Assessing the quality of anti-malarial drugs from Gabonese pharmacies using the MiniLab^®: a field study

This medicine quality field survey was registered in advance (30 Dec 2013) in The Netherlands Trial Registry (NTR): NTR4341 [26]. This report follows, where appropriate, the Medicine Quality Assessment Reporting Guidelines (MEDQUARG) [27, 28]. Also, the costs of this study are reported [29] (Additional file 2).

Scientific clearance was obtained from the Scientific Review Committee (SRC) of the Centre de Recherches de Médicales de Lambaréné (CERMEL), Albert Schweitzer Hospital (SRC number: 2013.11; Additional file 3). The Ethical Committee of CERMEL decided that ethical approval of this study was not required as this study is a quality assurance in healthcare study, no humans having been subjected to it [30].

Gabon (an upper-middle income country, GDP $19.34 billion, 2013) straddles the Equator. About 80% of its 267,667 km² area is covered by dense tropical rainforest. The population of Gabon is estimated to be around 1.6 million inhabitants (6.3 inhabitants/km²), 86.2% of whom live in urban areas. CERMEL is based in Lambaréné, the capital of the Moyen-Ogooué Province, a semi-urban town of about 30,000 inhabitants surrounded by villages. Gabon is administratively divided into nine provinces, with villages mainly located along roads and rivers. Gabon is a highly malaria-endemic country. The official first-line treatment for uncomplicated falciparum malaria is artesunate + amodiaquine (AS + AQ) [31] or artemether-lumefantrine (AL) and severe falciparum malaria is treated with intravenous quinine. Intramuscular use of artemether or intravascular artesunate is not common in Gabon.

The primary outcome was the proportion (percentage) of poor-quality anti-malarial drugs in pharmacies in Gabon. Secondary outcomes were the proportion of outlets selling poor-quality anti-malarial drugs and availability of anti-malarial drugs that are no longer recommended as first- or second-line treatment in Gabon or by WHO. The following secondary outcome was added during the study to assess the external validity of the field survey: to determine where people purchased their anti-malarial drugs.

The overarching term ‘poor-quality drugs’ is used to describe the different categories: falsified medicines are fake medicines that are designed to mimic real medicines; counterfeit medicines are medicines that do not comply with intellectual-property rights or that infringe trademark law.

The field survey was conducted in January 2014 in Gabon. The six (out of nine) most populated provinces (ISO 3166-2:GA) were selected: Estuaire; Haut-Ogooué, Moyen-Ogooué, Ngounié, Ogooué-Maritime, and Woleu-Ntem. Selected locations were: Libreville (capital), Franceville, Lambaréné, Mouila, Port-Gentil, Oyem, Bitam, Owendo, Fougamou, Makouke, Bifoun, Gamba, and Lopé (Figure 1).

There are approximately 75 open and fully functioning pharmacies in Gabon (2013) [32]. Pharmacies were randomly selected. The randomization procedure was performed by BJV using statistical software (nQuery Advisor^® Version 7.0. Statistical Solutions, Cork, Ireland) on the day before the actual sampling. A full list provided by the Health Authorities of Gabon and a list from the National Health Assurance Company (La Caisse Nationale d’Assurance Maladie et de Garantie Sociale du Gabon [33], CNAGMS) with registered/licensed pharmacies and dispensaries was prepared (before sampling) to allow for proper randomization procedures. This list was accomplished with (unlisted) pharmacies by local nurses and fieldworkers. In total, six pharmacies were found which were missing on the CNAGMS list of pharmacies in Gabon. In Lambaréné (Moyen-Ogooué Province), where the CERMEL is based, all known pharmacies (n = 7) were sampled during the first week of the sampling period and thus not randomly selected. In all other areas, approximately 50% of the pharmacies were randomly selected in all neighbourhoods. Hospitals (except for the Albert Schweitzer Hospital), markets, grocery shops, and street peddlers were not visited because regulations for drug selling are in place; previsits by local nurses yielded little evidence (if any) of anti-malarial drugs sold there. The appropriate sample size and strategy is challenging, since data of the prevalence of poor-quality anti-malarial drugs are very scarce for Gabon. Thus, the most conservative sample size is given by using an unknown prevalence [hypothesized 50% frequency of outcome factor in the population (p)]. To determine the actual prevalence of poor-quality drugs available in Gabon with a precision of 5% with 95% confidence intervals (z = 1.96), a random sample of 384 was needed. The following equation was used: sample size \( n = \, \left[ {{\text{DEFF}}*{\text{Np}}\left( {1 - {\text{p}}} \right)} \right]/ \, [({\text{d}}^{2} /{\text{Z}}^{2}_{1 - \alpha /2} )*\left( {{\text{N}} - 1} \right) + {\text{p}}*\left( {1 - {\text{p}}} \right)].\) The sample size was calculated with OpenEpi (Open Source Epidemiologic Statistics for Public Health) version 3.03 [34].

A Gabonese ‘mystery shopper’ (JM, nurse, of Gabonese nationality) conducted the actual sampling process in Moyen-Ogooué (177/432 samples) and was trained utilizing standard sampling guidelines. She dressed according to regular Gabonese standards and gave no indication that she was not a regular shopper. A standard scenario was used: she asked which anti-malarial drugs were for sale. Subsequently, she purchased one full child/adult treatment in their original packaging of each of the available anti-malarials and of each available brand, but not of each available batch. Samples included drugs sold in the manufacturer’s original packaging as well as those distributed loose, often in plastic bags. Surprisingly, sellers never asked questions. Although the US Pharmacopeia (USP) recommends 30 dosage units [35] for a single tablet of the same lot number from each location, this was not deemed practically feasible in the Gabonese setting, and also too expensive. For logistical reasons the other 255/432 (60%) samples were collected by the investigators. Anti-malarial drugs purchased included: AL, AS + AQ, AS + sulfadoxine, AS-mefloquine, dihydroartemisinin-piperaquine, dihydroartemisinin-piperaquine-trimethoprim, artemisinin-piperaquine, artemisinin-naphtoquine, quinine, sulfadoxine-pyrimethamine (SP), mefloquine, proguanil, atovaquone-proguanil, proguanil-chloroquine, pyrimethamine and chloroquine. Other anti-malarial drugs were not purchased. Only solid dosage forms were collected (no liquid formulations). To avoid potential bias in subsequent sampling rounds, the exact reason for sampling medicines was not shared with the seller. Results were not reported back to the seller. For every sample collected, the collector completed and signed the sample collection form (including GPS locations) (Additional file 4) as soon as possible after leaving the point of sale and before performing the next purchase. Once purchased, all drugs were stored until testing at room temperature (in an air-conditioned room) with no sunlight. Humidity could not be controlled. Tests were completed at the laboratory of the Academic Medical Centre (AMC, The Netherlands) within 3 months of sample collection.

To determine where people purchased their anti-malarial drugs, a questionnaire was presented to food market dwellers in Lambaréné as well as at PK (‘point kilomètre’) 8 Le Marché Bananes (a transportation hub) and Marché du Mont-Bouët in Libreville. The survey was conducted after the purchase of the anti-malarial drugs. The most important question in this survey was: “where do you buy anti-malarial drugs?”.

Before shipment by air, samples were stored at CERMEL under appropriate storage conditions. The samples arrived within 36 h at AMC in Amsterdam, The Netherlands. Samples were protected by appropriate packaging (primary container and additional packaging) during shipment by air.

Samples were analysed at the Internal Medicine Research Laboratory of the AMC between February 2014 and April 2014 by BJV and JMG. The chemical analysis was performed unblinded to packaging. The Global Pharma Health Fund (GPHF) Minilab^® (Merck Darmstadt, Germany) was used to run semi-quantitative thin-layer chromatography (TLC) and disintegration tests on each sample to determine the presence and relative concentration of APIs [36]. Expired drugs were also tested. The MiniLab^® protocols award products a ‘pass’ for TLC if 80% or more of the labelled active ingredient(s) is present. For fixed-dose combinations (e.g., AL) and SP, ‘pass’ was awarded only if both active ingredients met this standard. TLC is an accepted method to assess the quality of drugs [37, 38]. The MiniLab protocols have been reviewed by the Promoting the Quality of Medicines (PQM) programme operated by the USP Convention. Each sample/test was run twice on separate days (once by BJV, once by JMG), with the assumption that the result most consistent with the reference was recorded. Thus, every API in every sample was tested twice. Standard operating procedures (SOP) provided with the Minilab were used [36]. Quality control of the GPHF MiniLab^® was performed daily before the drug testing and consisted of performing TLC on Minilab-reference samples for the anti-malarial drug analysed. In addition, Minilab reagents were quality control tested, using reference samples when a new lot was introduced. Samples were also tested to see if they disintegrated in purified water, following the guidelines of the European Pharmacopeia (EP) [39]. For this, apparatus A as described in the EP 2.9.1. was used at the laboratory of the Pharmaceutical Technology and Biopharmacy Department of Utrecht University (Utrecht, The Netherlands). Since the disintegration test requires six tablets per test, not all samples could be analysed. To include as many samples as possible, it was decided to test the samples per batch number instead of per sample. All samples with the same batch (or LOT) number, were expected to be homogeneous [40]. Samples failing TLC were analysed using the high-performance liquid chromatography with ultraviolet photo-diode array detection (HPLC–UV-PDA [41] February 2015 in a reference laboratory at the London School of Hygiene and Tropical Medicine) to quantify the amount of APIs present in each sample. This was compared with the stated dose on the packaging and the spectra achieved using a quality assured sample. For artemisinin derivatives, the artemisinin derivative screening test (ADST) was conducted according to an earlier published method [38]. Not all samples were analysed by HPLC–UV-PDA (the gold standard) due to lack of funding.

The packaging analysis was developed in line with previously published research [5, 27]. For the packaging analysis, genuine samples were requested by email from the manufacturers using a standard letter; with two reminders sent 2 and 4 weeks after the first email. Unfortunately no genuine anti-malarial samples were received.

Descriptive statistics were performed using SPSS 20.0 statistical package (SPSS Inc., Chicago, MI, USA). The confidence interval of the prevalence estimate was calculated using the Wilson procedure with a correction for continuity. Fisher’s exact test was used to calculate the difference between the number of poor-quality drugs of the Gabonese mystery shopper versus two European researchers. Inter-observer reliability (for chemical analysis) was calculated using the Kappa (κ) statistic [42].

The results of this field survey were shared with the Medicine Regulatory Agency (MRA) in Gabon, the Director of Health of the Province Moyen-Ogooué and Rapid Alert of the WHO.