Is there a cure for drug errors?BMJ 1995; 311 doi: https://doi.org/10.1136/bmj.311.7003.463 (Published 19 August 1995) Cite this as: BMJ 1995;311:463
- R E Ferner
- Consultant physician West Midlands Centre for Adverse Drug Reaction Reporting, City Hospital, Birmingham B18 7QH
Better systems should be adopted to identify and eliminate errors
Drug errors are avoidable but difficult to avoid. The prescriber has to be knowledgeable enough to choose an effective treatment suitable for the individual patient, taking into account age, infirmity, and possible interactions with other drugs. Having selected the right agent and the correct dose, the prescriber has to transmit the message to the dispenser, who has then to hand the drug to the patient or to a carer or nurse, who has to see that the drug is given in the correct way and at the specified times. The process is complex, and, not surprisingly, errors occur. Their frequency is difficult to estimate. As with air travel, deaths are sufficiently rare that “near misses” are important. More important still is to devise a strategy for reducing errors.
Bates et al have recently reported a study from Boston in which they identified about 6.5 actual and 5.5 potential adverse drug events—errors or adverse reactions involving drug treatment—per 100 hospital patients.1 Over a quarter of the observed events resulted from errors, and these were generally more serious than the adverse reactions. Drug errors are clearly an important cause of morbidity in hospital2 3 4 and probably in general practice.5 They account for about a fifth of all deaths due to adverse drug events6 and are also an increasingly common stimulus to litigation.
The standard approach to drug errors “has focused on identifying and castigating” those responsible.1 Those who make errors can suffer severe penalties, and in England and New Zealand doctors have been convicted of manslaughter. Nurses who are reported to nurse managers for making errors are disciplined as a matter of routine.
The punitive approach blames individuals for errors that are inherent in the system, and it can achieve only secondary prevention. In the long term it might be possible to choose medical students who are not prone to making errors and to train doctors to avoid errors.7 8 In the meantime, however, the more doctors know about the drugs they prescribe the safer their prescribing is likely to be, so teaching of practical therapeutics should be strengthened, particularly before and during the years as a junior hospital doctor.
A distinctly different approach is to examine the systems that lead to error and to improve them.9 The companion paper to the Boston study attempted to do this.10 It analysed the errors both by type (for example, wrong dose or wrong drug) and by the stage in the process at which they occurred. The important conclusion was that most errors were due to lack of knowledge about drugs; mistakes in identifying drugs because of “look alike” packages or “sound alike” names; and lack of information about the patient's condition, laboratory results, or current treatment. The authors hope that prescribing by computer will improve the performance of the overall system by providing the information needed to prescribe safely while forcing prescribers to adhere to certain rules: “Do not prescribe ampicillin to a patient who is allergic to penicillin,” for example.
Computers can at present reduce errors due to bad handwriting and can make sure that drug and dose are consistent (“Do you really mean this patient to have 250 milligrams of digoxin?”). They do not, however, necessarily reduce errors of choice, since the press of a button will prescribe one of a series of drugs with similar names. Nor do they necessarily protect patients from receiving drugs to which they are allergic or with which there may be an interaction, since the information must be entered into the computer.11 Moreover, by hiding the intermediate stages in a calculation computers can make errors in setting up computations difficult to spot later.12
Errors in transmitting data from the prescriber to the dispenser and then to the patient or nurse could be reduced by sending the data by computer networks. Stock ordering by computer is well established and could in theory be extended to individual prescriptions for particular patients. Nevertheless, a moderately sized hospital, with many patients taking several drugs several times a day, might need 5000-10000 individual doses each day, so the problems of central unitdose dispensing are huge.
For many hospitals, however, drug errors exist today while computerised drug prescribing and dispensing are a long way off. What can be done in hospitals now? The first step is to make all those who are part of the system aware of its fallibility, so that they are encouraged to check at each stage. Junior doctors, who do most hospital prescribing, should carry a ready source of information on prescribing (the British National Formulary is the paradigm) and should have ready access to a drug information service. Their prescribing should be checked by the senior medical staff, by the pharmacist, and by the nurses who give the drugs. Patients, too, can take responsibility for their medicines, though only if the medical staff have explained to them what they are taking and why.
Certain simple rules—for example, asking that “micrograms” should always be spelt out and that drug names should always be printed in capital letters—may help to reduce the error rate. Simple changes, like introducing a distinctive ampoule for potassium chloride strong solution, can save lives.13 As for the errors that do occur and that could or do lead to harm, these need to be made known, examined, and acted on. A formal reporting system, in which all hospital staff report suspected or near miss errors confidentially to a named member of the medical, pharmaceutical, or audit staff, allows this to happen. “Critical incident reporting” in intensive care units provides an example of data gathering that uses mistakes constructively, without threatening individual reporters.
We should take note of the conclusions of the Boston study: that adverse drug events are a major cause of iatrogenic injury, that many are preventable, and that for every preventable event there are almost three potential ones. We should set in place systems that minimise them, not in the future but now.