Recording previous adverse drug reactions—a gap in the system

Abstract

Aims

To measure the accuracy of recording of previous adverse drug reaction (ADR) history in patients admitted to a teaching hospital before and after an education programme.

Methods

One month survey of patients on one medical and one surgical ward, repeated after a 1 month education programme. Patients answered a questionnaire about previous ADRs and this information was compared with that in all relevant sections of their medical records and medication charts.

Results

Of 117 patients at baseline, 50 had a total of 81 previous ADRs. Only 75% were recorded on medication charts and 57% and 64%, respectively, in medical and nursing notes. In the post education survey of 124 patients, 56 had 105 previous ADRs, 85% were recorded on medication charts and 64% and 70% in medical and nursing records. These differences were not significant. Serious ADRs were also poorly recorded at baseline but, due to intervention by ward pharmacists, their recording on medication charts improved significantly after education. Pharmacists also significantly improved the quality of description of previous ADRs in both parts of the study.

Conclusions

Previous ADR history obtainable from hospital patients is poorly recorded in medical records and an intensive education programme only produced a significant change in recording by ward pharmacists. Better strategies are needed to improve this essential aspect of history taking.

Introduction

Adverse drug reactions (ADRs) are commonly classified into type A, which are predictable because they relate to the pharmacological actions of a drug or to toxic plasma concentrations, and type B which are unpredictable because they are due to idiosyncrasies, allergies or unforeseen long-term outcomes [1]. There is another category of predictable ADR that occurs when patients are prescribed medications to which they have previously experienced an ADR. This event is used as a quality measure by the Australian Health Care Council [2] but there are surprisingly few reports about the accuracy of ADR histories in hospital records.

Cook & Ferner found considerable under-reporting of ADR history that could be readily obtained from hospital patients on direct enquiry [3]. We have reported similar findings in a study of geriatric patients where previous ADRs, a significant predictor of hospital admission with a drug related problem [4] were absent from the medical notes in up to 50% of patients [5]. In an attempt to improve reporting of previous ADRs we undertook an intensive staff education programme about the importance of ADR history. An audit of ADR history reporting was conducted for 1 month before and 1 month after the education programme.

Methods

Subjects were the patients on one medical and one surgical ward for two, 1 month periods. To be included they had to speak English and be able to understand and answer questions. They gave replies to a specially designed and structured questionnaire about previous ADRs. This included drug names (as remembered by the patient), a detailed description of the reaction and leading questions about eyedrops, inhalers and skin creams. They were also asked specifically about any reactions to penicillin and aspirin. This information was taken to be the ‘gold standard’ for history that could be obtained from the patient. All patients were interviewed by the same person (TR).

Relevant sections of their notes were searched for documentation and description of the ADRs reported to the researcher. For the purposes of analysis the findings were consolidated into groups as follows: Emergency Department medical admission, ward medical admission and Anaesthetic Room patient sheet were classified as notes used by medical staff (MN). Emergency Department nursing admission, Ward nursing admission, Patient Care Plan and Operating Theatre patient sheet were classified as notes used by nursing staff (NN). Medication charts (MC) are filled in by doctors and include a special section for drug names and details plus a space for a brightly coloured alert sticker and were reviewed separately as were ‘ADR Alert’ armbands. Ward pharmacists also take medication and ADR histories and record their finds on the medication charts in red ink. Their input was analysed separately.

After the baseline data collection an intensive, 1 month education programme was conducted throughout the hospital. This focused on the importance of taking ADR histories. Posters and handouts were designed using the theme ‘Want to be a Detective? Learn How to Recognize and Report Adverse Drug Reactions’. Content covered when to suspect an ADR and how to recognize one. The theme was to suspect an ADR as the cause of any unexplained symptoms or signs especially when drugs had a dose increase were started or new durgs. Lists were provided, with appropriate graphics, of ADR symptoms in different body systems, e.g. swollen lips, bronchospasm, postural hypotension, etc.

The importance of recording ADR history was strongly emphasized and procedures for documenting and reported ADRs were tabulated in an eyecatching manner. The programme included 36 inservice sessions for nursing staff, a lunchtime talk for ward pharmacists, two short talks for junior medication staff, information packs for all registrars and weekly poster displays in the hospital foyer. These were attended by the researcher and some ward pharmacists who answered any queries. A weekly prize was offered for the best answers to a list of questions. In addition posters were put up in all the wards and pamphlets made available for all staff.

In the following month the questionnaire and data collection and analysis were repeated on the two wards originally surveyed.

Recording of ADRs was expressed as a proportion of the number of ADRs reported in each section of the notes over the number reported in the patient interview. This was also done for description of ADR symptoms and signs.

Two physicians independently reviewed the entire list of reported ADRs and classified their severity. Serious reactions were defined as potentially life threatening ones where it would be dangerous to prescribe the drug again for that patient. The evaluators agreed initially on the classification of all but five reported ADRs. Differences related to interpretation of severity of symptoms and, after discussion, complete agreement was reached on categorization.

A two-tailed, chi-squared test (with Yates correction) was applied to each of the ‘accuracy’ proportions found before and after the intervention. P < 0.05 was considered significant. 95% Confidence intervals were calculated using McNemar's test for paired proportions.

Results

There were no differences in demographic features or number of previous ADRs between the medical and surgical patients and their results were therefore combined. At baseline collection there were 117 eligible, consenting patients and, after intervention, 124 patients.

Table 1 shows the total number of patients who gave a history of ADRs (some had more than one) compared with the percentage recorded in each section of the medical notes before and after the education programme.

Table 1.

Incidence and recording rates of ADR, before and after education programme.

				Precentage recorded in:
	Total patients	Patients with previous ADRs	Total ADRs	MN	NN	MC
Baseline	117	50	81	57%	64%	75%
Post education	124	56	105	64%	70%	85%

The percentage of omissions was high and the minor improvements shown after the education programme failed to reach statistical significance. ADR ‘alert’ bands were also poorly utilized. At baseline only 19 (38%) of the 50 patients with reported ADRs were wearing a band. This increased to 31 (55%) of the 56 with reported ADRs in the post education group (NS). The percentage of accurate recording was not significantly different between those ADRs independently classified by two doctors as minor or serious. At baseline 12 (15%) of the 81 previous ADRs were considered potentially life threatening and failure to record these was MN 2, NN 5, MC 3. In the posteducation survey 18 (17%) were considered serious with nonrecording being; MN 2, NN 6, MC 0. The improvement on the MC was entirely due to intervention by ward pharmacists.

During the course of the study no patient was given a drug to which they had experienced a previous ADR but previous serious reactions omitted from the records included worsening asthma with β-adrenoceptor blockers, angioedema with iodine and penicillin and neutropaenia with fusidic acid.

Pharmacists significantly improved the number of reports on MC (Proportional difference 23% [95% CI 14, 32%]) (Table 2 P < 0.05) and also quality of the description of the ADRs. Of the 81 ADRs reported at baseline only 11% had the nature of the reaction described on the MC prior to the ward pharmacist's visit and this improved to 36% after the visit (95% difference 25% CI 15.5, 24.5%) (P < 0.01). Of 105 posteducation ADR reports, only 20% had the nature of the reaction described on the MC prior to the ward pharmacist's visit but this improved to 85% after the visit (95% difference 65% CI 55.5, 74.5%) (P < 0.001). The improvement in pharmacist input to number of reports after the education programme was also significant (proportional difference 34.5% [95% CI 25.2, 43.8%]) (P < 0.05).

Table 2.

The contribution of ward pharmacists to ADR history recording on medication charts.

	Number of ADRs reported	Number (%) of ADRs recorded on MC prior to pharmacist visit	Number (%) of ADRs reported on MC after pharmacist visit
Baseline	81	41 (51)	61 (75)
Post education	105	52 (49.5)	88 (84)

Discussion

These results are similar to those of Cook & Ferner [3] and indicate a very poor level of previous ADR history taking by medical staff and nurses. Even those that were recorded were rarely described in detail. The equally poor recording of serious ADRs is of particular concern as the potential for life threatening incidents was high. If they occurred they would almost certainly be indefensible in the Courts. Patients are often confused and upset on admission to hospital and staff are frequently rushed and concerned with immediate care. We could not determine if in some cases doctors had taken an ADR history and failed to record it, but for appropriate communication with other staff, both during and after the admission, accurate recording is essential. The admission history may well have deficiencies and there is a good case for repeating it on a subsequent occasion. This is often done by ward pharmacists who provide a vital safety net but they are in short supply and may not be able to interview all patients. We suggest that Health Authorities should allocate increased resources to employ more ward or clinical pharmacists. A recent, as yet unpublished study, has shown that interventions by ward pharmacists can produce significant reductions in patient morbidity personal communication, Society of Hospital Pharmacists of Australia).

It was particularly disappointing that the very intensive and visually attractive education campaign had so little success with the only improvement in reporting of serious ADRs being achieved by the pharmacists. In both surveys they were more likely than other staff to record full descriptions of ADRs and they did this even better after the education campaign. Unfortunately neither doctors nor nurses showed a similar improvement.

It is clear that new strategies are needed to improve the performance of all staff in this important area. We have initiated two processes in an attempt to address the problem in our institution. Firstly we have introduced a preadmission clinic for all elective surgical patients. In addition to general history taking and examination by a doctor, a pharmacist takes a history of current medications and previous ADRs. This procedure has not been formally evaluated but has been enthusiastically welcomed by anaesthetists and surgeons. Secondly we have developed an ADR summary sheet which is placed at the front of the consolidated medical records. It stays in this position through all subsequent admissions and provides an ongoing ADR history for the patient. It can be added to if further ADRs occur. For those patients for whom it has been completed it has proved very useful but it will take time to achieve widespread usage. It is likely that in the next few years we will have electronic records linked to online prescribing. This should improve prescribing safety but accurate histories will still need to be taken.

We are also collaborating with other teaching hospitals in New South Wales to improve all aspects of ADR reporting. We plan to recruit assistance from educationalists, behavioural scientists and communication experts to develop appropriate material to train health professionals about the importance of taking accurate and detailed ADR histories.

Many serious problems can occur while patients are in hospital [5]. Not all are avoidable but the majority could be prevented with adequate information which may well be deficient in areas additional to the one addressed in this study. Some ADRs will always occur but in an ideal world no patient should be given a drug to which they have previously had a serious ADR. In addition a previous history of any type of ADR to any drug is a major risk factor for future ADRs to any drug [4]. If a decision is made that the likely benefits might exceed the risks, for example use of an essential antibiotic when the previous event was a minor rash, it should be recorded in detail in the medical records.

In this study we were not able to establish whether the patients' answers to our questionnaire corresponded to their ‘true’ ADR history. To find out would require another, more resource intensive study. However in real life there is rarely time to check details of events that have occurred elsewhere and the patients' recollection is the principle basis of all history taking and recording. We consider that our results are representative of everyday deficiencies in history taking. All health professionals should be more aware of this problem.

Ferner & Aronson have recently described a classification of all types of medication errors [7, 8]. They categorize the prescription of a drug for a patient with a known previous ADR to that drug as a ‘knowledge based mistake’. Their system is a useful one for documenting and analysing errors but different types of error will need different solutions. Perhaps to start with the standard classification of ADRs should be expanded to include ‘Type P’-preventable ADRs.