Background
Financing for malaria control has increased substantially over the last decade, facilitating significant progress towards international targets for prevention and treatment. Increased coverage of at-risk populations with vector control as well as effective case management with artemisinin combination therapy (ACT) is contributing to substantial reductions in malaria cases and deaths [
1]. The spread of artemisinin resistance in
Plasmodium falciparum malaria parasites would threaten recent malaria control progress across endemic countries. Alternative anti-malarial medicines with equivalent levels of efficacy are not expected to become available for at least seven to eight years.
P. falciparum resistance to artemisinin derivatives has already begun to emerge; the first case was confirmed in Cambodia, near the Thai border (Pailin province) in 2009 [
2,
3].
Factors believed to be contributing to emerging drug resistance in Cambodia include the unregulated sale of artemisinin monotherapies for over 40 years; limited access to ACT; co-blistered ACT which is not co-formulated (facilitating continued use of artemisinin monotherapy); and ubiquitous counterfeit and substandard drugs [
2]. Cambodia's resistance containment programme consists of a number of interventions to facilitate early diagnosis and appropriate treatment of malaria. These include a ban on the sale of artemisinin monotherapies introduced in 2009; ongoing efforts to strengthen capacity for drug quality monitoring; a new bureau for policing private drug sellers; and active efforts to close unlicensed pharmacies [
2,
4,
5]. To facilitate diagnosis and treatment with national first-line drugs, these services are made available at community level through a village malaria worker (VMW) programme implemented in remote provinces [
2,
6]. Co-blistered artesunate + mefloquine (ASMQ) is the first-line treatment for
P. falciparum and chloroquine is currently the first-line treatment for
P. vivax infections. In late 2008, the first-line drug in districts with confirmed multidrug resistance was changed to dihydroartemisinin + piperaquine (DHA+PPQ); public outlets began stocking the drug in 2009 [
2].
Other national malaria control efforts include provision of highly subsidized rapid diagnostic testing (RDT) and ACT treatment in the private sector. The international NGO, Population Services International (PSI) has led private sector malaria treatment in Cambodia since 2003. Private sector ASMQ is sold under the brand name Malarine and accounts for 75% of the ASMQ distributed nationally. Rapid diagnostic test kits (RDT) are sold under the brand name Malacheck. At the time of data collection for this study, Malacheck tested for
P. falciparum infections. In 2010, the diagnostic kit changed to test for both
P. falciparum and
Plasmodium vivax infections. Malarine and Malacheck are sold in over 1,700 outlets, including private hospitals, clinics, pharmacies, mobile providers and drug stores [
4]. In the public sector, ASMQ is available under the name A+M.
Malaria control in Cambodia is of high significance globally because of the risk of
P. falciparum resistance to artemisinin, which could develop and spread to sub-Saharan Africa, crippling malaria control programmes in high-burden countries. Despite its importance, national level data on the anti-malarial market and on household treatment-seeking behaviour are limited. Prior to this study, national data collection efforts were undertaken by the national malaria control programme (National Centre of Entomology, Parasitological and Malaria control, CNM) in 2007 [
7]. Given a dynamic policy and programming environment, the need for national-level data that can continually inform targeted malaria control and drug resistance containment efforts is important.
This study uses data collected in 2009 as part of the
ACTwatch research programme [
8].
ACTwatch is designed to provide a comprehensive picture of the anti-malarial market to inform national and international drug policy evolution [
9]. Cambodia is one of seven
ACTwatch study countries. Nationally-representative outlet and household surveys were conducted to examine the supply and demand side of the anti-malarial market. Data from this study complement existing data on malaria treatment in Cambodia by providing a comprehensive picture of the anti-malarial market, including the availability, price and volumes of various anti-malarials moving through both the public and the private sector. This study also presents national data on household treatment-seeking behaviour for "malaria fever." Results can be used as part of monitoring access to and demand for effective combination treatment, as well as availability and consumption of artemisinin monotherapies. Information on consumer and provider behaviour can inform communications aimed at changing behaviour in the context of increased access to diagnostic testing and ACT.
Methods
This study uses data from two nationally-representative cross-sectional surveys. A survey of outlets with the potential to sell or distribute anti-malarials was conducted in June through July, 2009. A household survey was conducted in October through November, 2009. Data collection coincided with the rainy season during which malaria transmission occurs; this season generally begins in late May and ends in December.
Design and sampling
Both outlet and household survey sampling began with drawing a sample of 38 administrative clusters (health centre catchment areas) selected with probability proportional to size (PPS) from stratified lists of the 255 clusters in malaria endemic areas of Cambodia. Equal allocation stratification was utilized to allow for separate estimates across two malaria-endemic strata: areas with confirmed or suspected multidrug resistance and areas without confirmed/suspected multidrug resistance. A census of all outlets in these clusters was completed. To allow for comparisons between outlet types, public health facilities were also sampled from a larger administrative "booster" area (district) in which the selected cluster was located. A total of 7,518 outlets were visited. In the public sector, these included public health facilities (government referral hospitals, health centres and posts, N = 211) and village malaria workers (VMW, N = 203). VMWs are stocked with RDTs and anti-malarials and provide testing and treatment to the community free of charge. Government health facility staff support VMWs with initial training, monthly supervision and stock. Private sector outlets include nationally registered pharmacies/clinics (pharmacies, clinical pharmacies, cabinets, and private clinics, N = 152); informal drug stores (N = 164); grocery stores (shops located in urban areas, N = 1,170); village shops (shops located in rural areas, N = 5,144); and mobile providers (N = 383). Mobile providers are found predominately in rural areas and serve the communities in which they reside. They typically possess medical training and are often currently or previously employed by a public or private sector facility. Mobile providers interviewed in the outlet survey provided answers with respect to their drug stock and practices operating outside the realm of any other outlet that they are affiliated with. However, some mobile providers reportedly offer diagnostic testing using microscopy as part of their practice, in which case they take blood slides from patients in their mobile provider practice and use a facility laboratory for diagnosis.
Within the 38 clusters selected with PPS as first stage, a second stage of selection was used to select six census enumeration areas per cluster with PPS. Within each census enumeration area, 100 households were randomly selected for screening. The number of households in each census enumeration area was divided by 100 to derive a skip interval for random systematic sampling of households. A total of N = 22,317 households were asked a series of screening questions concerning fevers occurring among household members in the previous two weeks. While N = 14,306 recent fevers (krun kdao) were identified, interviews were conducted only among the N = 1,617 people with recent self-reported "malaria fever" (krun chanh/krun gnak). Most "malaria fevers" included in this study had occurred among men (62%) and people age 15 and above (71%).
Materials
The core component of the outlet survey instrument is an exhaustive audit of all anti-malarials in stock at the time of the survey. For each anti-malarial in stock, identifying information including brand, generic and manufacturer names and drug formulation and strength was collected. Provider reports on unit cost and amount sold or distributed within the last week was also recorded. The survey additionally measured availability and price of microscopic and rapid diagnostic blood testing. Basic outlet and provider characteristics were collected and provider knowledge surrounding first-line treatment was assessed.
The household survey instrument collected detailed information on treatment-seeking behaviour, including type, timing, source, and price paid for diagnostic testing and drugs acquired for fever. Respondent recall and recognition of the type of treatment acquired was aided by the use of a comprehensive anti-malarial field guide with photographs and brand names of common anti-malarials available in public and private sector outlets. This field manual included pictures of anti-malarial tablets that are not part of pre-packaged therapies. Where respondents reported use of a combination of drugs - or a multi-drug cocktail, the anti-malarial field guide was used to help the respondent identify anti-malarials contained within the cocktail, where applicable. Non-anti-malarial cocktail contents (e.g. vitamins, antipyretics) were identified by respondent recall, unaided by a pictorial guide.
The household survey was administered to people age 15 and above with "malaria fever," and to children's primary caregiver where the person with fever was below age 15. In addition to fever management questions, respondents were asked a series of questions assessing their attitudes and knowledge regarding malaria diagnosis and treatment. A household questionnaire module, modelled after the Demographic and Health Survey (DHS) [
10], collected information on household characteristics and household assets to be used in assessment of relative socioeconomic status.
Training and fieldwork
Data collection teams received a six-day training focused on administration of the questionnaire and sampling procedures. The outlet survey was conducted in June and July, 2009. Household survey data were collected from people with "malaria fever" during the two weeks preceding the survey (or their primary caregiver if under the age of 15) during the months of September, October and November, 2009, falling within the peak malaria transmission season. All questionnaires were reviewed by the team supervisor and spot checks were conducted for at least 20% of all outlets and households. Microsoft Access (©Microsoft Corporation, Seattle, WA, USA) was used for double data entry and validation. All research activities operate under ethical approval granted by the Cambodian national ethics review board.
Measures
Anti-malarials identified during the outlet drug audit were classified according to information on drug formulation, contents and strengths with supporting information including brand or generic name and manufacturer. All outlets visited were classified as having at least one anti-malarial in stock at the time of the survey or not. Cotrimoxazole and medicines intended solely for malaria chemoprophylaxis were not categorized as anti-malarials. Among outlets stocking anti-malarials, variables were created to indicate stocks by type, including the broad category of ACT, as well as specific categories including all ASMQ, private sector ASMQ sold as Malarine, public sector ASMQ distributed as A+M, oral artemisinin monotherapy, and non-artemisinin monotherapy. Availability of diagnostic testing services was assessed among outlets stocking anti-malarial(s) on the day of the interview, or that reportedly stocked anti-malarial(s) within the past three months. Outlets were categorized according to availability of rapid diagnostic tests (in stock on the day of the interview) and microscopic blood testing (availability of services). Outlet possession of laboratory equipment in working order was not assessed. Provider reports on the availability of microscopic blood testing services - which could entail taking patient blood slides to another site for testing (particularly relevant in the case of mobile providers) - were used to classify availability of microscopic blood testing.
To calculate volumes of anti-malarials reportedly sold or distributed in the week preceding the survey as well as drug prices, drug courses were standardized using adult equivalent treatment doses (AETD). One AETD was defined as the amount of the drug needed for a full course treatment based on guidelines from the WHO where available. Where unavailable from the WHO, peer-reviewed literature was consulted, and if necessary, manufacturer guidelines were utilized. Provider reports on drug prices per unit (e.g. tablet) and amount of the drug sold or distributed during the week preceding the survey were used to calculate volumes and price according to type of anti-malarial: Malarine, A+M, other ACT, oral artemisinin monotherapy, non-oral artemisinin monotherapy and non-artemisinin monotherapy. The volume of each drug is therefore the number of AETDs that were reportedly sold/distributed during the week preceding the survey. The price of each drug was calculated for one AETD. Audited anti-malarials with missing data required to calculate AETD - specifically drug strength - were excluded from volumes and price estimation.
Household survey indicators of treatment-seeking behaviour and treatment of fever were constructed from respondent reports on sources where treatment was sought; whether or not the person with fever received a diagnostic test; self-reported diagnostic test result; source of diagnostic test and treatment(s); type of treatments acquired (brand names); and timing of treatments. Brand names were used to categorize drugs according to generic anti-malarial types (e.g. chloroquine, quinine, ASMQ). These were then further classified as ACT, artemisinin monotherapy, or non-artemisinin monotherapy. Indicators were calculated using the three classes of anti-malarial above, as well as an overall category for any anti-malarial. These anti-malarial drug categories do not include anti-malarials that were taken as part of a drug cocktail. A separate category was created to indicate cocktail treatments that contained an identifiable anti-malarial. Another drug category was created for people who received a cocktail of drugs that did not contain any identifiable anti-malarial.
Sources for diagnosis and treatment were categorized as either public or private sector. Public health facilities, non-profit health facilities and village malaria workers (VMWs) were classified as public sector. The private sector encompasses outlets with providers who have formal training (pharmacies, clinical pharmacies, cabinets, private clinics, mobile providers) as well as providers who do not generally have formal training (drug stores, grocery stores, village shops). Sector in which a person with "malaria fever" sought treatment was defined as public only, private only, or a mix of public and private outlets.
Respondent knowledge and beliefs about the first-line treatment, ASMQ, was assessed through two items. One dichotomous variable indicates whether or not the respondent names ASMQ (generic or a brand name) as an anti-malarial drug. A second measure captured respondent beliefs on the most effective treatment for adults with malaria through an open-ended question that was re-coded into categories including ASMQ and other ACT.
Socioeconomic status was assessed at the household level relative to other household respondents, using measures of housing, water, sanitation and household asset items modelled after the DHS. Wealth index items were assigned a weight through principal components analysis and standardized in relation to a standard normal distribution. Each respondent was categorized according to their household's socioeconomic score, and recorded into one of five wealth quintiles, ranging from lowest/poorest to highest/least poor [
11].
Data analysis
Frequencies were tabulated for availability of anti-malarials and diagnostic testing by outlet type. Volumes of anti-malarials sold or distributed during the previous week was calculated by considering the fraction of overall volumes accounted for by the number of AETDs sold/distributed per anti-malarial drug type (public sector ASMQ - A+M, private sector ASMQ - Malarine, other forms of ACT (dihydroartemisinin+piperaquine - DHA+PPQ, or artemisinin+piperaquine - A+PPQ), oral artemisinin monotherapy, non-oral artemisinin monotherapy, non-artemisinin monotherapy). Frequencies of AETDs reportedly distributed for free in the public sector were calculated across drug types: the most popular non-artemisinin monotherapy, chloroquine; oral artemisinin monotherapy; Malarine; A+M; and other forms of ACT. Median price and interquartile range of AETDs sold in the private sector were calculated across these drug types. Median price was also calculated for RDTs and microscopic blood testing available in the private sector.
At the household level, frequencies were tabulated for source of initial treatment, sector in which treatment was sought (for those who sought treatment outside of the home), and source of diagnostic testing and treatments. Frequencies were also tabulated for number of treatments acquired for one episode of "malaria fever," and type of treatments acquired. Logistic regression was used to test for an association between treatment-seeking sector (public only, private only, public and private) and: 1) diagnostic testing among all people with fever; 2) anti-malarial treatment of self-reported positive cases, self-reported negative cases, and undiagnosed "malaria fevers" for which treatment was sought in the public and/or private sector. Logistic regression was also used to test for an association between household socioeconomic status and diagnostic testing among all people with "malaria fever." Odds ratios and 95% confidence intervals are reported.
Household data were weighted to account for difference in the probability of being selected in the different strata. Standard error estimation accounted for clustering at the health centre catchment area and enumeration area levels. Given the use of a census approach, sampling weights were applied to outlet survey data based on the inverse probability of selection to account for differences in strata and cluster sizes and the oversampling of booster outlet types. Outlet distribution was assumed to be proportional to population size. Stata 11.0 (©Stata Corp, College Station, TX, USA) was used for all analyses.
Competing interests
Henrietta Allen is the Malaria Programme Technical Advisor for PSI/Cambodia.
Authors' contributions
KOC designed the ACTwatch Cambodia household and outlet surveys and data collection instruments. DC made contributions to the study design. ML and HG are responsible for the particular analyses presented in this paper. HG and SP made contributions to field work, data cleaning and analyses presented in ACTwatch outlet and household study reports. TS, SY and AS assisted with interpretation of findings. All authors read and approved the final manuscript.