The malaria testing and treatment landscape in Kenya: results from a nationally representative survey among the public and private sector in 2016

A nationally-representative, cross-sectional quantitative survey was conducted among outlets with potential to stock anti-malarials or malaria diagnosis. All potential public and private sector outlets were included in the survey. The public sector included all tiers of the health care system (hospitals, health centers, dispensaries, clinics and CHWs) owned by government or affiliated with the not-for profit organizations such as non-governmental and faith-based institutions. Outlets surveyed in the private sector included, private for-profit health facilities (hospitals, nursing homes/medical centers and clinics), pharmacies and chemists (registered and unregistered) and general retailers selling fast-moving consumer goods. Table 1 provides an additional description of the outlet types.

The 2016 survey was stratified to deliver estimates for each of the aforementioned malaria epidemiological zones in Kenya. Clusters were selected from the four malaria epidemiological zones and defined as (1) endemic areas, (2) epidemic-prone areas, (3) seasonal-transmission areas, and (4) low-risk areas. Considering that updated and comprehensive lists of all potentially eligible outlets were not routinely available at both national and sub-national levels, a cluster sampling approach with an outlet census was used to identify outlets for inclusion. A cluster was defined as an administrative unit ideally with a population of 10,000–15,000 inhabitants and this corresponded with a “location”. Using 2009 Kenya Population and Housing Census [13], a national sampling framework was constructed and survey clusters or locations were selected using the technique of probability proportional to population size.

The survey was powered to detect a minimum of a 10-percentage point change in availability of QAACT medicines within each strata at the 5% significance level with 80% power. The number of study clusters was calculated for each strata based on the required number of anti-malarial stocking outlets, assumptions about the number of anti-malarial stocking outlets per cluster and information from previous survey rounds including anti-malarial and QAACT availability, outlet density per cluster, and design effect. A total of 84 locations were sampled, this included 17 endemic locations, 22 epidemic-prone locations, 28 seasonal-transmission locations, and 17 low-risk locations. Within each sampled location, all outlets with the potential to provide anti-malarials or diagnostic testing services to patients or clients were screened for eligibility. In all sampled locations, the census boundary was extended to the higher administrative unit, the “division”, to allow for over sampling of public health facilities that are relatively uncommon at the location level but important outlet type in health service provision.

Data were collected between 7th June and 17th August, 2016 by 14 data collection teams. All fieldworkers attended a standardized training that consisted of classroom presentations, exercises and role plays as well as a field exercise. Additional training was provided for supervisors and quality-controllers that focused on field monitoring, verification visits, and census procedures. Data collection teams were provided with a list of sampled locations and official maps that illustrated their administrative boundaries. In each sampled location, fieldworkers conducted a systematic and full enumeration of all outlets.

Data were collected using a standardized ACTwatch outlet survey questionnaire and key informant interviews. Using the outlet survey questionnaire, the primary provider/owner of each potential outlet was invited to participate in the study and screening questions were administered to assess eligibility. Consenting providers were asked to show the interviewer all anti-malarials and malaria RDTs currently available. An anti-malarial audit sheet was completed to capture information for each unique anti-malarial product in the outlet, including formulation, brand name, active ingredients and strengths, package size, manufacturer and country of manufacture. Providers were asked to report the retail and wholesale cost for each medicine, as well as the amount distributed to individual consumers in the last week. Similarly, among the outlets found stocking malaria RDTs, an audit was completed to record information such as brand name, manufacturer, country of manufacturer, reported retail selling price and number of tests conducted or sold in the last 7 days for each of unique RDT product. Finally, a provider module was administered to assess provider’s knowledge and reported practices on malaria case-management policy recommendations. Outlet survey data were captured using Android phones fitted with customized forms created using DroidDB (© SYWARE, Inc., Cambridge, MA, USA). Interviews were conducted in local language using questionnaires that were translated from English to Swahili language and back to English to confirm translations.

The outlet survey protocol received ethical approval from Kenyatta National Hospital–University of Nairobi Ethics & Research Committee (Reference Number KNH-ERC/A/145). Provider interviews and product audits were completed only after administration of a standard informed consent form and provider consent to participate in the study. Standard measures were employed to maintain respondent confidentiality and anonymity, such as ensuring privacy during interviews, secure storage of completed questionnaires, and preventing any sharing of data between outlets. Providers had the option to end the interview at any point during the study.

Data were analysed using Stata version 13.1 (StataCorp College Station, Texas, USA). Standard indicators were constructed according to definitions applied across the ACTwatch project and have been described elsewhere [12, 14]. Descriptive analysis was undertaken, all point estimates and 95% confidence intervals were weighted to provide national estimates and calculated using Stata survey setting procedures to account for the complex clustered and stratified sampling strategy. The sampling weights were calculated as the inverse of the probability of cluster selection. Data were presented according to the four strata as well as by outlet types.

According to information on drug formulation, active pharmaceutical ingredients and strengths, anti-malarials were classified as non-artemisinin therapies, artemisinin monotherapies and ACTs. ACT medicines were further classified as either QAACT or non-quality-assured ACT (non-QAACT). QAACT was defined as ACT medicines that had World Health Organization prequalification status, ACT medicines in compliance with the Global Fund quality assurance policy, or ACT medicines granted regulatory approval by the European Medicines Agency. ACT medicines that did not meet requirements for QAACTs were categorized as non-QAACTs.

Availability of any anti-malarial was calculated with all screened outlets as the denominator. In the public sector, the availability of specific types of anti-malarials was calculated using the denominator of all screened outlets given that anti-malarials should be available at all public health facilities and among CHWs. Availability of specific anti-malarial categories in the private sector was calculated using the total number of private sector outlets stocking any anti-malarial as the denominator. Outlet “readiness” for malaria case management was defined as the extent to which an outlet had QAACT and malaria testing available.

This paper also presents market share and price indicators among different classes of anti-malarials. The measure of adult equivalent treatment dose (AETD) was deployed to analyse market share and price to allow for meaningful comparisons between anti-malarials with different treatment courses. The AETD was defined as the amount of active ingredient required to treat an adult weighing 60 kg according to World Health Organization treatment guidelines [15]. Provider reports on the amount of the drug sold or distributed during the week preceding the survey were used to calculate sales or distribution volume according to type of anti-malarial. All dosage formulations were included in the sales or distribution volume calculation to provide a complete assessment of anti-malarial market share to the consumer or patient. Volumes were therefore the number of AETDs sold or distributed by a provider in the 7 days prior to the survey. Additional public health facilities sampled as part of over-sampling for these outlet types were not included in market share calculations.

Price data were collected in Kenya Shillings and converted to United States Dollar based on official exchange rates for the data collection period. Anti-malarial price indicators are expressed as median unit cost for one AETD to allow for comparability between classes of anti-malarials. Only tablet formulations are reported due to the differences in unit costs for tablet and non-tablet formulations. The interquartile range (IQR) was calculated to demonstrate price dispersion.

Provider perceptions regarding the most effective first-line treatment was assessed by administering questions to the senior most provider at all anti-malarial-stocking outlets. Providers were asked to describe what medicine they believed was the most effective treatment for uncomplicated malaria in a child and in an adult.

A total of 17,852 outlets were screened for availability of anti-malarials and/or malaria blood testing services. Of screened outlets, 2291 were stocking anti-malarials or testing on the day of the survey or within the past 3 months, and 2271 were subsequently interviewed. A total of 1917 eligible and interviewed outlets were stocking anti-malarials on the day of the survey, 293 reportedly stocked anti-malarial(s) within the past 3 months while 61 were found stocking malaria diagnostics without anti-malarial products. A total of 6716 anti-malarial medicines and 846 RDT products were audited (Additional file 1).

Table 2 summarizes the availability of anti-malarials and diagnostics among the screened public sector outlets on the day of survey. Availability of any anti-malarial medicine was 91.8 and 2.4% among CHW, and this was highest in the endemic areas (99.0%) and lowest in the low-risk areas (81.0%). Among the screened outlets, 87.1% of public health facilities had QAACT medicines available and this varied by epidemiological zone: endemic areas (92.2%), endemic-prone areas (93.5%), seasonal-transmission areas (87.5%) and low-risk areas (77.0%). Public health facility availability of weight specific QA AL was variable: 66.4% for 6 tablet-pack QA AL, 63.3% for 12 tablet-pack QA AL, 37.0% for 18 tablet-pack QA AL and 72.7% for adult tablet-pack of QA AL (Additional file 2). Availability of non-QAACT medicines among all screened public health facilities was 12.3% and highest in the low-risk areas (17.8%). Overall availability of SP among the screened public health facilities was 17.6%, but this varied by epidemiological zone, where availability was 70.0% in endemic areas and less than 1% across all other epidemiological zones. Of all screened public health facilities, 46.0% stocked artesunate injection and this varied by epidemiological zone: endemic areas (59.8%), endemic-prone areas (63.9%), seasonal-transmission areas (48.9%) and low-risk areas (16.0%).

Of all screened public health facilities, 86.4% had capacity for malaria blood testing, more commonly through RDTs (69.7%) compared to malaria microscopy (44.2%) (Table 3). Malaria blooding testing was lowest among public health facilities in endemic areas (79.9%). Among all screened CHWs, availability of RDTs was 4.3% and was highest in the endemic areas (7.3%) and endemic-prone areas (8.0%), compared to the seasonal-transmission areas and low-risk areas where availability was less than 1%.

Among all screened public health facilities, 78.3% had both QAACT medicines and malaria blood testing available, and this varied by epidemiological zone: endemic areas (75.6%), endemic-prone areas (84.6%), seasonal-transmission areas (72.8%) and low-risk areas (81.4%) (Table 4). Only 8.8% of public health facilities had QAACT medicines available but no malaria blood testing, and this was highest in the endemic areas (16.7%).

Among all screened private sector outlets, availability of any anti-malarial varied by facility type: private-for-profit facilities (76.4%), registered pharmacies (93.3%), un-registered pharmacies (87.2%), and general retailers (2.4%) (Table 5). Among the anti-malarial stocking private sector outlets 46.7% had QAACT medicines, and this was highest among registered pharmacies (73.2%) and lowest among general retailers (9.9%). Private sector availability of QAACT medicines also varied by epidemiological zone: endemic areas (66.7%), endemic-prone areas (43.0%), seasonal-transmission areas (33.8%) and low-risk areas (42.2%) (Additional file 3). Availability of non-QAACT medicines was 37.9% in the private sector, and highest among anti-malarial stocking registered pharmacies (76.8%). Non-artemisinin therapy was stocked by 69.6% of the private sector and this was most commonly SP (57.6%). SP availability varied by outlet type, from 25.6% of private-for-profit facilities to 85.2% of general retailers.

The availability of diagnostic testing among anti-malarial-stocking private sector outlets was 20.8% (12.4% malaria microscopy and 12.4% RDT), and varied by outlet type: private for–profit facilities (66.9%), registered pharmacies (22.4%), unregistered pharmacies (12.1%) and general retailers (0.2%).

Figure 1 illustrates the anti-malarial market share (the amount of anti-malarials sold or distributed during the week preceding the survey) across the public and private sector, according to outlet type and by anti-malarial. In total, 70.6% of anti-malarials were distributed through the private sector. Most of the private sector distribution was through unregistered pharmacies (37.3%) followed by private for-profit health facilities (13.4%), registered pharmacies (12.8%) and general retailers (7.1%). Anti-malarial market share for the public sector was 29.4% and nearly all treatments distributed were QAACT medicines (25.6%). Within the private sector, 32.5% of the market share was accounted for by QAACT medicines, followed by non-QAACT medicines (15.0%) and non-artemisinin therapy (27.2%), which was typically SP (22.1%).

The private-sector market share of all anti-malarials varied across epidemiological zones (Fig. 2): endemic areas (59.2%), endemic-prone areas (68.8%), seasonal-transmission areas (81.8%) and low-risk areas (94.9%). QAACT medicines market share was highest in endemic areas where (74.9% of anti-malarials distributed were QAACT), and lower in other zones: endemic-prone areas (49.4%), seasonal-transmission areas (32.2%) and low-risk areas (37.9%). Market share for non-QAACT medicines was lowest in endemic areas (9.7%) and highest in low-risk areas (24.8%). SP market share was 14.2% in endemic areas, 27.5% in endemic-prone areas, 45.7% in seasonal-transmission areas and 33.9% in low-risk areas. SP was exclusively distributed by the private sector in endemic-prone and low-risk areas, and mostly by the private sector in seasonal-transmission areas (0.3% of the public sector and 45.4% of the private sector).

Figure 3 shows that 66.9% of malaria blood testing was performed in the public sector and within the sector the market share of malaria microscopy (35.6%) and RDTs (31.3%) was comparable. A third (33.1%) of blood testing market share was delivered through private sector outlets and testing was more commonly performed through malaria microscopy (21.6%) than RDTs (11.5%). Notably, majority of the private sector malaria tests were delivered primarily by private-for-profit health facilities (28.5%) and was rare among registered pharmacies (2.2%), unregistered pharmacies (2.4%) and general retailers (0%).

Table 6 summarizes the private sector median price for anti-malarials and blood testing. The median retail price for one QAACT AETD was $1.31 (IQR: 1.00–1.51). Non-QAACT AETD was $3.52 (IQR: 1.51–5.02). The median price for a SP AETD was $0.45 (IQR: 0.30–0.50) (Table 6).

In regard to the cost of malaria blood testing, an adult patient was required to pay a median cost $1.00 to receive a malaria test using microscopy (IQR: 0.50–1.51) or RDT (IQR: 1.00–1.00). Similarly, the price of testing for children, either with microscopy or RDT, was $1.00 [microscopy (IQR: 0.50–100) or RDT (IQR: 0.50–100)].

Figure 4 illustrates the extent to which providers perceived ACT as the most effective treatment for uncomplicated malaria in an adult. Nearly all providers (96.3%) in the public sector perceived ACT as the most effective treatment for uncomplicated malaria for adults. In the private sector, 64.1% perceived ACT to be the most effective, 16.4% perceived SP as the most effective, 9.2% cited another anti-malarial, and 10.2% did not know. General retailers and unregistered pharmacy providers were the most common outlet types to cite an anti-malarial other than an ACT or to report that they did not know what was the most effective treatment for malaria in an adult (81.1 and 14.8%, respectively). With regards to perceptions of the most effective anti-malarial for a child, a similar pattern was observed: 95.3% public sector cited an ACT, versus only 61.2% of private providers (Additional file 4).

Previous ACTwatch studies have demonstrated high availability of QAACTs and malaria blood testing since 2010 and 2014 respectively [16]. Nonetheless in 2016, one in five public health facilities did not have both QAACT and malaria blood testing available, pointing to the need to close this gap in order to achieve universal coverage. Furthermore, while availability of QAACT was generally high, the implementation of the AL policy includes delivery of four different AL pack sizes (6, 12, 18 and 24 tablets) suitable for management of four different weight categories of patients. Availability of the different weight categories was more variable among public health facilities, with less than 40% of outlets having the 18 tablet pack size in stock. Only half of the public health facilities stocked treatment for severe malaria, injectable artesunate, and this was similar to levels in 2014 [18]. SP for IPTp was available in 70% of public health facilities in endemic areas where it is recommended for IPTp, and rarely found among public health facilities outside of the endemic areas, illustrating alignment with national guidelines in these areas.

Some of the gaps found among public health facilities may be explained by changes in the supply and distribution of anti-malarial medicines and RDTs in the public sector. The Government of Kenya transitioned from a traditional central medical store using a push-distribution system to a pull-distribution system for malaria commodities, through the Kenya Medical Supplies Agency (KEMSA). However, the role of KEMSA to provide health commodities according to the requested needs (“pull” system) rather than allocated proportions of the total supply (“push” system) presented a number of challenges [16], and resulted in frequent stock-outs and erratic supplies of commodities to public health facilities. This has been attributed to limited capacity to quantify needs, insufficient budget, and challenges in maintaining a full pipeline of commodities to meet county specific needs [5]. As a means to prevent irrational anti-malarial medicines and RDT procurements and stock-outs within the public sector, more recently the National Malaria Control Programme and KEMSA have implemented a “smart push” system for malaria commodities. This strategy sets limits on the maximum quantity of malaria commodities that can be supplied to a facility, depending on the level of care and the epidemiological area. This strategy may help to further close the gap in availability of both testing and treatment in the public sector and will be important to safeguard the gains made nationally in malaria case management.

The reach of the public sector has been extended to the community-level through the training and equipping of CHW with malaria case management skills and supplies in endemic areas since 2012 [3]. The findings illustrate that a small portion of CHWs were ready for malaria case management and these were restricted to endemic and endemic-prone areas where interventions are being piloted. Low levels of QAACT and RDT availability even within these areas may be explained by the lack of a consistent public sector malaria commodity supply as previously discussed. Changes to the supply and distribution of anti-malarial medicines in the public sector has been described as contributing to delays in implementation of the malaria CHW strategy as public health facilities were either reluctant to give limited malaria commodities to CHWs or had no commodities to supply CHWs [5]. Maintaining a network of trained and equipped CHWs will be central to ensuring access to malaria case management services in rural areas. Key challenges to be addressed include gaps in CHW motivation and retention, training and maintaining supervision [17].

Consistent with previous studies [18], the private sector plays an important role in malaria case management given almost three-fourths of all anti-malarials were distributed through this sector. While unregistered pharmacies comprised most of the market share, other private facilities also contributed to the anti-malarial distribution, including general retailers. As national policy stipulates that only private for-profit health facilities and registered pharmacies are licensed to sell medications, the results from this study suggest many anti-malarial medicines are being distributed through unregulated sources. Indeed, the Kenya Pharmacy and Poisons Board stipulates that for outlets to legally administer medicines, including anti-malarials, they must be registered with the board. In addition, medicines may only be dispensed under the supervision of a certified pharmacist or pharmaceutical technologists [19]. However, given the widespread distribution of anti-malarials through unregistered pharmacies and general retailers, strategies are needed to engage with these outlet types. One option may be to permit the licensing of unregistered pharmacies through accreditation programmes, or through partnerships with the public sector, as a means to increase access to appropriate malaria case management services. Evidence from other countries has demonstrated that the registration, training and supervision of the private sector, can lend to improvements in malaria commodity ACT availability, distribution and provider performance [20‐23].

Among private sector outlets in the business of anti-malarial distribution, fewer than half had QAACT available in 2016, reflecting a decrease from levels reported in 2011 (60%) and 2014 (71%) [18]. Although the private sector CPM for QAACT administered by the Global Fund continued through 2016 in Kenya, important changes and challenges in the post-pilot period are likely to have contributed to the decline in availability. Since the AMFm, Kenya transitioned from a dedicated donor funding to a country specific grant funding mechanism, which was further amplified by a reduction in funding for co-paid ACTs. Indeed, the number of subsidized QAACT doses delivered to Kenya’s private sector in 2015 through the CPM was half of what it was in 2012 [9]. In this context, lower availability of QAACT in the private sector as compared to previous rounds can largely be explained by a more limited supply and availability of these anti-malarials given reduced funding.

Only one in five private providers stocking anti-malarial medicines had confirmatory testing available and this was highest among private for-profit facilities, reflecting national policy which only permits these types of facilities to administer confirmatory testing. While the private sector distributed the majority of anti-malarial medicines, only one-third of malaria blood tests were performed by private outlets, suggesting that presumptive treatment is common. Permitting other private sector outlet types to administer diagnostic testing may be one means to increase access to confirmatory testing as evidenced by a number of private sector RDT pilot initiatives [24‐26]. Studies from Kenya have shown that registered pharmacies participating in a RDT pilot almost always correctly managed negative patients as per the government recommended treatment algorithm [11]. In addition, there were few differences in the performance of pharmacies and private for-profit health facilities, lending to the author’s conclusion that malaria diagnosis through RDTs can take place safely and effectively across a variety of channels. Given the role of unregistered pharmacies in Kenya, these facilities in particular may be critical to increasing access to confirmatory testing prior to treatment, though future research is merited to assess the feasibility of introducing RDTs among these outlet types. This is particularly pertinent as Kenya is deciding whether to legalize and/or subsidize RDTs in registered pharmacies as part of its malaria control strategy.

Most of the anti-malarial market share comprised of QAACT, while one in four anti-malarials distributed were SP—and most commonly distributed through the private sector. Market share for QAACT was highest in endemic areas as compared with other epidemiological areas where distribution of SP was relatively common. In seasonal-transmission areas, SP accounted for nearly half of all anti-malarial distribution with nearly all of these treatments going through the private sector. SP accounted for about one-third of anti-malarials distributed in endemic-prone and low-risk areas. This finding suggests that SP is being used for malaria case management within the private sector and outside of endemic areas and points to the need to ensure continued availability of affordable QAACT in the private sector across the country, and not just focused in endemic areas. Furthermore, these findings suggest that unless effective private sector strategies are identified and implemented, further decline in private sector availability may be observed. In absence of suboptimal private sector engagement and support, and social and behaviour change communication for providers and consumers to raise awareness of QAACT, Kenya may see an increasing market share for SP.

Several reasons may explain the ongoing distribution of SP for case management. This may be in part driven by price as QAACT was approximately three times more expensive than SP in the private sector. While the price of SP has remained consistently low since 2010, the price of QAACT has fluctuated, initially dropping to nearly the same price as SP in 2011 but then increasing by 2014 [18]. The persistence of inexpensive SP may inhibit the uptake of QAACT even with a subsidy mechanism in place, and especially in the context of a reduced subsidy since the AMFm period. Furthermore, in the absence of supportive interventions to increase awareness of subsidized ACT (which ceased in 2015), providers and consumers may have less awareness of the subsidized ACT medicine. Indeed, the results also illustrate how some private sector providers perceived non-ACT medicines as the most effective treatment for uncomplicated malaria. Finally, while availability of QAACT was low in the private sector, the data also illustrate how there was variability according to different pack sizes, perhaps causing providers to ration ACTs. Other evidence has suggested that providers will administer other artemisinin therapies over ACT if supply is uncertain and availability is lower than non-recommended treatments [27]. Additional research to understand provider and consumer practices and demand for SP will be valuable to provide insights into the ongoing distribution.

Nearly one in five anti-malarials distributed were non-QAACTs. Non-QAACT was about 2.5 times more expensive than QAACT in the private sector. This raises the question of why consumers would pay more for non-QAACT when less expensive QAACT are available? Other ACTwatch research has shown that non-QAACT products were primarily distributed by pharmacies and drug stores in urban areas of Kenya. Thus the distribution of non-QAACT may reflect a higher purchasing power of urban consumers [28]. More research is needed to understand the reason for provider dispensing practices of non-QAACT as well as consumer preferences. Addressing the availability and distribution of non-QAACT will also require addressing elements of the supply chain, as well a product registration, manufacturing, importation laws and regulation.

General limitations of the study design which are applicable to all ACTwatch outlet surveys have been described in detail elsewhere [12, 14, 29]. In particular, the study is descriptive in nature and the results are unable to provide analytical information on the causal effects of interventions on the malaria testing and diagnostic landscape.