Morphine in the MS Contin® slow-release tablet is embedded in a complex dual matrix of hydroxyethyl cellulose and cetostearyl alcohol, designed to release the drug over 12 h . Crushing the tablet disrupts this matrix, allowing the rapid extraction of morphine. The amount of morphine in prolonged-release tablets is permitted to vary by 5%  so a 60 mg tablet could contain from 57 to 63 mg morphine sulfate. The extraction of morphine by cold water (56 mg) and hot water (59 mg) was therefore essentially complete. None of the filters bound morphine, and their retention of morphine was due to the volume of liquid which remained. Consequently, rinsing the filters increased the recovery of morphine. Repeated rinses brought diminishing recoveries of morphine, and increased the volume to be injected, and therefore the number of rinses used in the combined filtration method was a minimized but nevertheless gave a good recovery (84-93%) of the extracted morphine. Overall, the extraction of morphine and its recovery after filtration was similar after cold and hot extraction.
The MS Contin® tablet contains a number of constituents with low or no water solubility which are liable to produce particles in the extract . These include cetostearyl alcohol, which is a mixture of two waxes: cetyl alcohol (1-hexadecanol, mp 49°C) and stearyl alcohol (1-octadecanol, mp 61°C); magnesium stearate (mp 88°C); talc, a hydrated magnesium silicate; and hydroxyethylcellulose (a gelling agent which is insoluble in water). The coating contains other insolubles such as iron oxide, but this was usually removed in preparing the extracts.
There are advantages and disadvantages in counting particles by microscopy rather than an instrumental method, such as the Coulter Multisizer which has been used to study the effectiveness of filters for heroin injections . In this latter study, the instrument required considerable dilution of the sample (50 μl to 75 ml) with the possibility of dissolution of some particles which would have been present in the smaller volume to be injected. The dilution was made with an electrolyte (saline) which may also have affected particle solubility or aggregation. Microscopy avoided dilution and enabled examination of the appearance of particles, which gave insights into their origin (such as crushed solids, condensed wax droplets, and crystallised morphine) which in turn can indicate how they could be removed. However, microscopy necessarily examines only a small part of the total sample, adding to errors as discussed below.
Counting particles, especially in the unfiltered preparations, was inherently variable due to the large amount of insoluble material and its complex physical form. This variability also affected instrumental counting, and Scott  considered that variability in the particle counts made the exact values meaningless although useful for comparison of filters. In our study counts are presented as the number of particles in an injection volume in order to relate the data to health impacts. This required a large multiplier factor. For example a count of one 100 μm particle in 5 fields using the 5× objective would give an estimated 8,897 particles in the 3 ml dose volume, and one 10 μm particle (20× objective) would give 161,333 particles in 3 ml. If there were no similar particles in the other two replicate mixtures, then the mean particle counts would be 2966 for the 100 μm particle and 53778 for the 10 μm particle. The particle counts are therefore reported in thousands to avoid implying a level of precision which would be misleading. The counts are indicative estimates rather than precise determinations, but are nevertheless able to show that filtering can greatly reduce the number of particles injected.
The working area for preparing the injections was neither sterile nor particle-free, since the aim was to reproduce the typical conditions used for illicit preparations by injecting drug users. Consequently, a significant number of particles and fibres were found when control injections were prepared, showing that particles are ubiquitous unless removed by specific cleaning procedures. Fibres, however, were not counted as particles since they were present on control slides and were not considered to be tablet-derived. Environmental particles will vary widely according to local conditions but will add to the total particle burden in the injection. Not using a clean workplace became a limitation in counting particles in the cigarette plus 0.22 μm filtrate, in which virtually all tablet-derived particles had been removed.
The form of the waxes was evidently altered after melting and re-solidifying, with the formation of wax droplets of various sizes. Hot extraction resulted in a shift in particle size distribution, with the formation of more small particles (< 5 μm) and fewer larger particles. However, the remaining particle burden in unfiltered preparations was still too large for this to be considered other than harmful to inject. Pharmaceutical standards require that, measured by microscopy, injections of less than 100 ml must have, in total, no more than 3000 particles > 10 μm and no more than 300 particles > 25 μm . The hot unfiltered morphine tablet preparations had, in the total volume of 3 ml, an average 1.1 million particles > 10 μm and 368,000 particles > 20 μm. For the cold preparations, the numbers of particles were 7.2 million > 10 μm, and 4.0 million > 20 μm.
The aqueous solubility of morphine is critically dependent on its ionization and therefore the pH, as the ionized form is freely soluble and the free base has a low water solubility (0.25 mg/ml at 35°C) . These authors found that, at 35°C, the solubility of morphine in water was 13.39 mg/ml at pH 6.35 and 5.75 mg/ml at pH 6.69. This change in solubility with pH could be explained by the change in ionization and the low free base solubility. A 60 mg tablet of morphine sulfate (MW = 758.9) contains 45.1 mg morphine (MW = 285.3), or 15.0 mg/ml in the 3 ml extract. Using the amount of morphine sulfate recovered in cold extracts, this concentration would be 14.0 mg/ml. In either case, the concentration of morphine in the 3 ml extract will be critically close to, or exceed, its solubility, especially as the pH is slightly higher (6.4) and the temperature was considerably lower (about 20°C). From the pKa of morphine (8.08) and the buffer equation, pH = pK
+ log10([base form]/[acid form]), it can be calculated that morphine is 1.8% unionized at pH 6.35 and 2.1% unionized at pH 6.40.
It was considered that morphine crystal formation was an artifact due to alkalinity in the glass microscope slide or cover, since they did not form on acid-washed glass or plastic slides. However, there is a significant risk of formation of morphine crystals in the tablet extracts,, and conditions of preparation could reduce this problem by decreasing pH or, preferably, increasing the volume of water. Although morphine will eventually dissolve in blood this will take some time, and any crystals which remain undissolved during the brief transit time from injection site to capillary bed are liable to cause embolisms.
The unfiltered tablet extracts must be considered extremely harmful as they contained many particles of all sizes. After intravenous injection, particles will flow through ever-widening vessels back to the heart and then they will enter the pulmonary circulation, where the smaller arteries which are 300-400 μm diameter  could be occluded by the largest particles found in unfiltered mixtures. Arterioles (9-40 μm diameter) and capillaries (7-9 μm diameter) could be blocked by the smaller particles. Even particles too small to embolize may cause vascular injury. Small airborne particles (< 2.5 μm) have been implicated in cardiac and vascular damage, including endothelial dysfunction and promotion of atherosclerotic lesions . Large numbers of particles of this size were present in the unfiltered mixtures.
The cigarette filter reduced the number of particles, especially the larger particles. This filter was more effective when used after hot extraction, but the remaining particle burden remained too high for injection. Of course, cigarette filters are not designed for liquids. The morphine recovery from the cigarette filters was nearly complete (90%) after two rinses. The unfiltered mixtures caused a block of the syringe filters, but the cigarette filtrate passed through them, as did the rinse volumes. Scott  found that both 0.22 μm and 0.45 μm syringe filters blocked with heroin injections, and abandoned them in favour of 5 μm filters. However, these blockages can be prevented by the use of a preliminary, coarse filter, such as the cigarette filter applied here.
The combination of cigarette filter then syringe filter mostly gave a good recovery of the extracted morphine. The 0.22 μm filter is considered to be sterilizing because, unlike the 0.45 μm filter, it will remove bacteria. In a trial with injecting drug users , it was found that 0.22 μm syringe filters were effective in removing bacteria from 3 out of 4 injections, while larger pore filters (15 - 20 μm) were completely inadequate.