The following information on the results for each drug is presented in Annex 1: shelf-life, assay method and specification limits, number of samples collected (GMS samples, facility samples, longitudinal series), number of samples in each risk group (age > 50% of shelf-life, facility type - PHC, hot climate, slow transport method) and summary assay results (mean with 95% confidence intervals) with corresponding mean age. When a study drug is supplied by more than one manufacturer, data are summarized separately for each manufacturer. Expired drugs were excluded from reported results with the exception of epinephrine injection, as these are presented separately. Individual sample results arc illustrated in X-Y scatter diagrams showing the assay value against the age of sample. Key findings are summarized for each drug and a summary of all results appears in Tables 5 and 6.

GMS samples

Table 5 shows that in eight out of 12 drugs, all GMS assay results were within the pharmacopoeal limits. For four drugs, samples were found with the content of active ingredient below specifications: ferrous sulfate tablets, retinol tablets, ampicillin injection and ergometrine injection. In three of these (retinol tablets, ampicillin injection and ergometrine injection)/the mean assay value (or 95% confidence limits of the mean) was also below the lower limit of pharmacopoeal specification.

Facility samples

Table 5 also shows summary results of the facility samples. For the following four drugs, all assay results were within the pharmacopoeal limits: amoxicillin capsules, acetylsalicylic acid tablets, phenoxymethylpenicillin tablets and tetracycline capsules. The following nine drugs had some samples with the content of active ingredient below specification: ampicillin capsules, doxycycline capsules, ferrous sulfate tablets, retinol tablets, epinephrine injection, ampicillin injection, benzylpenicillin injection, ergometrine injection and penicillin procaine injection. Again, for three of these - retinol, ampicillin injection and ergometrine injection - the mean assay (or 95% confidence limit of the mean) was below the lower limit of the pharmacopoeal specification.

Longitudinal series

Table 6 shows the summary data for GMS/facility sample pairs for 12 drugs. Two drugs showed a statistically significant mean loss: ergometrine injection (-7.1%, p < 0.0005) and penicillin procaine injection (-3.6%, p < 0.001). The other drugs showed non-significant changes (loss or gain) at facility level. The gain was no more than 5% in all cases with the exception of retinol tablets which apparently gained 12.2%.

Table 5 - Summary results for GMS and facility samples for each study drug

Notes:

^a number of samples tested.

^b in months

^c % of stated content (95% confidence limits)

^d number (% of total) of samples with active ingredient below lower limit of pharmacopeia specification.

Table 6 - Summary results for longitudinal studies

Notes:

^a number of batch-paired sample pairs
^b in months
^c % of stated content (95% confidence limits)
^d the time interval between sampling at GMS and sampling at facility, in months
^e difference as a percentage of initial content (GMS assay - facility assay)/(GMS assay) × 100
^f students' t-test for paired samples, one tail

Risk factors for sub-standard quality

The number of samples in various "risk groups" is shown in Table 4, No significant association was found between the failure to meet pharmacopoeal specifications and any of these risk factors.

With ergometrine injection, there was a significant difference between the three manufacturers. Table 7 shows the results of a chi-square test (p < 0.00001). A comparison of longitudinal series data (sample age, assay values and potency loss) for individual manufacturers of ergometrine injection is presented in Table 8.

10 (15%) ergometrine injection field samples were found in storage at room temperature. The relative risk (1.33) for failure in samples at room temperature storage, compared with samples found under refrigeration, was marginally significant (1.01 < RR < 1.74).

Table 7 - Other factors influencing quality

Table 8 - Longitudinal series for individual manufacturers - ergometrine injection

Notes:

^a number of batch-matched sample pairs
^b in months
^c % of stated content (95% confidence limits)
^d the time interval between sampling at GMS and sampling at facility, in months
^e difference as percentage of initial content (GMS assay - facility assay)/(GMS assay) × 100
^f students' t-test for paired samples, one tail