assessment climbed from 14% at pretest to 40% at posttest (Table 1). However, while the retraining apparently resulted in wider use of syndromic management, it is clear that not all women received these services.
We used data from the mini-situation analyses to explore how risk assessment, the pelvic exam and the collection of lab specimens contributed to the length of a visit. (Because some of the cell sizes are small, we also show the median visit length.) As can be seen in Table 2 women who had a risk assessment but who did not receive either of the two other services spent about the same amount of time with a provider as did women who received none of these services (10 minutes, on average). Thus, it would appear that conducting a risk assessment added little or no time to a visit.
A pelvic exam added about 16 minutes to the average length of a visit, an increase that was not dependent on whether a woman had a risk assessment. The collection of lab specimens added another 17 minutes to the average visit for women who had both a pelvic exam and a risk assessment. (The lab specimens were collected for validation purposes only; they would not be performed as part of normal service provision once the study ended.)
Thus, we can conclude that if syndromic management (or a pelvic examination) were to become a routine part of family planning visits for selected groups of women, it would add only about 16 minutes to their visits.
Contact Time for Syndromic Approach
We calculated how much the total time spent on visits would increase if syndromic management were provided to all targeted clients. To do so, we estimated the proportion of revisits that were for one-year check-ups (one-third) and the proportion of women in the "other" group who should be targeted for the syndromic approach (one-half). No changes were expected for other visits, such as those by STI clients or by women coming to the clinic because of problems with their contraceptive method. All new clients were supposed to be targeted, and additional time expenditures were calculated based on the difference between 100% and the percentage actually receiving services in that group.
At the Spilhaus clinic, the providers spent an average of 22.7 minutes with clients during each visit, and handled an average of 9.5 visits per day (Table 3 page 61).§ Thus, following retraining in syndromic management, providers spent an average of 217 minutes per day with clients. About half (47%) of clinic clients who either accepted a method at their visit, made a return visit or made a visit for some "other" purpose would receive syndromic management if it were extended to all targeted women, but the approach was actually used with only 41% of such women.
Thus, at Spilhaus, the number of additional clients receiving syndromic management would increase by 0.4 per provider per day ([47% - 41%] X 6.3 provider visits per day). These visits would require an additional 6.1 minutes per provider per day (0.4 visits multiplied by 16 minutes, the net change in contact time per provider per day). However, the use of lab tests to verify infection status would end, as this is not part of the syndromic approach. Thus, visit time would be reduced by 17 minutes for the 21% of targeted clients for whom specimens were collected, or by a total of 22 minutes per provider per day (21% X 6.3 X 17). Overall, the total average time each provider spent with clients at Spilhaus would decrease from 217 minutes to 201 minutes per day if all targeted clients received syndromic management on a routine basis.
Results for Mpilo were similar. The mean length of a patient contact was 17 minutes, and there were nearly 12 visits per provider per day, on average. Thus, each provider spent about 200 minutes with clients each day. Expansion of syndromic management would increase the proportion of clients receiving this approach by 11 percentage points in the group consisting of new acceptors, return clients and women attending the clinic for "other" reasons. Although this expansion would require almost 19 additional minutes per provider per day, it would save more than 30 minutes per provider per day by avoidance of lab tests. Altogether, providers at Mpilo would spend, on average, an estimated 188 minutes per day with clients if syndromic management were fully implemented, slightly less than the 200 minutes spent on clients after retraining.
At Lister, providers would need to add the most time for syndromic management (43 minutes), because they were seeing the smallest proportion of targeted clients. However, the total time spent with clients would increase by only 35 minutes—because eight minutes would be saved by eliminating the lab tests—to 196 minutes per provider per day, if all targeted clients received syndromic management.
Impact on Provision of Other Services
Information for the three clinics on how time spent in different activities varied over the course of the day (taken from the time-motion study) was available only for the posttest period; the provider interviews, which were conducted pretest and posttest, did not prove to be valid, and could not be used.** The percentage of time spent in direct care increased throughout the morning, with the maximum proportion (56%) reached between 11 a.m. and noon (Figure 1). The proportion of time spent on lunch and tea breaks was highest from 10 a.m. to 11 a.m. and from 1 p.m. to 2 p.m. Time spent unoccupied was highest in the early morning (when staff sometimes had not yet arrived) and in the late afternoon.
For all clinics, about 38% (mean, 192 minutes) of time during the full day was spent with clients, while an additional 29% (150 minutes) was spent in various administrative tasks or on other work activities (Table 4). The percentage of time spent with clients was higher at Spilhaus and Mpilo than at Lister.
Cost of Visits
The overall estimated cost of visits for different family planning services ranged from $2.49 per pill visit to $3.77 per IUD visit (Table 5). These estimates are in close agreement with those found in a previous study in Zimbabwe,23 although our estimate of the cost of an IUD visit is toward the lower end of the range given in that study ($2.94-8.70).††Our estimates include the costs of provider labor (the nurse's time), gloves and contraceptives.
The differences in costs at the various clinics are partially explained by how providers spend their time. As the number of clients increases, the proportion of time spent with clients increases, and the labor cost per visit decreases. That proportion ranged from 32% at Lister to 43% at Spilhaus.
Labor costs will vary depending on the share of provider time that is spent with clients. Visit costs are lower when providers spend a higher percentage of their time with clients, since the same amount of labor is used to provide more visits. For example, at Spilhaus, if the proportion of time spent with clients were to increase to 50%, the total time spent with clients would rise from 201 minutes (Table 3) to 255 minutes (not shown), or by 54 minutes per day. This would reduce labor costs from $1.73 per visit to $1.46 per visit (Figure 2, page 64). Increasing the share of time with clients to 60% by adding more visits would lower the cost of labor per visit still further, to $1.22.
This time could be drawn from that spent on nonwork activities or could be made up by reducing work activities not involving clients. On average, each provider would then be able to see an additional 2.5 clients, or 26% more clients per day. This change would be even greater at Lister, since providers there now see the fewest clients and spend the smallest amount of their work time with clients. Overall, the labor cost of a client visit at Lister would be reduced from just under $2 to about $1 if the time spent with clients were increased to 60% (Figure 2).
The decrease in cost would be least pronounced in Spilhaus (a drop of 29%), where the proportion of time spent with clients is highest, and most pronounced in Lister (48%), where the share of time spent with clients is lowest.‡‡ Overall, the average provider cost of a family planning visit would be reduced by more than one-third if providers saw more clients and therefore increased the share of their time spent with clients to 60%.
Our analysis assumes that over time, additional clients will seek services at these clinics. If they do not, average costs will not decrease. Moreover, at some point, providers will be fully occupied and additional staff will have to be hired to serve more clients. Thus, labor costs do not increase continuously, but follow a step function: The marginal costs of labor are zero until administrative time and unproductive time are minimized, but increase substantially when additional staff are hired; however, they once again are zero as the new staff use their unproductive time to see more clients.
An important limitation of this analysis is that we are unable to say definitively whether retraining and reinforcement in the syndromic approach, combined with the collection of lab specimens, affected the quantity and quality of other services provided during client visits. However, we believe it very unlikely that other services suffered from the expansion of syndromic management. While visits were substantially longer for women who had pelvic exams or had lab specimens collected, and the total time spent with clients undoubtedly increased as a result of retraining, our calculations indicate that providers still had sizable amounts of time that could be used to expand service provision.
Even with substantial numbers of clients receiving syndromic management, the providers in this study had some unoccupied time at work, and some time when they were simply not at the clinic. They also devoted quite a lot of time to administrative activities. In all three clinics, about as much time was spent with clients as on these other tasks. Although additional time was required to clean and prepare instruments for pelvic exams and to obtain lab specimens, it appears that there was more than enough time to accomplish these tasks. Therefore, the time required to perform syndromic management could have been drawn from unused time or from work time used for administrative activities. Moreover, we may have overestimated administrative time and underestimated downtime if providers made an effort to appear busy.24
On the other hand, downtime was not evenly distributed throughout the day, and there may have been periods during which it was not possible for providers to see additional clients. If scheduling were improved, providers could take advantage of downtime by making appointments during times when fewer clients were obtaining services.
We did not investigate the costs of adding other reproductive health services, but because concerns regarding excess capacity are similar, regardless of the service added, the methodology described here could be used to examine the costs of adding any other reproductive health service. This article is not about syndromic management per se, and the conclusions are likely to be unaffected by the particular service examined.
Because providers have some downtime (likely more than we have estimated), we argue that such time needs to be considered when program managers estimate the labor costs of providing a new service. Many family planning clinics in the developing world operate under conditions similar to the three in Zimbabwe, in which providers are not directly engaged in client-related activities for a fairly high percentage of the day. Cost estimates such as those provided in earlier analyses25 may be adequate for determining the current average cost of a visit of a particular type, but they do not provide the information on time usage that is needed to determine the incremental cost of adding additional visits or of offering new services during visits.
Our findings have implications beyond the cost of syndromic management provision. Many estimates of the cost of adding new services to existing programs assume a multiplicative relationship between the number of visits and costs and do not take into consideration the way that staff spend their time. Consequently, these estimates are likely to be too high. Instead, cost studies of service integration should determine whether staff members can add new services or visits without having a negative impact on the quality of existing services.
Realizing these results will require an increase in the efficiency of health service provision and a corresponding improvement in clinic management. Such changes will not be easy to achieve. Incentives to increase performance are lacking in the public sector of many developing countries. Without such incentives, providers may choose to spend fewer hours providing services in the public sector so they can augment their incomes in the private sector, or they may expect to be compensated for their very low salaries through nonmonetary benefits, such as a light client load.
Given the resource constraints facing most reproductive health programs, changes are necessary if programs are to expand the quality and variety of services that they provide. Health-sector reform needs to be concerned with incentives for providers. Of course, increasing providers' salaries will raise labor costs, but improving productivity will still reduce the costs per unit of service, especially if productivity gains are large. The challenge, therefore, is to find ways to improve the use of resources rather than to argue that they are inadequate.