Why good Root cause analysis is hard to achieve

Root cause analysis has real value. It gives organisations a structured way to investigate incidents, failures, delays, or repeated underperformance.

It forces attention beyond the immediate symptom, encourages examination of contributory factors, and can create a clearer shared account of what may have gone wrong. In settings where problems are messy, emotionally charged, or operationally complex, that structure matters. It is one reason RCA remains so widely used in healthcare. [1]

But RCA also has a familiar weakness. It can create more confidence in explanation of problems or symtoms rather than in their correction. The organisation may feel that the problem has been understood because a timeline has been reconstructed, contributory factors have been listed, and actions have been recorded. Yet that does not mean the analysis is technically strong, nor that the service is any closer to stopping the problem from recurring. Critiques of RCA in healthcare make this point clearly: the method has value, but it is often applied in ways that do not do justice to the complexity of the problems it is being used to investigate. [1]

Why RCA remains valuable

At its best, RCA does several useful things. It imposes order on a problem that may initially appear chaotic. It requires teams to move beyond the visible failure and ask what conditions, decisions, handoffs, constraints, or assumptions allowed that failure to occur. It can surface contributory factors that would otherwise remain implicit. It also creates a record of reasoning, which is useful for governance, organisational memory, and shared learning. These are not minor benefits. In complex services, even getting to a clearer and more disciplined explanation can be important. [1] So, the problem with RCA is not that it lacks value. The problem is that its technical promise is often perceived being greater than the quality with which it is carried out. The more accurate question is not whether RCA has value, but why good RCA is often so difficult to achieve in practice. [1]

Why good RCA is harder than it looks

Good RCA is not just a matter of filling in a standard template. It requires the problem to be examined with enough discipline that the organisation does not settle too early on a tidy but weak explanation.

Peerally and colleagues identify several recurring challenges in healthcare RCA, including overly linear reasoning, hindsight bias, variable investigation quality, poor problem framing, weak action design, and limited organisational learning.

Their argument is not that RCA never works, but that its common form often underestimates the methodological difficulty of investigating events in complex systems. [1]

That point is important because it explains why superficially competent RCA can still be technically weak. A team may believe it is being rigorous while still compressing a complex problem into a simple cause-and-effect sequence. It may name contributory factors without really testing them. It may rely too heavily on what appears obvious in retrospect. It may privilege the explanation that feels most coherent rather than the one that has been examined most critically. In that sense, good RCA is harder than it looks because it asks investigators to resist the human preference for tidy stories in settings where the truth is often untidy. [1]

Common reasons good RCAs are not achieved

One common weakness is linearity. Many problems in healthcare are produced by several conditions interacting over time. Traditional RCA can push investigators towards a sequence that looks cleaner than reality. A delay, incident, or recurring failure may be explained as though one factor led neatly to another, when in fact several factors were building pressure simultaneously. That kind of reduction can make the output easier to read, but less useful. [1]

A second weakness is hindsight bias. Once the outcome is known, it becomes easier to treat certain signals, decisions, or actions as more obvious than they really were at the time. That can distort the investigation, because what looks like a clear warning in retrospect may not have been clear in the live operational context. RCA is therefore vulnerable to producing explanations that are more convincing after the fact than they would have been beforehand. [1]

A third weakness is variable investigation quality. RCA is often treated as though the method itself guarantees rigour. It does not. The quality of the output depends heavily on how evidence is gathered, how questions are asked, how competing explanations are tested, and how willing the organisation is to tolerate ambiguity rather than rush to closure. Where time is short, hierarchy is strong, or the pressure for reassurance is high, the quality of investigation often suffers. [1]

A fourth weakness is broad thematic causation. Problems are sometimes framed in language that sounds explanatory but is too general to support action: “poor communication”, “capacity pressure”, “training issues”, “lack of escalation”, “human error”. These may be contributory themes, but they are not yet precise enough to tell the organisation what actually needs to change. A broad cause can make the RCA look complete while still leaving the real operating problem underdefined. [1]

A fifth weakness is generic action design. Once the causes are broad, the actions usually become broad too: improve communication, provide training, review the process, remind staff, strengthen governance. These actions are easy to agree and hard to oppose, but they often do not alter the mechanics of delivery that produced the issue. This is one reason recurrence remains a problem. Kellogg and colleagues argue that RCA outputs in healthcare have not consistently translated into reliable safety improvement, in part because the solutions generated are often weak or poorly matched to the problem. [2]

Why complexity makes good RCA more difficult

This becomes even more difficult in complex pathways and service models, where problems are rarely driven by one cause acting in isolation. Complexity science is useful here because it describes healthcare as a complex adaptive system: one made up of many interacting people, processes, technologies, constraints, and local adaptations, where cause and effect are not always linear or predictable. In these settings, outcomes often emerge from interactions across the system rather than from one simple failure at one point in time. [5]

That matters for RCA because traditional investigations can push teams towards a cleaner causal story than the system really supports. Causes interact. Constraints build on one another. A visible problem in one part of the pathway may be the downstream result of decisions, delays, bottlenecks, or adaptations elsewhere. In that kind of environment, an RCA that looks for one dominant root cause can become too linear for the problem it is trying to explain. Peerally and colleagues make this criticism directly in their discussion of how RCA is often applied in healthcare. [1]

A waiting list problem, for example, may not simply reflect demand. It may be reproduced by the interaction of referral growth, outpatient template design, diagnostic turnaround times, theatre availability, scheduling rules, and variation in how existing capacity is used. If the analysis reduces that complexity too quickly, the resulting action is likely to be too narrow. A local fix may then move the issue elsewhere, suppress it temporarily, or allow it to reappear later through another route. Complexity science is helpful here because it reminds us that some problems are better understood as patterns produced by a system than as the result of one easily isolatable cause. [5]

That does not diminish the value of RCA. It means its limits need to be understood. In a complex system, there may be no single root cause in the simple sense implied by the term. There may instead be a set of interacting conditions that together keep reproducing the problem. Good RCA therefore requires more than identifying “the cause”. It requires careful reasoning about which conditions mattered, how they interacted, and which of them are actually actionable. [1]

Why even useful RCA still may not change delivery

Even where RCA is done reasonably well, there is still another problem. Useful analysis does not automatically change how the service runs. A plausible cause is not yet a delivery intervention, and a completed investigation is not yet a changed system. Between the analysis and the improvement sits another set of tasks: translating findings into specific actions, assigning ownership, sequencing dependencies, defining measures, and reviewing whether the chosen action has changed the process it was intended to change. [2]

This is where many organisations stall. The RCA may have produced useful insight, but the route from insight to action is underdeveloped. The actions may still be generic. Ownership may be unclear. Follow-up may be weak. Measures may not distinguish between whether an action was agreed, whether it was introduced, and whether it changed the conditions that mattered. At that point, the RCA has value as explanation, but limited value as correction. [2]

The wider improvement literature reinforces this point. Taylor and colleagues’ review of PDSA found that the method was often used in ways that did not reflect its intended discipline, particularly around iteration, prediction, and learning. That is relevant here because the pattern is similar: an improvement method may be accepted and used, but still applied in a way that weakens its practical value. [3]

What stronger RCA practice looks like

Stronger RCA practice usually begins by being more technically disciplined about the analysis itself.

That means resisting overly linear explanations, testing competing interpretations more carefully, avoiding thematic vagueness, and distinguishing between broad contributory factors and causes specific enough to support action.

It also means recognising when the problem is too complex to be reduced honestly to a few root causes. [1]

But stronger practice also requires something beyond the investigation. RCA becomes more useful when its findings are treated as the start of intervention design, not the end of the thinking. If the organisation cannot link its analysis to specific actions, named owners, review points, and measurable follow-through, then even a technically decent RCA may do little more than describe the problem more elegantly. [2]

Conclusion

RCA can bring structure, discipline, and shared understanding to difficult problems. But good RCA is harder than it looks, and it often doesn't achieve its intended purpose. Investigations become too linear, hindsight shapes the explanation, causes are framed too broadly, actions become generic, and complex systems are reduced to tidy narratives that do not hold up well enough in practice. [1]

And even where RCA does produce useful insight, that still does not guarantee change. The method helps organisations understand why a problem may be happening. It does not by itself change the conditions that keep reproducing it. That is why RCA adds most value when it is both technically stronger as an investigation and more deliberately connected to the practical work of changing delivery. [2]

References

[1] Peerally MF, Carr S, Waring J, Dixon-Woods M. The problem with root cause analysis. BMJ Quality & Safety. 2017;26(5):417–422.

[2] Kellogg KM, Hettinger Z, Shah M, et al. Our current approach to root cause analysis: is it contributing to our failure to improve patient safety? BMJ Quality & Safety. 2017;26(5):381–387.

[3] Taylor MJ, McNicholas C, Nicolay C, Darzi A, Bell D, Reed JE. Systematic review of the application of the plan–do–study–act method to improve quality in healthcare. BMJ Quality & Safety. 2014;23(4):290–298.

[4] Dixon-Woods M, McNicol S, Martin G. Ten challenges in improving quality in healthcare: lessons from the Health Foundation’s programme evaluations and relevant literature. BMJ Quality & Safety.2012;21(10):876–884.

[5] Greenhalgh T, Papoutsi C. Studying complexity in health services research: desperately seeking an overdue paradigm shift. BMC Medicine. 2018;16:95.