Study design and setting
Kisumu County is part of the endemic lake region, whose malaria prevalence in children under 14 years was 26.7% in 2015  and 19% in 2020 . Between 2018 and 2019, a total of 1804 health facilities in the lake endemic counties reported testing a total of 12.8 million outpatients, with about 6 million of the tests turning positive (test positivity rate 46.7%, mean 309 cases in primary health facilities) in the 24 months [17, 18]. A survey conducted in schools in Kisumu County revealed that each person experiences at least 31 infective bites annually .
The study was a cross-sectional survey. Private drug outlets were sampled from Kisumu Central and Seme sub-counties, and an assessment was carried out on the adherence to malaria treatment guidelines from May to December 2021. Knowledge and practices on malaria case management were evaluated using a developed tool with established thresholds based on specific indicators (Fig. 1).
Kisumu County has over 120 government health facilities, 17 faith-based health facilities, 17 non-governmental organization facilities, and over 50 private health facilities [20, 21] is divided into seven administrative units called sub-counties (Fig. 2). Seme sub-county has the county’s highest population-weighted mean malaria test positivity rate (about 60%), while Kisumu Central sub-county has the lowest at around 35% . However, Kisumu Central hosts the more significant part of the cosmopolitan city of Kisumu and most of the region’s most extensive public and private health facilities . In 2020, there were over 1.8 million and 110,000 new outpatient department attendances and admissions, respectively. Of the 637,298 suspected malaria cases, 336,302 (53%) were confirmed, resulting in 1373 deaths (case fatality rate 0.41%). Children under five years comprised over a quarter of the 3126 confirmed severe malaria admissions, with 15 fatalities reported . The number of cases is usually highest around July .
Insecticide-treated mosquito net (ITN) coverage is 70%, while Intermittent Preventive Therapy in Pregnancy (IPTp) coverage is 35.3%; the mean number of doses of IPTp received by pregnant women in Kisumu County is 3.8, with over 40% receiving at least five doses . A challenge with data on malaria is that reporting rates among private health facilities are very low . The county experiences bimodal rainfall distribution, with the long rains falling between March and June while short rains fall between September and December. Temperature varies with altitude and season, ranging between a minimum of 9 °C and a maximum of 35 °C .
Study population and sampling technique
The study population consisted of staff in stand-alone private drug outlets (registered and unregistered) in Kisumu Central sub-county and Seme sub-county. A comprehensive list of potentially eligible drug outlets is not available in the national or county databases, so the identification of outlets for inclusion was carried out through an outlet census. The sub-county pharmacists were contacted for clues on drug outlets’ locations. A team consisting of the researcher and three research assistants was then deployed to both sub-counties to locate the mentioned outlets. Each outlet’s name and GPS coordinates were taken, and the available staff asked where the neighbouring outlets were. After the location of all the mentioned outlets, the team conducted a mop-up exercise by identifying all the private drug outlets by ward and estate/village. A total of 85 outlets met the criteria for selection. Due to the relatively small number of outlets, all of them were considered for inclusion in the study without sampling. However, only 70 outlets consented to participate. With the outlets included, it was possible to measure the prevalence of adherence to treatment guidelines to within five percentage points of the actual value with a confidence of 95%.
Inclusion and exclusion criteria
Private drug outlets that diagnose malaria and sell anti-malarial medicines on retail terms.
Private drug outlets that fill prescriptions for malaria treatment from other health facilities.
Private drug outlets owned and operated jointly with medical laboratories or clinics.
Data collection methods and data quality control
Standard indicators were constructed according to definitions used by the Ministry of Health as captured in the Kenya Malaria Monitoring and Evaluation Plan. Under the KMS’s second objective (to manage 100% of suspected malaria cases according to the Kenya malaria treatment guidelines by 2023), some of the indicators used are the proportion of suspected malaria cases tested with RDT and microscopy and the proportion of suspected malaria cases managed according to the guidelines . Adherence to the guidelines was assessed by evaluating testing or asking to see laboratory results, selecting appropriate treatment, and patient counseling.
Outlet staff interview
Study data collectors introduced themselves to the outlet staff and sought an audience with the superintendent or person dispensing drugs. Permission to collect data was sought from the outlet’s owner/manager/director as applicable in case of outlets where the staff did not have the authority to allow access to such information. In such a case, the interviewer provided the information contained in the consent form to the owner/manager/director to ensure clarity and avoid suspicion. After obtaining consent for both the interview and the mystery client, the interviewer started administering the first question of the survey questionnaire but was interrupted by the entrance of the mystery client (study staff), as described below.
After the exit of the mystery client, the interviewer proceeded with the interview, collecting information on the outlet’s drug dispensers’ demographics, pre-service training, access to guidelines, retrospective exposure to in-service training and supervision, availability and storage of various anti-malarial drugs, the availability of malaria diagnostic services, and the availability of appropriate basic equipment. Appropriateness of documentation was assessed by verifying the availability of ledger books or other inventory records (physical or electronic), copies of delivery notes and invoices, and pharmacovigilance reporting forms (pink and yellow forms) or proof of previously submitted reports (paper or electronic). The outlet assessment was conducted using both direct observation and interview.
Mystery client technique
To reduce suspicion and minimize information bias (Hawthorne effect), the study staff pretending to be a client (mystery client) presented immediately after consent had been obtained by the drug outlet staff interviewer, posing as a client (mystery client), and the interviewer excused themself and moved away to allow the mystery client to consult and be attended to privately. The mystery client helped check the usual practice at the drug outlet in terms of diagnosis, choice of medicine, and dispensing practices using a standard case scenario. The standard mystery client script described a 4-year-old child at home with malaria. If asked for the symptoms, they mentioned hotness of body, refusal to feed, and generally sick-looking over the previous two days. The mystery client had a copy of positive malaria microscopy results indicating malaria parasites (MPS + +) seen. This was only presented if asked for by the staff. Interviewer bias was minimized by advanced training of the research assistants on the case scenario and the indicators to look out for.
After the encounter, the information was recorded on a standard tool designed to answer critical dispensing and counseling processes observed during the visit.
Variables of the study
Adherence to malaria treatment guidelines was defined as dispensing the right malaria medicines only to a client with evidence of a positive malaria test result from microscopy or RDT, accompanied by the provision of appropriate instructions on the use of the medicines. This outcome was assessed based on the mystery shopper tool.
Secondary outcomes included appropriate drug outlet infrastructure, adequate stocking levels, staff skills in malaria case management, and good record-keeping practices. These outcomes were assessed based on the drug outlet staff tool.
Data processing and analysis
Continuous variables were summarized using central tendencies and dispersion measures, while categorical variables were summarized using frequencies and proportions. The Chi-square test was used to assess the association between dependent (adherence to treatment guidelines) and independent variables (provider socio-demographics, other provider characteristics, outlet characteristics, and health system factors) calculated for the adherence to treatment guidelines. Any variable with a p-value of less than 0.05 at the bivariate level was subjected to multivariable logistic regression. All variables with a p-value less than 0.05 at the multivariate level were regarded as independently associated with adherence to malaria treatment guidelines. Analysis was done using Microsoft Excel and EpiInfo software.