278
Letters to the Editors
Br. J. clin. Pharmac.
(1990), 29
References Bennett, P. N. and The WHO Working Group (1988). In Drugs and human lactation, pp. 72-75. Amsterdam: Elsevier Science Publishers. O'Hare, M. F., Murnaghan, G. A., Russell, C. J., Leahay, W. J., Varma, M. P. S. and McDevitt, D. G. (1980). Sotalol as a hypotensive agent in pregnancy. Br. J. Obs. Gynaecol., 87, 814-820. Singh, B. N., Deedwania, P., Nademanee, K., Ward,
A. & Sorkin, E. M. (1987). Sotalol, a review of its pharmacodynamic and pharmacokinetic properties, and therapeutic use. Drugs, 34, 311349. Woods, P. B. & Robinson, M. L. (1981). An investigation of the comparative liposolubilities of ,Badrenoceptor blocking agents. J. Pharm. Pharmac., 33, 172-173.
The value of therapeutic drug monitoring Much of the medical literature is unsound because it relies on references themselves dubious or taken out of context, because it is based on the implicit but false assumption that, if no statistical connection between variables is uncovered, the variables are therefore not connected, and because the material presented, even when correct, is less than useful from a practical point of view. Two recent articles on therapeutic drug monitoring are thus flawed (McInnes, 1989; Spector, 1988). The former also contains the statement that therapeutic ranges are suspect (inter alia) because 'it is likely that the initial samples [for analysis, presumably for determination of the ranges] came from a biased population of patients identified on clinical grounds as presenting a particular therapeutic problem and therefore containing a disproportionate number of values outside the target range'. Surely, if there is any validity in the concept of therapeutic ranges, this is exactly what one would expect to see! Yes, of course therapeutic ranges are based on inadequate firm data, of course there is much that is anecdotal, of course the ranges do not apply to all patients under all conditions, of course the results of analysis are often misunderstood and misused. This author, at least, has never pretended otherwise to doctors at Groote Schuur Hospital (GSH) or to his students. Our department issues full reports, looking for drug interactions and giving warnings of conditions (low potassium levels in digoxin therapy, for instance) which invalidate the quoted range, and so on. The more bizarre cases are examined in the wards by our staff. Therapeutic drug monitoring (TDM) implies interpretation as well as assay, but that is not to say that it is useless. We should keep our feet on the ground. Clinicians rarely complain about the validity of
therapeutic ranges because they know from their training about therapeutic indices and feel intuitively that the higher the dose that can be given with least risk to the patient, the better the chance of control. Drug assay provides a benchmark, a step on the way to therapeutic success. A result gives the doctor an estimate: a drug is apparently not working because it is inefficacious or because the level is low or because the patient is poorly compliant. To be used thus, therapeutic ranges do not need to be universal or immutable, though improvements in quality are appreciated. Financially, for every assay that saves one patient-day in hospital, it is cost-effective to do up to 14 others less obviously valuable. Critics steer clear of discussing the use of TDM to assess compliance, as well they may. In Cape Town, at least, this is certainly the main economic and practical justification for pharmacological laboratories. As yet, our studies are incomplete but we conclude tentatively that the introduction of TDM to two local day hospitals improved patient compliance towards antiepileptic medication from about 30% to about 60%. In passing, for both hospitals, the doctor! examining the patients did not believe our results, did not believe that the rate of compliance was so poor, in the initial stages of the investigations. In general, it is our experience that doctors prefer to believe their patients' words more than their symptoms. The suggestion that clinical observation is all that is needed for adequate dosage adjustment is not half true. Our observations are that overdosage is often confused with underdosage, for digoxin and phenobarbitone at least. With theophylline, dosages used (assessed against blood levels) are typically very low even though the indication given for assay is lack of response to treatment. For certain patients, we have used
Br. J. clin. Pharmac. (1990), 29 the therapeutic range for phenytoin'as a lever to persuade doctors to give doses of 1 g or more daily to control ongoing status epilepticus. A problem usually ignored, too, is the placebo effect of being in the (relatively) low stress environment of hospital; an asthmatic or epileptic patient may well appear to be controlled by 'subtherapeutic' drug levels in bed but it does not follow at all that he will remain so outside, cold, frightened, hungry and jobless; adjustirng up the dose to achieve entry to the therapeutic range not only seems intuitively correct but appears to work. Anecdotal? Many procedures used in hospitals are strictly unproven but doctors persist in them because they appear useful. Because they work, there is no incentive to research them further. Frequently the research needed is nearly im-
Letters to the Editors
279
possible to do properly; establishing firm therapeutic ranges in local epileptic outpatients falls into this class, though we intend to continue our studies in this and related areas. All this does not mean that the procedures are valueless, as anyone working with expert systems will know. When the procedure accords with scientific theory and with common sense, it can be argued that the burden of proof lies upon those who criticise TDM, not its proponents. J. S. CRIDLAND Department of Pharmacology, Medical School, Observatory, 7925 South Africa
Received 2 October 1989, accepted 9 October 1989
References McInnes, G. T. (1989). The value of therapeutic drug monitoring to the practising physician-an hypothesis in need of testing. Br. J. clin. Pharmac., 27. 281-284.
Spector, R., Park, G. D., Johnston, G. F. & Vessel, E. S. (1988). Therapeutic drug monitoring. Clin. Pharmac. Ther., 43, 345-353.
Letters to the Editors
Br. J. clin. Pharmac.
(1990), 29
References Bennett, P. N. and The WHO Working Group (1988). In Drugs and human lactation, pp. 72-75. Amsterdam: Elsevier Science Publishers. O'Hare, M. F., Murnaghan, G. A., Russell, C. J., Leahay, W. J., Varma, M. P. S. and McDevitt, D. G. (1980). Sotalol as a hypotensive agent in pregnancy. Br. J. Obs. Gynaecol., 87, 814-820. Singh, B. N., Deedwania, P., Nademanee, K., Ward,
A. & Sorkin, E. M. (1987). Sotalol, a review of its pharmacodynamic and pharmacokinetic properties, and therapeutic use. Drugs, 34, 311349. Woods, P. B. & Robinson, M. L. (1981). An investigation of the comparative liposolubilities of ,Badrenoceptor blocking agents. J. Pharm. Pharmac., 33, 172-173.
The value of therapeutic drug monitoring Much of the medical literature is unsound because it relies on references themselves dubious or taken out of context, because it is based on the implicit but false assumption that, if no statistical connection between variables is uncovered, the variables are therefore not connected, and because the material presented, even when correct, is less than useful from a practical point of view. Two recent articles on therapeutic drug monitoring are thus flawed (McInnes, 1989; Spector, 1988). The former also contains the statement that therapeutic ranges are suspect (inter alia) because 'it is likely that the initial samples [for analysis, presumably for determination of the ranges] came from a biased population of patients identified on clinical grounds as presenting a particular therapeutic problem and therefore containing a disproportionate number of values outside the target range'. Surely, if there is any validity in the concept of therapeutic ranges, this is exactly what one would expect to see! Yes, of course therapeutic ranges are based on inadequate firm data, of course there is much that is anecdotal, of course the ranges do not apply to all patients under all conditions, of course the results of analysis are often misunderstood and misused. This author, at least, has never pretended otherwise to doctors at Groote Schuur Hospital (GSH) or to his students. Our department issues full reports, looking for drug interactions and giving warnings of conditions (low potassium levels in digoxin therapy, for instance) which invalidate the quoted range, and so on. The more bizarre cases are examined in the wards by our staff. Therapeutic drug monitoring (TDM) implies interpretation as well as assay, but that is not to say that it is useless. We should keep our feet on the ground. Clinicians rarely complain about the validity of
therapeutic ranges because they know from their training about therapeutic indices and feel intuitively that the higher the dose that can be given with least risk to the patient, the better the chance of control. Drug assay provides a benchmark, a step on the way to therapeutic success. A result gives the doctor an estimate: a drug is apparently not working because it is inefficacious or because the level is low or because the patient is poorly compliant. To be used thus, therapeutic ranges do not need to be universal or immutable, though improvements in quality are appreciated. Financially, for every assay that saves one patient-day in hospital, it is cost-effective to do up to 14 others less obviously valuable. Critics steer clear of discussing the use of TDM to assess compliance, as well they may. In Cape Town, at least, this is certainly the main economic and practical justification for pharmacological laboratories. As yet, our studies are incomplete but we conclude tentatively that the introduction of TDM to two local day hospitals improved patient compliance towards antiepileptic medication from about 30% to about 60%. In passing, for both hospitals, the doctor! examining the patients did not believe our results, did not believe that the rate of compliance was so poor, in the initial stages of the investigations. In general, it is our experience that doctors prefer to believe their patients' words more than their symptoms. The suggestion that clinical observation is all that is needed for adequate dosage adjustment is not half true. Our observations are that overdosage is often confused with underdosage, for digoxin and phenobarbitone at least. With theophylline, dosages used (assessed against blood levels) are typically very low even though the indication given for assay is lack of response to treatment. For certain patients, we have used
Br. J. clin. Pharmac. (1990), 29 the therapeutic range for phenytoin'as a lever to persuade doctors to give doses of 1 g or more daily to control ongoing status epilepticus. A problem usually ignored, too, is the placebo effect of being in the (relatively) low stress environment of hospital; an asthmatic or epileptic patient may well appear to be controlled by 'subtherapeutic' drug levels in bed but it does not follow at all that he will remain so outside, cold, frightened, hungry and jobless; adjustirng up the dose to achieve entry to the therapeutic range not only seems intuitively correct but appears to work. Anecdotal? Many procedures used in hospitals are strictly unproven but doctors persist in them because they appear useful. Because they work, there is no incentive to research them further. Frequently the research needed is nearly im-
Letters to the Editors
279
possible to do properly; establishing firm therapeutic ranges in local epileptic outpatients falls into this class, though we intend to continue our studies in this and related areas. All this does not mean that the procedures are valueless, as anyone working with expert systems will know. When the procedure accords with scientific theory and with common sense, it can be argued that the burden of proof lies upon those who criticise TDM, not its proponents. J. S. CRIDLAND Department of Pharmacology, Medical School, Observatory, 7925 South Africa
Received 2 October 1989, accepted 9 October 1989
References McInnes, G. T. (1989). The value of therapeutic drug monitoring to the practising physician-an hypothesis in need of testing. Br. J. clin. Pharmac., 27. 281-284.
Spector, R., Park, G. D., Johnston, G. F. & Vessel, E. S. (1988). Therapeutic drug monitoring. Clin. Pharmac. Ther., 43, 345-353.
