SAMPLING PROCEDURES AND ERRORS
To collect a ‘representative sample’ forms a vital aspect of analytical chemistry, because the samples subjected to analysis are assumed to be perfectly homogeneous and truly representative. Thus, sampling may be considered as the most critical aspect of analysis. In other words, the accuracy and significance of measurements may be solely limited by the sampling process. Unless and until the sampling process is performed properly, it may give rise to a possible weak link in the interpretation of the analytical results. For instance, the improper handling of a blood sample both during and after sampling from a patient prior to an operation may not only pose serious complications but also may prove fatal.
A definite instruction with regard to the sampling of given materials have been duly put forward by a number of professional societies, namely :
· Association of Official Analytical Chemists (AOAC),
· American Society for Testing Materials (ASTM), and
· American Public Health Association (APHA).
However, a good deal of the wisdom of the analyst supported by the application of statisical results and wealth of experience may go a long way in achieving reasonably accurate and reproducible results.
Samples may be categorized broadly into four heads, namely :
(a) Gross Sample : A sample that represents the whole lot and may vary from a few grams or less to several pounds based on the nature of the bulk material.
(b) Sample : A sufficiently small size of the sample exclusively for the purpose of analysis and derived from the representative gross sample.
(c) Analysis Sample : An aliquot or portion of the ‘sample’ being subjected to actual analysis.
(d) Grab Sample : A single sample usually taken at random and assumed to be representative. It is considered to be the most unreliable way to sample a material.
Sampling of solid materials are comparatively more difficult than other materials because of the follow-ing three reasons, namely :
(a) Variation in particle size.
(b) Inhomogeniety of the material.
(c) Variation within the particle.
Sampling of solids can be best accomplished by adopting the following procedures :
· To take 1/50 to 1/100th of the total bulk for gross samples.
· To take larger gross samples for products having larger particle size.
· To sample large bodies of solid materials while they are in movement to obtain aliquots representing all portion of the bulk.
· To handle tissue samples, several tiny parts of an organ may be taken and combined together.
Sampling of liquids may be carried out by following these procedures :
• Small heterogenous liquid samples are first shaken thoroughly and then followed by immediate sampling.
• Large volumes of liquids are best sampled immediately after a transfer; or if in a pipeline, after passing through a pump where it has undergone the most vigorous mixing.
• Large volumes of stationary liquids are normally sampled with a ‘thief sampler’, i.e., a device for collecting aliquots at different levels.
• Samples are best drawn (with a ‘thief sampler’) at various depths diagonally instead of vertically down so as to have a better cross-section of the bulk liquid.
• Either separate aliquots of liquid may be analyzed individually and the results combined duly, or the various aliquots may be first combined into one gross sample and replicate analysis carried out. However, the latter method is preferred for obvious reasons since the analysis shall have a better hold on the accuracy and precision of the analysis.
• For sampling of biological fluids the ‘time factor’ is of utmost importance and hence, should be performed by qualified pathologists attached to clinical laboratories under adequate supervision. A few specific examples are stated below :
(a) A 24 hour urine sample collections are usually more reliable than single specimens.
(b) A sample for blood-sugar analysis is more reliable in a fasting patient.
(c) A sample of cerebro-spinal-fluid (CSF) from the vertebral column by lumber puncture in patients having suspected pyogenic meningitis.
A grab-type gas sample is usually satisfactory in certain cases. For example :
(a) A breathe sample may be collected by allowing the subject to blow into an evacuated bag. (Persons driving automobile under the ‘influence of alcohol’ on high-ways during festive seasons).
(b) Auto exhaust may also be collected in large evacuated plastic bag to monitor the pollution by vehicles run by gasoline/diesel/CNG in cities and metropolis.
The famous adage—‘to err is human to forgive divine’—literally means that it is natural for people to make mistakes. However, errors in analytical chemistry or more precisely in pharmaceutical drug analysis are normally of three types, namely :
(a) Determinate Errors
(b) Instrumental Errors
(c) Personal Errors
These above mentioned errors would be discussed briefly here with specific examples. It is pertinent to mention here that errors outside the range of ‘permissible errors’ in the analyses of pharmaceutical substances may cause serious problems because most of these substances are usually highly toxic, potent and used exten-sively in life-saving processes across the globe.
Errors caused due to either incorrect adoption of an assay method or an incorrect graduation read out by an analyst are termed as determinate errors. Such errors, in principle may be determined and corrected. In usual practice the determinate errors are subtle in nature and hence, not easily detected.
A few typical examples of determinate errors are stated below :
(a) Gravimetric Analysis : Where a compound is precipitated from a solution and the analyst believes that the analyte has been removed from the solution completely. Actually a small portion of the substance under investigation shall remain in solution. This sort of error is normally so insignificant that it is often neglected.
(b) Incomplete Chemical Reaction : Where a chemical reaction fails to attain the chemical equilibrium, thus virtually invalidating most calculations entirely based on chemical equilibrium characteristics. It may be eliminated by carrying out a detailed study of the reaction kinetics.
(c) Colour-change at Endpoint : Where a colour change is employed for an endpoint signal in a volumetric analysis. It may require an excess quantity of reagent to affect the colour change which ultimately shows completion of the chemical reaction between reagent and analyte. Hence, it is absolutely necessary to determine this excess amount of added reagent, otherewise the analytical results may give a positive error. Therefore, in all such analytical procedures a ‘blank titration’ is performed simultaneously to determine how much reagent is required to affect the colour change when no analyte is present.
The past three decades have witnessed a quantum progress and advancement in the field of analytical chemistry. Nowadays, both microprocessor based and computer-aided analytical instruments have more or less replaced the manually operated ones in any reasonably good analytical laboratory. One of the most prevalent determinate errors is caused by analytical intruments which are found to be ‘out of calibration’. Hence, it is very essential that such instruments need to be calibrated periodically, for instance, a pH meter is calibrated using a buffer solution of known pH, say adjusting the meter to read pH = 7.00 when a buffer of pH 7.00 is measured ; a single-pan electric balance is calibrated by using standard certified weight box; an UV-spectrophotometer is calibrated using standard solutions of known substances.
In a similar manner, the calibration of glassware, such as : volumetric flasks, pipettes, burettes, measur-ing cylindres are duly carried out by specific methods recommended by Indian Standards Institution (ISI), British Standards Institution (BSI), National Physical Laboratory (NPL), United States Pharmacopoeia (USP) at specified temperatures.
In addition to errors caused due to improper assay methods or faulty instruments, it may also be due to the analyst. A few typical examples are cited below :
(a) Physical Impairment : A person suffering from colour blindness may not be in a position to assess colour-changes precisely ; or if he uses bifocals he may not take the burette readings accurately.
(b) Learning-Curve Syndrome : An analyst must practise a new assay method employing ‘known’ samples before making an attempt to tackle an unknown sample, thereby minimising the scope of personal errors.