How I have used root cause analysis

In this article, Graham Lavender, a supplementary prescriber in a Southampton GP practice, describes three cases in which medication reviews did not result in an intended outcome and how applying root cause analysis allowed him to learn from each.

The HOMER study,1 which reported on home-based medication review by pharmacists, resulted in significant concerns over the value and, indeed, safety of pharmacist medication review. The study (a randomised controlled trial involving 872 elderly patients recruited during an emergency admission) indicated a statistically significant higher rate of hospital admissions as a result of pharmacist medication review. Although the study design has been criticised,2 there has been no detailed look at the individual pharmacist interventions that may have resulted in the hospital admissions.

In recognition of the increasing clinical role of pharmacists, the Royal Pharmaceutical Society issued a revised “Clinical governance framework for pharmacist prescribers” in October 2005 (PDF 170K). One recommendation of the framework is that pharmacist prescribing should be considered in clinical risk management programmes, including root cause analysis (RCA). An RCA investigation traces the cause and effect trail from a failure. It is a step-by-step method that leads to the discovery of a fault’s first (or root) cause.

To date, no attempt has been made to identify the root cause of the increased hospital admissions in the HOMER study and, until such an analysis is made, no sound conclusions can be made on the validity of the individual pharmacist interventions. Although RCA is not the only clinical governance tool recommended by the Society, it has the advantage over clinical audit, for example, because it can analyse a sequence of events.

As a prescribing pharmacist of 30 months’ standing I have followed the Society guidelines to assess clinical medication reviews that I have carried out and which have resulted in other than the desired outcome. The process is similar to a practice encouraged during my supplementary prescribing course, in which the supplementary prescriber had to write up and analyse all significant events. I continued with this practice after the course, entering patient reviews in which a significant event occurred, onto my online continuing professional development file. Significant events are any incident where further reflection might be advantageous in identifying not only the cause of an adverse incident but also any training requirements and plans needed as a result.

RCA has enabled me to reflect in detail on medication reviews where the outcomes were not as intended. For each case, a root cause can be identified from an examination of the sequence of events and subsequent training can be used to prevent a repeat occurrence. In two of these three cases, the patients required secondary care and, although my GP practice was happy with my reviews, it was shown that, in fact, it might have been possible to avoid the hospital admissions. It should be noted that it is not necessary to prevent the root cause from occurring — it is merely necessary to break the chain of events at any point so that the final “failure” does not occur, for example, with Mrs KB (case 3), where there was the opportunity to break the sequence when I later reviewed the patient.

For Mr TM (case 2) incomplete counselling skills were identified as the root cause of the incident. Assumptions were made that the patient had a good understanding of diabetes and a need to review the quality of the information given to patients and to check their understanding was identified.

RCA allows a detailed breakdown of a sequence of events and the identification of the actual cause or steps in the sequence of events which might have caused the adverse outcome. The HOMER study included no such analysis and without one it is not possible to identify the root causes of the increased hospital admissions. This reduces the validity of its conclusion that pharmacist medication review increases hospital admissions. There will always be adverse outcomes when a significant number of interventions are made; the challenge is to identify the causes, recognise any training needs and meet those needs.

Clearly if RCA identifies individuals with unacceptable numbers of adverse outcomes and training is not addressing the problem, then questions must be asked about the suitability of the individual for the role. With pharmacists now taking on increasing roles, the potential for adverse outcomes is ever present. Only by using tools, such as RCA, can we ensure that patients have the quality of care they deserve and the profession of pharmacy can meet the standards of care that its new roles demand. The value of medication review as an opportunity to break the sequence of events leading to an adverse outcome must be recognised as well as the inherent advantage of regular medication review for all patients on regular medication.

Case 1: Stopping a patient’s medicine in order to follow guidelines

Mrs JG, a 62-year-old female had had a gastroscopy two years ago, which was normal aside from some gastric inflammation attributed to the use of non-steroidal anti-inflammatory drugs. Her appointment was part of a review of all patients on long-term proton pump inhibitors (PPIs) using local guidelines similar to the National Institute for Health and Clinical Excellence dyspepsia guidelines.

When I saw Mrs JG, she was taking omeprazole 20mg od, and co-proxamol prn for intermittent osteoarthritis. No over-the-counter medicines were being taken. In her notes, the gastroscopy report suggested prescribing a PPI for four to six weeks to resolve irritation to the gastric mucosa after which the drug was to be stopped and only used if symptoms returned. However, it appeared from the patient’s notes that she had been taking omeprazole 20mg for the two years because no medication review had taken place (this in itself is an example of why it is important to conduct regular medication reviews).

Having confirmed that the patient was asymptomatic it was agreed that she would stop taking the omeprazole and this would be reviewed at a later date. I was comfortable with this course of action because there was no reason to continue a relatively powerful drug when, in this case (ie, after gastroscopy), there was no indication of long-term need, and I advised Mrs JG to return if she experienced any gastrointestinal symptoms.

At a practice meeting some months later I was informed that Mrs JG had been admitted to hospital with a gastrointestinal bleed but had made a full recovery. I seemed to be the only one at the meeting concerned that stopping her omeprazole had been the cause. The comment by the GP prescribing lead was: “We looked at your intervention. It followed guidelines. It was in the interest of the patient not to have unnecessary medication and we have no problems with it. Unfortunately any intervention you, or any of us, make can have adverse consequences which, despite best practice, are simply unpredictable and should not change practices unless evidence suggests otherwise.”

Initially, I was happy to accept the GP’s assessment, but when I applied RCA a sequence of events that might have prevented Mrs JG’s gastrointestinal bleed was highlighted. The step that caused me concern was the sudden stopping of omeprazole after two years’ continuous therapy. Although the local guidelines and the gastroscopy report said that therapy should be stopped after four to six weeks because the patient had been on omeprazole for two years a slow reduction in the dose of omeprazole would have been more appropriate. Although the literature is sparse on this point, it could be assumed that the lower acid levels in the stomach as a result of long-term omeprazole therapy might have led to some reduction in gastric protection and a sudden rebound in acidity may have had adverse consequences.

Case 2: Starting a patient on a new medicine for a newly diagnosed condition

Mr TM, a 69-year-old man who had been treated for prostate cancer in 2004, had an appointment with me following a raised fasting glucose level and an HbA1c of 8.4. He had been first seen by a GP in my practice (my independent prescriber) who agreed a clinical management plan. I prescribed metformin 500mg tablets, one with breakfast for the first week, adding one with the evening meal for the second week and, finally, adding in one with lunch thereafter. The patient’s BP measurements averaged 150/90 over the past three readings and his cholesterol was raised but I explained that we were first going to treat his high blood glucose levels before looking at his blood pressure and then his cholesterol level. Mr TM had no questions and I asked him to come back to see me in six weeks, following another HbA1c test.

At the next appointment, however, his HbA1c was almost the same (8.6) and before I could ask him any questions he said “I took the course of tablets, am I now cured of diabetes?”

RCA requires a detailed record of events if the reason for an undesirable result is to be determined. In this case, although the notes on examination, therapy and goals were detailed, no record was made of the counselling I gave on starting the patient on metformin. Several weeks after an event it is difficult to remember an individual consultation but, given the sequence of events and the outcome, I am prepared to admit that I had not been clear about the long-term nature of the therapy. It is relatively easy to make an assumption, based on your own knowledge, but from a patient’s perspective there is no reason to assume that a treatment is for life. My RCA for this patient identified a failure in my counselling. It indicated the need to be more specific in future and, if necessary, to provide written instructions and information.

Case 3: Learning to question an independent prescriber’s decision

Mrs KB, a 76-year-old woman with a history of chest pain, had been referred in 1999 to hospital for further investigation for heart disease and, considering her symptoms and age, was prescribed aspirin, a statin (simvastatin 40mg od) and an antihypertensive agent (losartan 50mg od). The result of an exercise test was equivocal and a follow-up stress test showed no indication of cardiac ischaemia. However, the hospital suggested that Mrs KB continue taking the aspirin, statin and the antihypertensive as a precaution. The patient was a non-smoker, with a body mass index of 24.

During the medication review, I found that, at the patient’s request some months earlier, the GP had stopped prescribing aspirin. This was because Mrs KB had experienced some bruising (a common side effect in elderly people taking aspirin). She had no chest pain at rest, but experienced tightness in the chest on excessive exercise, which responded to glyceryl trinitrate. However, these episodes were rare. BP was 148/90 and readings over the past 12 months had all been in excess of 140/85. Electrolytes and renal function were normal. I advised Mrs KB that, because of her high BP readings, we should increase the dose of losartan to 100mg and to repeat electrolytes and BP measurements after a month. I accepted the reason for stopping the aspirin and recorded in the patient’s notes that aspirin was contraindicated due to excess bruising, with a reference to the date of the consultation where this had been decided with the GP.

Less than a month later, Mrs KB was seen at the surgery having woken with chest pain that did not resolve with GTN, sweating, nausea and palpitations. She was admitted to hospital with an acute myocardial infarction and discharged after having a stent inserted. I saw her after she was discharged, now taking aspirin and clopidogrel. Again she was heavily bruised but she had accepted that the risk of a repeat MI outweighed the concern about bruising. I later discussed the patient at a practice meeting. The conclusion was that the patient’s overriding concern about unsightly bruising and the subsequent stopping of prescribing must be seen as a clear example of patient’s choice. Clearly, this was most likely to have caused the MI but, using RCA analysis, there was evidence of opportunities, on a number of occasions, to break the sequence that led to the MI.

The exercise test is known to have a limited diagnostic value (especially for elderly patients) and, although followed by a stress test, a negative result does not completely exclude coronary artery disease — it is well established that around 30 per cent of atheroma grow into the artery wall, giving little obstruction of blood flow and hence not necessarily identified on exercise and stress testing. The aspirin had been initially prescribed with this possibility in mind and this was not recognised on at least two occasions: the GP had failed to balance the risk of stopping aspirin against the patient’s concern about bruising and I had failed to challenge the GP’s decision. In this case, RCA not only found the primary event that led to the adverse outcome but also identified subsequent events that presented an opportunity to reverse the original error.


1. Holland R, Lenaghan E, Harvey I, Smith R, Shepstone L, Lipp A et al. Does home based medication review keep older people out of hospital? The HOMER randomised controlled trial. BMJ 2005;330:293.

2. Hay JW. Pharmacist medication review study design concerns. BMJ 2005;330:347.

Last updated
The Pharmaceutical Journal, PJ, December 2006;()::DOI:10.1211/PJ.2022.1.149947

    Please leave a comment 

    You might also be interested in…