CHAPTER 1. INTRODUCTION
2.7 Discussion
2.7.1 Main findings
An MAE rate of 4.3% for non-IV doses was identified. Overall findings indicate the processes of administering apparently straightforward non-IV doses appear to have a number failings in addition to MAEs that potentially lower their quality and safety; some were related to individual procedural violations (such as not confirming the patient’s identity prior to administration and inaccurate administration documentation) while others indicate potentially more organisational-related problems (such as unavailability of medication and less than four hour intervals between consecutive drug rounds). In the current study, six quality and safety measures were combined to reveal that only 11.8-25.5% of doses observed complied with the six nominal standards of good practice for medication administration at the study site.
‘Failures’ were apparent for each of these six quality and safety measures; two variables associated with the lowest ‘quality’ were timeliness of drug rounds and confirming the patient’s identity prior to administration. Findings relating to the following are next discussed in detail: (1) MAEs and considerations for use as a quality and safety measure, (2) discrepancies between prescribed time, scheduled drug round time, and actual drug round times – an underlying latent failure? (3) patient identity check – an inadequately used defence barrier, and (4) medication retrieval and storage.
2.7.2 MAEs and considerations for use as a quality and safety measure
The primary measure of quality and safety in the current study was the MAE rate, which was 4.3%; or approximately one MAE in every 25 doses. This MAE rate is lower than the 7.0% for non-IV doses previously reported by Franklin et al (2007) that was used to derive the sample size, but consistent with MAE rates of 3.0-8.0% for non-IV doses reported in other similar observational studies (Dean et al., 1995; Ho et al., 1997; Cavell & Hughes, 1997; Taxis et al., 1999). Exploratory sub-analysis of MAE rates at different times of day suggested that more
MAEs occurred during the 18:00 and 22:00 drug rounds than at 08:00. This is different from other studies which have found the 12:00 drug round to be associated with the most MAEs (Franklin et al., 2008; Ho et al., 1997), and may reflect differences in inherent common causes of variation (chapter one) associated with the specific ward setting. For example, Ho et al (1997) found higher MAE rates during the first 48 hours of admission and in the first 48 hours of prescribing on an acute admissions ward; the patient turnover on that ward was much higher than on the current study ward of general medical patients and therefore the potential effect of patient admissions may not have been a factor in the current study. Franklin et al (2008) attributed a higher MAE rate at midday to a potentially greater number of interruptions and activity on the ward. While the study by Franklin et al (2008) was not designed to assess the effects of interruptions on MAEs, a separate study was identified that investigated the relationship between interruptions and MAEs. Westbrook et al (2010) observed a total of 4,271 drug administrations by nurses at two major teaching hospitals in Australia. The researchers identified a correlation between the number of interruptions and errors; each interruption was associated with a 12.1% increase in procedural failures and a 12.7% increase in clinical MAEs (Westbrook et al., 2010). However, findings from the present study suggest there were more interruptions at 08:00 and 22:00 than at other times (although not statistically significant), however, the MAE rate was lowest at 08:00 (1.6% of OEs; 95% CI 0.2-3.0%) and highest at 18:00 (10.4%; 95% CI 4.1-16.6). Overall, these findings suggest other factors, in addition to or other than interruptions, played a more prominent role in affecting MAE rates in the present study. Based on analysis of medication administration documentation, more doses were inaccurately documented during 18:00 and 22:00 than at other times (although statistical significance was not explored due to small sample); these were mainly due to doses that were omitted and not signed, suggesting potential oversight to be more problematic later in the day. Overall, findings from this study suggest that analysis and interpretation of MAE rates alone may not be sufficient to explain the potential causes.
Concomitant data collection of other parameters associated with the medication
administration provides an enhanced understanding of the safety of medication administration process.
2.7.3 Discrepancies between prescribed time, scheduled drug round time, and actual drug round times – an underlying latent failure?
Observations revealed that 82% of doses administered were prescribed for the same time as a scheduled drug round time; this meant that 18% of doses administered were not planned to be given around the prescribed time. This was because doses were also prescribed regularly for 06:00, 14:00, and 20:00. The mismatch between the prescribed times and the scheduled drug round times present a potential technical problem in the measurement of quality and safety. Timeliness is generally measured against the prescribed time, however in practice, timeliness is usually only relevant for a relatively small group of drugs: for example, time-critical medicines such as anti-Parkinsonian drugs, and time-interval time-critical medicines such as those that require administration four or more times a day. Thus, while some doses prescribed for 06:00 may be administered at the 08:00 drug round, these are not always a problem and do not infer ‘lower’ quality or safety. However, for time-interval critical doses, timing of administration may be more problematic. In particular, findings from the current study suggest that some consecutive drug rounds were started less than 4 hours apart, and others over 11 hours apart; these could potentially result in sub-optimal drug profile levels of interval critical drugs. Consequently, it may be more useful to assess timeliness for time-critical and time-interval time-critical medicines rather than for all.
Unfortunately, it was rarely clear when timeliness was critical on the drug chart without knowledge of the medications prescribed; such information was generally not provided.
Instead, much of the responsibility to ensure that time-critical doses are recognised and administered on time seems to be burdened on nursing staff; this is likely to require a
combination of medicines knowledge and memory to identify and keep track of such doses for all patients. This highlights a potential underlying limitation of the medication prescribing and administration system; which may also be considered a latent failure.
2.7.4 Patient identity check – an inadequately used defence barrier
Of all the measures recorded in the current study, the largest deviation from standard good practice was the percentage of doses given after the patient’s identity was confirmed by the nurse (37.5% of OEs). This figure is higher than 17.4% of doses previously reported in a similar observational study (Franklin et al., 2007). One possible reason for higher compliance may be related to the presence of an observer. In the current study, the observer noticed that some newly observed nurses (i.e. not observed during the piloting stage) tended to check the first few patients’ identity during the drug round but would then revert to addressing subsequent patients by their first name as the drug round progressed. Although the absence of a check may not lead to patient harm, 66% non-compliance is high and evidence from other research suggests this preventable risk to patients has yet to be resolved (Franklin et al., 2007; Koppel et al., 2008). Furthermore, it has been suggested that not checking a patient’s identity prior to administration may be indicative of non-compliance to other procedures that also increase the risk of error (Westbrook et al., 2011). In the UK, in 2007, there were 2,781 incidents of mismatch between patient and medicine reported to the NHS NRLS; of these, two patients suffered severe harm and there was one patient death (Cousins et al., 2007). These incidents may have been avoided if the patient’s identity was confirmed as a match to their medication order prior to drug administration.
2.7.5 Medication retrieval and storage
The findings from the current study suggests that despite the use of PODs and OSD stored in patient bedside medication lockers, 11.3% of doses administered were not available at the
patient’s bedside. Having the right medications at the patient’s bedside should help to minimise time spent looking for medication and it was inferred that measuring the percentage of doses that were unavailable at the patient’s bedside provides an indication of the amount of
“excess travel” nurses make during drug rounds. However, from observations, a number of nurses would routinely store medications in their pockets in anticipation that a medication will not be available at the patient’s bedside, for example, syringes of enoxaparin and paracetamol tablets. This suggests that potential problems or inefficiencies associated with medication storage exist and are perhaps common. While experienced nurses may take preventative actions to manage the potential inefficiency, others may take more time to retrieve medications during drug rounds unless the underlying potential medication storage inefficiency is addressed.
2.7.6 Strengths and limitations
A limitation of the current study was that the process improvement initiative was not implemented on the study ward and therefore the potential effects on the quality and safety of the medication process could not be examined. However, a strength of the current study was the inclusion of a range of quality and safety measures to study the medication administration process. Another strength was the use of observation; this allowed MAEs to be recorded more accurately and consistently, and also provided the context that facilitated data analysis. In general, nurses did not seem to mind being observed and no obvious change in behaviour was identified other than those associated with confirming patient’s identity. A limitation was that a number of other potential confounding factors that may affect the quality of the medication administration process were not collected: nurse experience, number of admissions and number of new medication orders. It would be useful to know what the extent (if any) of these factors had on the MAE rate as there is the possibility that some of the MAEs may be restricted to a small number of patients, staff or medication order. Furthermore,
severity assessments of the MAEs were not conducted. It would have provided an additional dimension to the potential impact of the MAE rate and therefore a stronger safety indicator of the quality of the medication administration process. Finally, the current study was based on one ward, by one observer which limits the generalisability of the study.
2.7.7 Implications for practice
In the current study, identifying time-critical and time-interval critical doses was found to be a potential problem; EPMA systems potentially offer an opportunity to resolve this by employing a design function that alert nurses to time-critical and time-interval critical doses; thus minimising the need for individuals to rely on their memory. Additionally, use of bar-code technology has been associated with a greater compliance in confirming a patient’s identity prior to administration in a UK hospital (Franklin et al., 2007). However, care should be taken when designing, choosing and/or implementing technologies in health care; as discovered from research of technological workarounds (Koppel et al., 2008), implementation of electronic systems may not resolve the problem entirely, and potentially create new problems.
2.7.8 Future research
Several definitions for MAEs and associated subcategories have been reported in past studies in the UK, the US, and other countries. While it was outside the scope of the current study to review all the definitions used, it was soon realised that the implications of such diverse terminology posed a potential barrier for interpreting and comparing research in the area; a systematic literature review was therefore carried out separately to the current study and is described in chapter three of this thesis.
In addition, the potential problem of inefficient medication storage also warrants further research. Findings from the current study suggest nurses took preventative action to avoid
having to travel back and forth between the patient and the stock cupboard for some drugs.
While some preventative practices may have been considered risky (unused medicines may be left in the pocket and expire, nurses may accidently take medicines home), these practices may have potentially increased drug round efficiency and also reduced the number of interruptions to nurses during drug rounds. Research has found 22% of interruptions occurred when nurses were in the medication storage room (Potter et al., 2005). As highlighted earlier, Westbrook et al (2010) identified an association between the number of interruptions and errors. Furthermore, the researchers also found the risk of a major error doubled (4.7% of errors) when there were four interruptions than when there were no interruptions (2.3% of errors). Thus, by taking preventative action to minimise excess travel and interruptions, it is possible that the risk of MAEs may also be reduced; however, this inference is outside the scope of the present study.
While the problem of medication not being available may be partly because the ward no longer used conventional drug trolleys during drug rounds, previous research have found that doses are not always available from drug trolleys even when they are used (Dean et al., 1995;
Taxis et al., 1999). Thus potentially more efficient methods of storing medications for timely retrieval and administration may be required. However, the problem (and thus the solution) of medication storage is likely to affect many parts of the NHS. Consequently, a more systems-based approach to understanding this is required. First however, we need to identify what current types of hospital medication systems exist, including ward-based medication storage facilities. A national survey of hospital medication systems in English hospitals is presented in chapter four.