Table of Contents Acknowledgements.... ii List of Abbreviations..... v Executive Summary.... vi Introduction and Background.... 1 Study design.... 5 Description of the target population and selection criteria for human subjects. 5 Statistical analysis.... 7 Outcome measures..... 7 Provider and CBD agent characteristics.... 9 Study results..... 12 IUCD Uptake.... 12 AMKENI status.... 13 Reported IUD service provision.... 14 Exposure to the intervention.... 15 Other sources of information about the IUCD... 17 Effect of the detailing intervention... 18 Access to essential supplies.... 20 Costs of the detailing intervention... 22 Knowledge, attitudes and self-efficacy.... 25 Data interpretation workshop with local stakeholders... 28 Conclusions and Recommendations... 30 References..... 32 Appendix 1: Knowledge, attitude and self-efficacy tables... 33
List of Abbreviations CBD community based distribution DMPA DPHN DRH FHI FP GTZ IEC IUCD MOH OCP WHO depot-medroxyprogesterone acetate, AKA depo provera, injectables District Public Health Nurse Division of Reproductive Health Family Health International family planning German Agency for Technical Cooperation (Gesellschaft fur Technische Zusammenarbeit) information, education, and communication intra-uterine contraceptive device Ministry of Health oral contraceptive pill World Health Organization
Executive Summary This operations research study tested the effectiveness of an outreach intervention to clinic-based family planning providers and community-based distribution (CBD) agents, in promoting use of the intrauterine contraceptive device (IUCD) in Kenya. Rationale for the study In July 2002, the Kenyan Ministry of Health (MOH) signed a letter of support for the Reintroduction of the IUCD to the family planning services of the country. Although IUCD use at one time comprised 31 percent of the total modern contraceptive use in Kenya, that percentage has declined in recent years to 7.6 percent in 2003. Research conducted by Family Health International (FHI) indicated that one of the major barriers to IUCD use in Kenya is provider behaviors. A simulated client study found that providers spontaneously mentioned the IUCD in fewer than half the monitored interactions. The IUCD is a safe, effective, reversible contraceptive method. The length of its effectiveness makes it a low cost method for clients and the public health system. Furthermore, the IUCD requires fewer follow-up visits by clients than short-term methods thus over time providers will be required to spend less time with clients using the IUCD than clients using methods that require regular re-supply visits to the clinic. Study objectives The study tested a practical approach to reducing clinic provider and CBD agent biases about the IUCD while at the same time giving them information and tools to better counsel their clients regarding this method. The primary objectives of the study were to determine: (1) Whether detailing among clinic-based providers and CBD agents was an effective intervention to increase IUCD use in Kenya. (2) Whether detailing among clinic-based providers and/or among CBD agents had a positive impact on provider attitudes, knowledge, and self-efficacy regarding the IUCD. (3) The costs of implementing a detailing intervention among clinic-based providers and CBD agents. Study design The study used an experimental pre/post-test factorial design to test the effectiveness of the educational outreach intervention among both clinic-based providers and CBD agents. The study was designed to give information on whether the detailing intervention was more effective when applied to clinic-based providers, CBD agents, or both groups at the same time. The clinics chosen were MOH clinics in Western province, some AMKENI-supported, that had CBD programs attached to them, and had providers trained to insert and remove IUCDs. Forty-five study sites were chosen, and then randomly assigned to four groups: (1) Received the intervention among clinic based providers only; (2) Received the intervention among CBD agents only; (3) Received the intervention among both providers and CBD agents; and (4) Received no intervention at all. All family planning providers in the study sites and CBD agents attached to those clinics were invited to participate in the intervention. The study was reviewed and approved by the Protection of Human Subjects Committee at FHI and approved by Kenyatta National Hospitals ethics review board. Design of Intervention The exact design of the detailing intervention was created working closely with the Division of Reproductive Health (DRH), Ministry of Health (MOH), and in collaboration with the AMKENI

and MOH/GTZ projects. The intervention trained existing MOH supervisors (District Public Health Nurses) from the districts where the study took place. This model was selected so that the detailing intervention would not be too costly to be replicated on a wider-scale. The intervention consisted of two visits to family planning providers in health clinics and/or to CBD agents by the trained detailer. For both provider and CBD agent groups, the detailers asked the participants at the first visit to commit to activities to promote IUCDs. Detailers were instructed to proactively present both the advantages and disadvantages of the IUCD to participants within the context of client informed choice, and worked to directly address rumors and misconceptions about the IUCD. Providers and CBD agents were presented with IEC brochures and promotional materials to emphasize the visits key messages, and to use in their interactions with clients. Approximately one month after the first visit, detailers returned to each clinic and/or CBD program to discuss providers and CBD agents experiences in implementing their action plans. At this time, detailers also assisted participants in identifying solutions to any problems that surfaced. Data collection At baseline and six months after the intervention, indicators of clinic-based provider and CBD agent knowledge, attitudes and (for providers only) self-efficacy in counseling about and inserting the IUCD were measured in all 45 sites. These were measured through intervieweradministered questionnaires after having completed informed consent procedures and obtained participant signatures. Fieldworkers also collected service statistics information on the number of IUCD clients as a proportion of all family planning clients (new and continuing) for the entire twelve-month period of the study. The intervention costs were assessed through record reviews and interviews with key informants. Results The intervention only modestly increased the provision of IUCDs and only when both clinicbased providers and CBD agents were targeted. We concluded that detailing is most effective when it is done with both clinic providers and CBD agents. The clinic providers represent the supply side of service provision and the CBD agents the demand side. Poor provider attitudes and technical skills are not adequately addressed by this intervention and may constitute the major obstacles to increasing IUCD uptake. Furthermore, the two detailing visits provided do not appear sufficient to sustain the effect of the intervention or to completely address poor provider attitudes and lack of technical skills. The number of IUCDs provided dropped immediately after the intervention ceased, indicating that continued promotion and education about the method are required. A cost-effectiveness analysis reveals that the cost per 3.5 years of pregnancy protection was $49.57 for the detailing intervention plus the cost of the IUCD compared to $15.19 for the commodity costs of the current standard of care, DMPA provision. Thus we conclude that although the effectiveness of provider-based activities was somewhat amplified when concurrent demand creation activities were carried out, we cannot recommend the expansion of the detailing intervention due to its high cost and modest outcome.

Detailer reviews basic anatomy with CBD Agents to reassure them that the IUCD cannot travel in the body.
The detailers asked the participants to create a list of the barriers they perceived to IUCD provision. One frequent concern raised by participants was the prevalence of rumors about the IUCD among clients. Some of the more common rumors are that the IUCD can travel to other parts of the body, a child could be born while holding the IUCD or that the IUCD causes infertility and/or delays pregnancy. Since the detailers had received up to date information about the IUCD and had already developed many message confronting these myths during training, they were able to use concrete evidence to demystify the rumors. For example, they explained
In some facilities other support staff members were also invited because they come into contact with clients within the community, are often regarded as doctors, and may be confronted with situations where they might counsel clients on FP.
that the IUCD cannot travel to the other parts of the body because the only opening out of the uterus is via the vaginal cavity. Through sessions with the CBD agents, detailers concluded that some of the myths circulating in the community were originating from CBD agents themselves thus making them an essential group to target with education and motivation. The detailer then assisted the group in developing activities that addressed these barriers and that they could carry out over the next month to increase IUCD uptake. Some of these included: Clinic based providers Continue health talks (micro-teachings) to clients in the waiting room, placing more emphasis on the IUCD Give updates to other service providers not present during the detailing session Order any equipment or supplies necessary for inserting IUCDs that were missing from their facilities Intensify sensitization to womens groups, CBD agents and clients who go for VCT services Conduct re-orientation of IUCD insertion and removal for providers who were not comfortable with their skills in these areas Ensure quality counseling so that clients can make informed choices CBD agents Sensitize community members about the IUCD Refer clients not suitable for other methods who are interested in the IUCD to the local health clinics Conduct health education talks discussing the IUCD during home visits and/or meetings for womens groups In addition to the group talks, detailers also held one-on-one discussions with some clinic providers and CBD agents to address their individual concerns. Since the detailers actually supervised the health facilities in the areas they were assigned, they were able to help clinic staff and CBD agents to confront any supply problems that existed. For example, some detailers delivered missing supplies themselves, while others arranged for clinics to borrow missing supplies or equipment from other facilities. Detailers also distributed MOH-created IEC pamphlets for providers and CBD agents to use to educate their clients. The IEC materials distributed included 250 IUCD advocacy briefs for the providers, 18,000 IUCD specific client brochures and 5300 general family planning brochures. Clinic providers and CBD agents were given promotional items such as key chains, badges (e.g. buttons) and pens, embossed with the IUCD Reintroduction Initiative theme, A new look at IUCDs, to prompt them to remember to counsel their clients about the IUCD whenever appropriate. Approximately 950 key rings, 2000 badges and 1000 pens were distributed.

Study design The study tested a practical approach to reducing clinic based provider and CBD agent biases about the IUCD while at the same time giving them information and tools to better counsel their clients regarding this method. The objectives of the study were: To determine whether detailing among clinic-based providers and CBD agents was an effective intervention to increase IUCD use in study sites To determine the cost-effectiveness of implementing a detailing intervention in study sites among clinic-based providers and CBD agents To determine whether detailing among clinic-based providers and CBD agents had a positive impact on provider and CBD agent knowledge, attitudes, and self-efficacy regarding the IUCD Description of the target population and selection criteria for human subjects Study sites were public sector rural health facilities in five districts of Western Province with family planning programs and a CBD program attached. Western Province, an underserved area of the country, is also the same province where a large, integrated reproductive health intervention, supported by USAID/Kenya, the AMKENI project is currently taking place. The study utilized some sites that are included in the AMKENI intervention and some that are not in order to determine whether the intervention works equally well outside the AMKENI project. Also in Western Province, the MOH cooperates with the German development agency (GTZ) to manage a large CBD program. The MOH/GTZ program has over 2800 active CBD agents in Western Province, who work as part-time volunteers, receive non-monetary incentives and are attached to specific public sector health facilities. The study targeted those health care providers and CBD agents attached to the selected study sites. A complete list of health facilities with family planning services was obtained for all chosen districts. The eligibility criteria for the study sites stated that facilities must: Have an active family planning program Be capturing data on their family planning clients Have at least one FP provider trained in IUCD insertion and removal who was expected to remain at the facility during the time of the study Have active CBD agents attached to the facility The study used an experimental pre/post-test factorial design to test the effectiveness of a detailing intervention among both clinic-based providers and CBD agents. A factorial experiment tests the effect of more than one treatment (factor) using a design that permits an assessment of interactions between the treatments.(10) In this case the two treatments were the detailing intervention with clinic providers and the detailing intervention with CBD agents. The benefit of using a factorial design is its power to assess the effect of the intervention in each target group separately, and to examine if there is an interaction effect when both target groups receive the intervention. In the case of an interaction effect, the impact of the intervention in one group is affected by the presence of the intervention in the other group. Thus, the study was able to provide information on whether the detailing intervention is more effective when applied to clinic-based providers, CBD agents, or both groups at the same time.

Forty-five public health facilities were selected as study sites using the selection criteria. They were then stratified based on whether they were AMKENI sites, and randomly assigned to receive the intervention in neither, one, or both target groups (Table 1). The majority of the study sites were health clinics (36), but 7 hospitals and 2 dispensaries that met the site selection criteria were also included. The planned sample size of 20 clinics per study arm was selected to have at least 80% power to detect a 5% difference of IUCD use between arms under type I error rate = 0.05 with the assumption of at least 0.5 correlation between outcomes at the pre-and post-intervention and no interaction between factors. Research assistants visited each facility at baseline and follow-up and collected monthly service statistics for 12 months (Jan-Dec 2004) using the daily activity logs in each facility. All available clinic based service providers offering FP services in the study sites were interviewed during baseline as well as follow up surveys. Due to their larger numbers, the study interviewed a sample of CBD agents at each facility. A list of codes of active CBD agents was used to randomly select 15 CBD agents per facility (using a random numbers table). In facilities where there were fewer than 15 CBD agents, all were invited to answer the study questionnaire. CBD agents selected to participate in the study were asked to report to the facility on one of the two days that research assistants were there collecting data.
Table 1: No. sites by factorial study design intervention assignments AMKENI Sites CBD Detailing 10 No CBD Detailing 10 Total 20
Detailing in Clinics No Detailing in Clinics Total Non-AMKENI Sites
Detailing in Clinics No Detailing in Clinics Total

CBD Detailing 13

No CBD Detailing 12

Total 25

During baseline data collection in March and April 2004, research assistants conducted interviews with 131 clinic-based family planning service providers and 480 CBD agents. At follow-up data collection (January and February 2005), 120 providers and 402 CBD agents were interviewed. Every attempt was made to speak to the same providers and CBD agents at the follow-up interviews, and the interviewers succeeded in re-interviewing 83.8% of the CBD agents at follow-up. Unfortunately, we were unable to match identities of providers between baseline and follow-up due to issues of respondent confidentiality. CDD agents had pre-assigned code numbers that could be used to match their responses at pre and post-test data collection, rather than collecting their names. Such code numbers did not exist for providers, thus we would have had to collect their names for the matching and this would have increased the risk of a breach of confidentiality. Ten percent of provider respondents at follow-up said they had been working in that facility for less than a year, so we are certain there were at least some providers interviewed at follow-up who did not contribute information at baseline.

To measure the effects of the detailing intervention on individuals in the target groups, provider and CBD respondents were read a series of statements designed to evaluate their knowledge about and attitudes towards the IUD during interviewer-administered face-to-face interviews. For example, one of the attitude statements was, I would recommend the IUD to a friend or family member. Several statements were also posed to determine the respondents sense of self-efficacy with regard to counseling about and providing the IUD. For each statement, respondents were asked to respond either true/false or agree/disagree. In some cases, providers and CBD agents were given different statements, based on the assumed level of their knowledge and specific issues that applied to one target group but not the other.
Provider and CBD agent characteristics The great majority of both providers and CBD agents were female (CBD: 91.3% female; providers: 84.0% female), which mirrors the overall gender distribution of providers and CBD agents in this area. The CBD agents were aged 25 to 74 years old, with a median age of 45 years. The average CBD agent interviewed had been a CBD agent for 10.6 years (range: 3-30 years). Providers were on average slightly younger than CBDs, with a median age of 42.6 years (range: 27-54). However, the providers had spent a good deal more time as health care providers than the CBD agents. The CBD Agent presents her concerns about the IUCD to detailer. average provider-respondent had been working in health care for 18.2 years (range: 2-33). On average, providers said they had worked in the facility in which they were interviewed for 6.7 years (median 4.0 years). However, the distribution is highly skewed. Nearly thirty percent of respondents (29.0%) had worked in their clinic for one year or less at the time of interview, reflecting the high rate of staff transfers in the MOH system. The distribution of respondents at follow-up was similar to baseline, with respondents predominantly being female (CBD: 91.5%; providers: 84.2%) and median ages in the mid-40s (CBD: 45 years; providers: 44 years). The average amount of time spent as a provider and CBD agent were also similar to baseline (CBD: 11.3 years; provider: 17.8 years). Providers were asked to give their current primary responsibility in the clinic. (Table 15) Less than half (baseline: 43.8%; follow-up 34.2%) of the providers said that their current primary responsibility was family planning. This proportion may be lower at follow-up because some providers interviewed at baseline were assigned to the family planning clinic at that time, but had been assigned different responsibilities in the meantime.

Table 2: Providers Current Primary Responsibility in Facility Baseline Follow-up Current Primary Responsibility No. % No. % Family planning 57 43.34.2 Child health 17 13.35.8 Supervision/management 16 12.5.0 General curative (inpatient or outpatient) 9 6.10.0 Antenatal care 9 6.7.5 Labor and delivery care 4 3.0.8 STI/HIV/AIDS/PMTCT 3 2.1.7 Counseling 2 1.1.7 Pharmacy 0 0.0.8 All services 13 10.12.5 Total 120 100
At baseline, nearly one-quarter of providers (22.9%) and over one-third of CBD agents (37.5%) reported currently using no method of contraception or being menopausal or widowed, implying no need for contraception. (Table 3) The numbers of respondents independently reporting that they were menopausal or widowed at baseline were so high that they were added as separate response categories at follow-up, explaining why they are even higher at the second data collection. The providers, younger on average than the CBD agents, were less likely at both baseline and follow-up to report being menopausal or using no method (baseline: 22.9%; followup 31.7%). Interestingly, the proportion of CBD agents reporting personal IUCD use doubled from baseline to follow-up. Respondents tended to use contraceptive methods that they have easy access to. For example, providers were more likely to use clinical methods. Among providers, the most commonly reported method was female sterilization, followed by the IUCD, condoms and the injectable. CBD agents were most likely to use oral contraceptive pills (OCP) at baseline, which they themselves distribute to clients. At follow-up, however, the proportion on OCPs decreased, as the proportion using IUCDs increased. Among the CBD agents, the most popular form of birth control was the oral contraceptive pill (baseline), condoms, female sterilization and injectables.
Table 3: Respondents Current Use of Contraception Providers CBD Agents Baseline Follow-up Baseline Follow-up Contraceptive Method* No. % No. % No. % No. % Female sterilization 32 24.19.16.18.4 IUCD 22 16.17.3.7.0 Condoms 20 15.11.18.14.2 Injectable 19 14.10.11.11.9 OCP 5 3.5.23.14.2 Natural method 1 0.9 1.2.7 Norplant 4 3.4.1.0.5 Male sterilization 1 0.2 0.0.5 Menopausal/Widowed 12 9.18.30.30.8 No method 18 13.13.7.11.2 Pregnant 2.1 0.2 Total *Note: Percentages do not add up to 100%, since more than one answer was possible.

Figure 1 demonstrates the effect in the study sites assigned the intervention among both provider and CBD agent target groups using the unadjusted proportions of IUD among all family planning clients. The detailing visits (intervention) took place in June and July. Immediately thereafter, the proportion of FP clients initiating IUD use increased in clinics receiving detailing among both clinic-based providers and CBD agents. This increase started to drop off almost immediately after the intervention although it remained elevated as compared to baseline. It is worth noting that implants were out of stock for much of the study period. It was back on the shelves of clinics in September 2004 and its provision increased at the same time that IUD provision was decreasing following the post-intervention bump. We cannot directly link the restocking of implants to the lack of success of the detailing intervention, but implants appear to be popular with both providers and clients. Data after December 2004 have not yet been collected to determine if any increase was sustained over a longer period of time.
Figure 1: Average quarterly percent of all family planning clients accepting IUCD by intervention group (unadjusted), 2004
% FP clients accepting IUCDo
2.0 1.5 1.0 0.5 0.0 Jan-Mar

Clinic Only

Apr-Jun

CBD Only

Jul-Sep

Clinic + CBD None

Oct-Dec
For facilities that received detailing among clinic-based staff, support from the AMKENI project was an important, statistically significant predictor of IUCD provision (data not shown). AMKENI-supported facilities that received the intervention in the clinic provided more IUCDs on average than non-AMKENI facilities, and were better able to maintain the increase in IUCD provision over time than non-AMKENI facilities. The single most important predictor of IUCD was the baseline provision of IUCDs, which was highly statistically significant (p<.0001) in both models (data not shown). Differences among districts, facility type and the effects of access to essential supplies were not important predictors of IUCD or statistically significant and were excluded from the model. We conducted a sensitivity analysis accounting for those clinics that experienced cross-over (e.g. contamination) of intervention groups and found that this made no difference in the statistical findings of the study effect.
Access to essential supplies One possible reason for the continued low provision of IUCD could have been lack of essential supplies in the clinics. To assess the availability of commodities, expendables and equipment designated by the Kenya IUCD Reintroduction Initiative Task Force as the required supplies and equipment for IUCD provision, clinic providers were asked whether supplies were in stock and equipment was functioning on the day of each interview. The sterilizing or high level disinfecting equipment was the most likely piece of equipment to not be in working order, and cotton wool was the most commonly missing supply. (Table 17) The proportion of facilities with working sterilization or high level disinfecting equipment increased from baseline to follow-up.

Table 19: Actual intervention-related costs for 34 health facilities Kenyan Shillings (KSh) Training costs MOH Trainer for detailer training Airfare and airport transfers for trainer Accommodation for detailers and training space Per diem for detailers Transport for detailers Meals during training Stationery Sub total Materials production Promotional bics (pens) Key rings IUCD briefs @US$2.50 ea x 250 IUCD brochures/General FP brochures Sub total Intervention costs Transport reimbursement to CBD agents Detailers transport during field work Communication Sub total Grand total Exchange rate (2005): US $1 = KSh 75 25,000 32,257 29,050 12,500 3,800 7,870 4,950 115,427 11,903 92,118 95,881 664,126 864,027 125,217 59,559 6,882 191,659 1,171,113 US Dollars 66 1,1,228 1,278 8,855 11,520 1,92 2,555 15,615
The detailers in this operations research study were provided with per diems for their participation, as stipulated by Ministry of Health, Division of Reproductive Health protocol. We have not included detailer per diems as part of the cost calculations presented here, since the payments to detailers were study-related costs. Discussions with MOH staff members have confirmed that providing detailers with money for transportation to make the detailing visits, and adding detailing to their job responsibilities should be sufficient to ensure that MOH personnel carried out detailing activities. Regular salary costs are also not included in the estimate, since we assume that current MOH staff members will perform the detailing intervention as part of their normal supervisory duties, incurring little additional staff time. Furthermore, the detailing visits took up relatively little time. During the study, detailers spent an average of 2.8 hours in the facilities during the first visits and 2.1 hours during their second visits to the clinic. Not all of the 34 facilities receiving the detailing intervention in the study received detailing in both target groups, therefore we calculated a per facility cost for implementing detailing targeting both clinic-based providers and CBD agents following the two-visit model utilized in the study. (Table 20) The training costs include a 5-day workshop for 5 detailers, utilizing the curriculum written for the study. Because most of the costs associated with the training are per participant (e.g. everything except the expenses for the trainer), there would be only limited economies of scale in increasing the size of the training cohort.
Table 20: Unit cost per facility of implementing detailing among both providers and CBD agents Kenyan Shillings (KSh) Training costs MOH Trainer for detailer training Airfare and airport transfers for trainer Accommodation for detailers and training space Per diem for detailers Transport for detailers Meals during training Stationery Sub total Materials production Promotional bics (pens) Key rings IUCD briefs @US$2.50 ea x 250 IUCD brochures/General FP brochures Sub total Intervention costs Transport reimbursement to CBD agents Detailers transport during field work Communication Sub total Grand total

CBD Agent examines an IUCD to discover how soft the strings are.
CBD agents to promote the method in communities, they did not see an increase in demand for the method. The lack of time in facilities might also reinforce provider biases against the method as being too time-consuming. Finally, there were isolated instances of missing supplies and more widespread problems with provider skill levels, despite the amount of perceived selfefficacy they reported in the study Workshop participants acknowledged that CBD agents could play an important role in promoting the IUCD. Some activities they suggested to facilitate this are conducting refresher courses for CBD agents and recruiting new CBD agents as needed. They also suggested introducing nonmonetary incentives for CBD agents to strengthen their commitment to serving as community resource persons. The final conclusions and recommendations of the data interpretation workshop were that continuous medical education, including on-the-job training in technical skills is essential for maintaining quality services and client method choice. The involvement of stakeholders such as the MOH headquarters, DPHNs, and the provincial health administration are key for successfully replicating the detailing intervention. In addition, participants acknowledged the usefulness of IEC materials, but suggested that the detailing intervention use less expensive forms of these materials, such as posters, to improve the sustainability of the intervention. Remarks from the Provincial Medical Officer, read on his behalf by the Provincial Nursing Officer, closed the workshop. The Provincial Medical Officer endorsed sustaining the detailing intervention and collaborating at all levels of the MOH and with all partner organizations to more effectively market the IUCD to clients. He also recognized the key role of community resource persons, such as CBD Agents in encouraging IUCD uptake and recommended strengthening referral networks between communities and their rural health facilities.
Conclusions and Recommendations IUCD provision was low in the participating facilities at baseline and remained low 6 months after the detailing intervention. IUCD clients made up an average of 0.4% of all FP clients at follow-up in all facilities. The intervention had a moderately statistically significant effect in increasing IUCD provision among facilities that received detailing for both providers and CBD Agents, but the effect was still very small and did not meet our definition of programmatically significant (5% increase). In facilities where only one target group received the detailing intervention, there was no increase in IUCD provision observed. This leads to our primary conclusion which is that in order to increase uptake of the IUCD, efforts must be made on both the supply side (e.g. providers) and the demand side (e.g. clients). The inclusion of CBD agents as one of the target groups was intended to serve the purpose of increasing client demand and appears to be important to achieving the goals of IUCD promotion. Without a community-oriented demand creation element, clients may arrive at the clinic with their minds already made up to request the typical contraceptive which providers will then feel obliged to supply, regardless of its appropriateness to the womans reproductive needs. We saw an immediate drop-off of IUD provision after the detailing intervention ceased, indicating that two detailing visits is probably not sufficient to sustain increases. The intervention was specifically designed to have only two visits so that it might be realistically replicable for the MOH. A more intensive intervention might produce stronger results, but at the expense of any likelihood that the intervention would be scaled up after the research was completed. Why wasnt the intervention more successful in significantly increasing IUD provision in any meaningful way? We examined several factors and found that facility type, access to essential supplies and district were not significant determinants of success. The sensitivity analysis suggests that the contamination among study arms did not significantly change the results. It is worth noting that implants were out of stock for much of the study period. They were back on the shelves of clinics in September 2004 and implant provision increased at the same time that IUD provision was decreasing following the post-intervention bump. We cannot directly link the re-stocking of implants to the lack of success of the detailing intervention, but implants appear to be popular with both providers and clients. Examining changes in knowledge, attitudes and self-efficacy, the intervention appears to have been successful in increasing knowledge and improving attitudes overall, but on some key issues respondents remained unmoved. For example, only about half of respondents agreed that unmarried women could use IUCDs, and most respondents were strongly opposed to nulliparous women using IUCDs. We also noted that the promotional detailing visits did little to improve self-efficacy among providers. Although providers at both baseline and follow-up reported relatively high levels of self-efficacy about inserting and removing IUCDs, participants in the data interpretation workshop confirmed the researchers impressions that technical proficiency is still a major obstacle to IUCD provision.

The detailing training focused considerable attention on finding messages to convince providers that increasing IUCD uptake would decrease their workload in the long run. However, postintervention approximately half the respondents, whether they were subject to the intervention or not, agreed that they are often too busy to insert IUCDs. This indicates that providers are still not enthusiastic about offering this service to clients. By increasing client demand and promotion efforts about the IUCD among providers, we hypothesize that an expanded and sustained detailing effort may be able to more significantly increase IUCD uptake. Among the strengths of this study is its experimental design including a control study arm and random assignment to intervention study arms. On the other hand, there was a large amount of intervention contamination. The contamination points to the difficulties inherent in conducting research in a real-life field setting. Only just over half of clinic providers in sites assigned to receive the detailing intervention reported that they actually heard the detailers presentation. This may be due to staff transfers between job duties or facilities, or due to providers simply not having been available on the day the detailer came. CBD agents assigned to the intervention were more likely to report receiving it, which makes sense given that they are more stable members of the community than clinic providers. In addition, many providers and CBD agents who were not assigned to receive the intervention reported having heard the detailing presentation. This may have taken place since some providers assigned to receive the detailing intervention may also have been CBD supervisors, and may have been present during a detailing session with CBD agents. Furthermore, since the detailers are district supervisors they may have given messages about the IUD to nonintervention groups, despite requests that they refrain from doing so before the study was over. Or, the respondents may simply have been mistaken about whether the source they heard about the IUCD from was the detailer. The nationwide IUCD Reintroduction Initiative was on-going at the same time as the study and approximately thirty percent of all respondents (33/120 providers, 128/402 CBD agents) reported that they had received training or education about the IUCD from a source other than the detailer in the past six months. We examined the cost-effectiveness of the detailing intervention to determine if it could spur enough IUCD use to save the family planning program money in commodity costs and contribute to its overall sustainability. We found that if replicated exactly as done during the study, implementing the intervention would cost $49.57 per additional 3.5 years of contraceptive protection, whereas continuing DMPA provision for 3.5 years is associated with a cost of $15.26. Clearly detailing is not cost-effective in this light. To make the detailing intervention worthwhile, it would have to improve its effectiveness three-fold and spur provision of an additional 23 IUCDs per quarter rather than the 6.5 IUCDs per quarter we observed in this study to make it comparable to DMPA provision. Thus we conclude that although the effectiveness of provider-based activities was somewhat amplified when concurrent demand creation activities were carried out, we cannot recommend the expansion of the detailing intervention due to its high cost and modest outcome.

93.5 97.5

96.8 97.8
Table 23: Changes in correct/favorable responses by intervention group from baseline to followup: General support for the method Providers Statement/Question (Correct/Favorable Response) In detailing intervention facility? I would recommend the IUCD to a friend or family member. (A) Given the advantages of the IUCD, it should be used by more Kenyan women. (A) There are too many issues to consider when deciding if a woman can use an IUCD. (D) The IUCD should always be included when discussing family planning choices. (A)
Change in Proportion Correct/ Favorable Yes No
Change in Proportion Correct/Favorable Yes No

-0.4 +6.5 +0.8

+6.6 -3.0 -3.7
Table 24: Correct/favorable responses: Efficacy & sequelae Statement/Question (Correct/Favorable Response) Providers A copper IUCD is effective in preventing pregnancy for up to 12 years. (T) The IUCD is more effective at preventing pregnancy than oral contraceptives. (T)* Over a five-year period, the IUCD is as effective as sterilization in preventing pregnancy. (T) IUCDs can lead to infertility. (D) IUCDs cause the majority of the cases of pelvic inflammatory disease (PID) in Kenya. (F) CBD Agents A copper IUCD is effective in preventing pregnancy for up to 12 years. (T) The IUCD is more effective at preventing pregnancy than oral contraceptives. (T)* The IUCD is more effective at preventing pregnancy than implants/Norplant. (F) Sterilization is the best contraceptive method for a woman who doesnt want any more children. (D) 62.4 76.0 36.5 2.0 76.6 90.0 76.8 1.0 Baseline No. % Correct/ Missing Favorable 27.7 81.7 66.7 90.0 88.Follow-up
% Correct/ Favorable No. Missing

47.3 83.3 70.7 92.5 91.3

After an IUCD is removed, a woman can become 84.90.pregnant again right away. (T) * Note: for this question, providers were asked to evaluate the opposite statement, that IUCDs are less effective than oral contraceptives. The results above are reported for the percentage of correct responses to both statements.
Table 25: Changes in correct/favorable responses by intervention group from baseline to followup: Efficacy and sequelae Providers Statement/Question (Correct/Favorable Response) In detailing intervention facility? A copper IUCD is effective in preventing pregnancy for up to 12 years. (T) The IUCD is more effective at preventing pregnancy than oral contraceptives. (T)* The IUCD is more effective at preventing pregnancy than implants/Norplant. (F) Over a five-year period, the IUCD is as effective as sterilization in preventing pregnancy. (T) IUCDs cause the majority of the cases of pelvic inflammatory disease (PID) in Kenya. (F) IUCDs can lead to infertility. (D) After an IUCD is removed, a woman can become pregnant again right away. (T) Sterilization is the best contraceptive method for a woman who doesnt want any more children. (D) +8.6 +8.1 +6.4 -1.7 -3.0 -2.3 +9.2 -1.1 +3.0 -1.9