A cross-sectional survey was carried out between January 2011 and June 2011 in the Kintampo Health Research Centre (KHRC) study area located in 8 neighbouring districts, made up of Kintampo North, Kintampo South, Nkoranza North, Nkoranza South, Techiman North, Techiman South, Tain and Wenchi in the central part of Ghana. These districts are predominantly rural with close to 80 % of the resident population living in rural communities. The tribal groups mainly consist of Bono and Mo origins with a significant part of residents having migrated from the northern parts of the country. The occupation is mainly subsistence farming of crops such as yam, cassava and maize . Malaria is the disease of highest burden in the area. [21–23].
The health systems in these districts are largely basic and modelled so that they follow a uniform pattern used in district health systems in Ghana. The lowest level (referred to as level A) of health care is at the community level and manned by Community Health Nurses. The next level is Health Centre level (referred to as level B), where middle level health professionals (physician assistant, midwife, nurses, laboratory and dispensary technicians are in charge of the system. The district hospital is level C and employs senior-level health professional including physicians, anaesthetists, senior nurses/midwifes, pharmacists and laboratory technologists. Patients presenting at the lower level with any condition that cannot be managed are referred to the next level for management; with cases beyond district hospitals referred to regional hospitals or teaching hospitals. Health seeking by most community members needing treatment, however, does not always follow this conventional arrangement as home treatment of malaria using drugs purchased from licensed chemical sellers is the commonest health seeking behaviour [24–26].
The study area consisted of 237 medicine outlets (MO); these comprised 37 health facility dispensaries, eight pharmacies and 192 license chemical shops. Mobile vendors and some grocery shops, drinking bars and homes were other sources of medicine sales; these were mostly accessible to community members since they are located within the community. The quality of drugs are to be regulated by the Ghana Food and Drugs Authority in collaboration with the Pharmacy Council of Ghana and various security agencies (including police, Customs Excise & Preventive Services among others) are charged with ensuring that only quality medicines are available on the market.
Medicine outlet selection
The Kintampo Health and Demographic Surveillance database, which has geolocations of all communities, MOs and health facilities, was used to create MO clusters. Cluster A included MOs within a 5-km radius from a district/municipal hospital, cluster B included MOs within 5–10 km radius from a district/municipal hospital and cluster C included MOs >10 km from a district/municipal hospital. Within each cluster, 10 MOs were randomly selected if the number of known MOs was less than forty and 20 MOs were randomly selected if the number of outlets was 40 or higher. All pharmacies were, however, purposively selected for medicines to be purchased due to their limited number  in the studied area. All mobile vendors and other outlets in the selected communities were included for medicine procurement. Medicines outlets were stratified by clusters as defined above and 141 of these outlets were randomly selected from a total of 237 medical outlets within the study.
A “mystery shopping” strategy was used to buy ACTs from the selected shops . Trained mystery shoppers were made to visit each selected MO, present themselves as a client with malaria symptoms and engage the shop attendant in a discussion regarding what antimalarials were available for sale to clients. From the list communicated, the mystery shopper purchased some of the ACTs available in stock and then made a mental note of all other ACTs that were available and could be obtained from the MO. The information was then relayed to another mystery shopper who later visited the shop to purchase the other brands of ACTs available in the shop. The process of mystery shopping was repeated in all identified shops until samples of all brands of ACTs available in each selected shop were obtained. The trained shoppers carried through this process in such a way that there was never any suspicion by the sellers.
Mystery shoppers wrote down the description of the outlet on exiting the MO. ACTs purchased were transported to the Kintampo Health Research Centre Pharmacy on daily basis. All samples purchased were stored as per the manufacturer’s instructions by the study pharmacist.
The Regional Medical Store was visited to ascertain the state and storage conditions of ACTs that were in stock.
Samples were shipped within two months to the bioanalytical facility at the London School of Hygiene and Tropical Medicine (LSHTM) for analysis. The packaging on each sample was scanned electronically and/or photographed at LSHTM and a detailed description of the ACT was recorded on a Microsoft Excel spreadsheet. The description included brand name, stated active ingredients, dose form, date of purchase from MO, name of manufacturer, country of origin, batch number, expiry date, number of tablets per packet and retail price paid. The database created together with the scans of the packets and their contents was shared with the Ghana team.
Chemical content analysis and classification of samples
Laboratory assessment of the quality of ACTs purchased in the central part of Ghana was carried out at the London School of Hygiene and Tropical Medicine (LSHTM) and the Georgia Institute of Technology (GT). Sample packaging and appearance were assessed, coded and recorded, as well as the weight, and dimensions of each sample. The amount of active pharmaceutical ingredient (API) in each sample was determined using high performance liquid chromatography (HPLC) and the presence of any compounds other than the stated APIs was screened by ambient mass spectrometry (MS) following published methods . Formulations were analysed for the amount of each active pharmaceutical ingredient (API) present using high performance liquid chromatography (HPLC). Briefly each tablet was pulverised, dissolved in solvent depending on the stated API; artesunate (AS), artemether (AM), dihydroartemisinin (DHA), amodiaquine (AQ) and sulfadoxine/pyrimethamine (SP) were dissolved in methanol; samples containing lumefantrine (LUM) were dissolved in 10 % acetic acid in methanol, and piperaquine (PIP) in methanol/0.1 M HCl (1:1; v/v). Solvent extracts were sonicated followed by centrifuging, and the supernatant injected into the HPLC column to determine the amount of API present.
HPLC analysis was conducted using a Dionex Ultimate 3000 system (Thermofisher, Hemel Hempstead, UK) and separation achieved using a GENESIS AQ 4 μm column (150 × 4.6 mm, Grace Materials Technologies, Cranforth, UK). The mobile phase was a gradient of ammonium formate (10 mM, pH 2.7) and acetonitrile (v/v; 60:40 to 85:15 over 7.0 min). A photo-diode array unit (UV-PDA; DAD 3000) was set at 204 nm for the artemisinin derivatives (AS, AM, DHA), 360 nm for PIP, AQ and LUM. In all cases, the flow rate used was 1.0 ml/min. Calibration curves of each compound were generated by Thermofisher Scientific Dionex Chromeleon 7.2 chromatography data system (CDS) software using known amounts of the corresponding chemical standards. Reference standards of artemisinin, artesunate, artemether, dihydroartemisinin, amodiaquine dichlorodihydrate and pyrimethamine were purchased from Sigma Aldrich, UK. Lumefantrine was purchased from WHO, Switzerland. Results were expressed as a percentage of the stated amounts of API on the package.
Quality of ACT was assessed by comparing the amount of API detected with that stated on the packaging label and indicated as a percentage of the stated value.
We adopted a range between 85 % and 115 % of the stated API content for both compounds in the ACT to classify samples as being of acceptable quality. Medicines with less than 85 % or over 115 % of the stated API content of any of the partner compounds were therefore classified as substandard.
Data management and analysis
The questionnaires completed by mystery shoppers as well as basic data collected by the pharmacist were double-entered, range and consistency checks and verification were carried out using Microsoft Access. The sampling was randomised and all data were cleaned and analyzed using STATA version 11 (Stata Corp., TX USA). Simple descriptive analysis such as proportions are used to summarize various variables including the prevalence of substandard drugs. Bar graphs were used to summarize some results such as sources of ACTs and country of their origin using Microsoft Excel.