A stock solution of calibration standards (CS) of Amoxicillin (AM) 5.0 mg∙mL⁻¹ was prepared in phosphate buffer at 50.0 mM potassium dihydrogen phosphate adjusted to pH of 5.0 ± 0.1 with a 45% (w/w) solution of potassium hydroxide. Several dilutions were performed in the same buffer in order to obtain running solutions at 3 different concentration levels to establish and evaluate the response function. In the same way, the stock solution of calibration standards with a concentration 500 µg∙mL⁻¹ of amoxicillin and 62.5 µg∙mL⁻¹ of potassium clavulanate (CL) were prepared in phosphate buffer at 50.0 mM sodium hydrogen phosphate and adjusted to pH of 4.4 ± 0.1 with solution of potassium hydroxide at 45 % (w/w).

For the validation standards (VS), the solutions were prepared as for calibration standard including the reconstituted matrix powder for amoxicillin suspension and for mix of amoxicillin and potassium clavulanate. The stock solutions were obtained as in the case of calibration standards to which was added a corresponding amount of the matrix. Dilutions were performed in phosphate buffer in the same way as described for the CS in order to obtain five solutions at different concentration levels as shown in Table 1.

The solutions were freshly prepared and protected from light. After centrifugation at 5000 rpm for 5 minutes, they were filtered through 0.45 µm PTFE syringe filter discs before being placed into autosampler set in the LC analysis systems. Finally, 3 set of solutions were run and used for calibration standard and of 5 sets for validation standard considering that the 100% concentration level is the nominal level of the amoxicillin and of potassium clavulanate corresponding to the claimed values of samples.

After their validation, the methods have been implemented in a quality system environment to the use in routine analysis. This environment is under a European Union Good Manufacturing Practices (UE-GMP) licence authorisation (N˚1813H) of the Pharmacy Department that is also WHO prequalified laboratory. The whole analyses were carried out according to an analytical procedure as shown in Figure 2.

The main tests were the pH measurements and the assay of active substances after their identification by two orthogonal tests namely comparison of

Legend: (+): presence of the substance mentioned on the left; (-): absence of the substance mentioned on the left. (LM): local manufacturer; (EU): imported from Europe; (AS): imported from Asia; (AM): Amoxicillin; (CL): Potassium Clavulanate.

the peak retention time with that of the reference substance and superposition of the respective UV-Vis spectra. The assay specifications were those of USP 37 that are 90.0% - 120.0% for amoxicillin; 90.0% - 125.0% for potassium clavulanate and 5.0 - 7.5 for pH of the reconstituted suspension [12] . The non-conforming samples were reanalyzed by another analyst before concluding on the result.

The sampling strategy for routine analysis was inspired from P. Newton proposal [26] .

We would prefer large size but due to financial limitation, however we collected 278 samples. Samples were bought by anonymous shoppers as patients in random sampling method. Each collector behaved as in real shopping situation called technique of “blind procedure”. This sample size seems enough for a pilot and preliminary study to assess the quality of pediatric formulations. Purchasers who were unaware of the study were recruited from the Kinshasa School of Public Health, from the Faculties of Pharmacy of Lubumbashi and of Kinshasa and the Medical Institutes of Kolwezi and of Matadi. Purchases were done in Circuit Urban Controlled (CUC) and Peri Urban Circuit Deficiency Controlled (PCDC) between November 2014 and November 2015. We were able to collect a total of 278 samples, 120 from CUC and 158 from PCDC. Then a first visual inspection based on the identification of the product, its validity (expired date) and the integrity of its packaging was carried out in Kinshasa by pharmacist’s collaborators. A first encoding of the samples was done to identify samples as per their

Legend: p = number of series; n = number of repetitions; LOD: limit of detection; LLOQ: Lower Limit of quantitation; ULOQ: Upper Limit of quantitation.

origin (manufacturer), place of purchase, expiry and manufacturing date, lot number, international non-proprietary name, brand name and the reported or claimed unit dose. Only 92 from CUC and 108 from PCDC were able to be analyzed in Belgium, the remaining were withdrawn due to not fulfilling our requirements above. Indeed, there were samples with not satisfying appropriate expired date (exceeded or less than one year expired date), the powder aspect that was clogged, the primary and secondary packaging that was not readable, some were even absent. In addition, insufficiency of unit number per sample was applied to discard not valid samples. To prevent retest analysis, the eligible samples should have at least two units. Than samples were stored in controlled environmental conditions (see section 2.5.1.). From the total eligible samples (200 samples), 48.0% samples were manufactured in the DRC, 30.5% in France and 21.5% in India.

Depending on the sampling place and thus their storage conditions by located population, the samples were categorized into two circuits: The Circuit Urban Controlled (CUC) and the Peri Urban Circuit Deficiency Controlled (PCDC). The controlled circuit is the one which is under the control of the Pharmacy Direction of the Ministry of Health (Direction de la Pharmacie et du Médicament or DPM) that is the National Medicines Regulatory Authorities (NMRA) in the DRC. Apart the verification of authorization administrative documents, their duties are the control of the conservation conditions of medicines since DRC belongs to the climatic zone IVa (max 30˚C, max 65% RH) [27] . The CUC samples were purchased from authorized wholesalers. These provide medical products at all levels in the public, private and denominational sectors in urban, peri-urban and rural areas. In addition, some of them are exclusive importers and distributors of products for the entire country. Normally, in this circuit the risk of finding falsified and degraded medicines is low. The PCDC is a circuit where control is almost not existing with high risk of encountering falsified and degraded products [28] .

The samples were powder for pediatric suspension of Amoxicillin (AM) or associate with clavulanate of potassium CL). The selection criteria mostly applied was the importance of drug in terms of distribution volume and therapeutic aspect; the active pharmaceutical ingredients have to be widely used in the DRC in the most recurrent infections in children according to the statistics of the National Health Development Plan 2010 [29] . Moreover, infant mortality in the DRC remains of concern [30] [31] . AM and CL fulfilled to these criteria. Other inclusion criteria mainly on technical aspects were valid expiring date, not clogged powder and intact primary packaging.

When received in Belgium, the samples were checked again to confirm they initial conditions and if they are still correct for analysis. An identification code and a bar code were assigned to the eligible samples. This code allowed administrative and physical traceability of the sample from receipt to the results issuing. Administrative traceability was supported by the following documents: request for analysis, receipt, raw results of the analysis, certificate of analysis, copy of the laboratory booklet and all associated records or documents. Identification and assay of active ingredients were done pH measurements as well.

For the analysis of amoxicillin samples, we used method 1 in which the concentrations of standards solutions were 100 µg∙mL⁻¹ diluted from a stock solution of amoxicillin at 1.0 mg∙mL⁻¹ (100 mg of amoxicillin up to 100 mL).

For samples containing both amoxicillin and potassium clavulanate, we used method 2 in which standards solutions contained also 100 µg∙mL⁻¹ of amoxicillin but with 12.5 µg∙mL⁻¹ of potassium clavulanate that were prepared from 100 mg of amoxicillin and 12.5 mg of clavulanate up to 100 mL.

In the two situations, two independent standard solutions were prepared. Since in solution, amoxicillin and potassium clavulanate are not stable especially at room temperature, we took care to perform the run of the set within less than 6 hours. In addition, the set was accepted if the recovery of the second standard compared to the first is between 98.0% - 102.0%.

Samples which are powder to be reconstituted with water were freshly prepared following the manufacturer’s instructions on the package leaflet.

The overall results indicated that 28% (CI; 21.3% - 33.7%) samples were declared out of specification (OOS). This finding is very close to the overall prevalence (30%) of low-quality drugs in sub-Saharan countries as estimated by WHO [12] . In addition, a recent review of 44 prevalence studies in 25 different countries, including 11 low-income countries, shows that the median prevalence of substandard/counterfeit drugs was 28.5% (range 11% - 48%) [32] . On the other hand, this gross result seems to be lower than the estimates of the review of the literature on the global prevalence (more than 50%) of antimicrobial directed in 2015 by T. Kelesidis [14] .

From the detailed results we noticed 7 samples (or 3.5%, CI: 0.9% - 6.1%) with pH values not complying, of which 3 cases were the sole cause of the non-specification of the sample. Also 52 samples were OOS because of AM and/or CL content, 4 cases of overdosage, 2 cases of absence of potassium clavulanate, one case with a pH greater than 7.5.

Table 4 shows the OOS results divided into two circuits, the CUC that is recognized by DPM and the PCDC with uncertain or deficient control. There was non-significant difference in quality between locally manufactured medicines and those of European manufacturers in the CUC.

According to the result and considering the GMP of European manufacturers, the case of OOS noticed in certain European products may be due to the falsification of which these products are often victims since there is no traceability of original source. They might have a very probable hypothesis with the existence of mafia drug trafficking networks especially between Kinshasa and Brazzaville. The heads of African states, especially those of the two Congo, expressed their concern at the France Africa summit held in January 2017 in Bamako (Mali) [33] . In the light of these results, the frequencies of the poor-quality medicines in the periurban zone represented by the PCDC is much higher than those of the CUC. The prevalence of non-specification products is close to that observed by T. Kelesidis (see above). Of the 200 samples of powders for suspension based of amoxicillin a 46/55 (83.6%, CI: 73.9; 93.4) of the OOS samples were from PCDC. This indicates that the application of an effective quality control system and a

strong political commitment by the NMRA may reduce the prevalence of counterfeits [34] . With regard to the different origin the frequency of the drugs of bad is almost five times higher in the PCDC for the locally manufactured products (LM) and Indian origin (AS). This can be due to non-compliance with GMP rules during the manufacturing process or a counterfeit. Also, for the LM to the problem of packaging (non-hermetic primary packaging) and the conservation conditions are still not respected; Almost 10% of the samples were not analyzed because of the presence of visible traces of the powders on the primary packaging or because of the bulk of the powder. We can also observe that further away from Kinshasa (the central distribution point) the prevalence increases. It was in Kolwezi, the least accessible site of all, that there were two cases of CL. absence in two samples manufactured in India, whose official distributor based in Kinshasa did not recognize the batch. This reflects the lack of traceability of pharmaceutical products in DRC even among official distributors. (Figure 6 illustrates the revelation of the absence of clavulanate in a sample).

Apart from two cases due to non-compliant pH, the overall assay results were presented according to their circuits the sampling site and the manufacturers (see supplementary data). We could notice that the distribution of OOS was dominated mainly by under dosage content than higher. However, the results within specification were somewhat in high proportion. Figure 7 confirms that distribution in particular for amoxicillin content for samples from peri-urban areas (PCDCs) that tends towards low values (under dosage).

The distribution of the OOS samples according to their origins in the different

circuits reveals that the local manufacturers L2 and L3 showed much dispersed results in the two circuits (CUC and PCDC). A failure to comply with GMP rules would be a probable cause of poor quality. However, the trend of these results to the lower median values in the PCDC indicates a probable degradation of these products because in this circuit the storage conditions are not always favorable. Reversing the trend of the local manufacturer (L3) when moving from the CUC to the PCDC seems to show that the trend of the amoxicillin content to the low values does not distribute same way. If one accepts that the poor conservation conditions would be a significant factor of the low content of amoxicillin in the PCDC then it does not affect all products in the same way. For example the sealing of the bottle closure system could be one of the reasons. Indeed, the visual inspection carried out on these samples led to the discarding of 10 samples collected in the CPDC, 8 of which were from the local manufacturer L3 and 2 from the local manufacturer L2. These samples showed many lump, a sign of exposure to moisture. Interpretation for European manufacturer EU1, however it may suggest a counterfeit for the medicines of European manufacturers whose local representatives did not recognize the existence of certain batches (4 in total). It should be noted that information on the traceability of batches was not always easy to obtain either at the Ministry of Health or at the local manufacturers or the representatives of European manufacturers.