SafetyInsights.org

This explored the application of the hierarchy of control (HOC) in the context of healthcare. 42 risk controls from four clinical teams were evaluated against the HOC.

In setting up the context, it’s said that “Though healthcare is often exhorted to learn from ‘high-reliability’ industries, adopting tools and techniques from those sectors may not be straightforward” (p39).

Methods from other industries are often deployed in healthcare, including RCA, proactive structured risk assessments (e.g. FMEA), and HOC.

They argue that the most variants of HOC is based on three principles:

1. Safety incidents are seen to occur as a result of exposure to hazards – and that eliminating exposure to the hazard via risk controls will be the most effective
2. HOC methods assume that “humans are fallible” (p40) and thus controls that “rely on ‘hard stops’ and forcing functions—and thus minimize reliance on human behaviour—are seen as more effective” (p40)
3. An assumption that risk controls higher on the hierarchy are, on the balance, harder to design and implement.

HOC typically classify administrative controls as weak as “they are thought to address only the symptoms of more institutionally engrained problems rather than the true causes” (p40). Existing healthcare data suggests that admin controls, which are ranked as the least effective by HOC methods, are the most frequently proposed solutions to hazards.

A recent systematic review found that ~3.3% of risk controls classified as elimination and by comparison 78% were administrative. They concluded in this review that “that some of the most widely used risk control strategies in healthcare, such as training and education, ‘might do more harm than good” (p40).

Results

As expected, the majority of the 42 proposed risk controls were administrative (35), 6 were engineering-related and 1 aligned with substitution. The strong substitution control was new pumps for mechanical prophylaxis (‘strong’ according to the NIOSH HOC).

The six engineering-related controls were things like new software for checking prescriptions or other pieces of technology.

Admin controls were heterogeneous: 10 involved changes to the organisation and delivery of ward care, ten other controls involved formalising roles and responsibilities and the use of a skill matrix and others involved training or improving communication.

Discussion

It’s said that “if a hierarchy of control model adapted from high reliability industries is applied to risk controls introduced by clinical teams in response to proactive identification of hazards in their clinical pathways, most risk controls would be deemed ‘weak” (p41, emphasis added).

These findings are consistent with previous data, suggesting that “a proactive hazard detection approach does not result in a distinctive pattern of risk controls” (p41).

They say that while one interpretation of these findings is that healthcare organisations are “simply very poor at generating risk control”, another interpretation is that the HOC method may not be “appropriate for the specifics of healthcare” (p41, emphasis added).

They argue that the ability of the HOC approach to predict success or failure of risk controls in clinical settings is challenged by three issues: (p41)

1. HOC approaches tend to categorise interventions based largely on superficial and visible characteristics without “sufficient attention to the heterogeneity of risk controls and the quality of the design, delivery and intervention” (p41). For instance training may be classed as a weak action under a HOC but can be delivered in a wide range of styles, including immersive group-based simulations. Thus, classifying training as a weak admin control is “reductive and misleading” (p2).
   1. Further, other “weak” actions under the HOC (like redesigning team structures or introducing or modifying structured communication strategies) can be very effective as shown by prior research.
2. HOCs don’t account for the fit between hazards and planned interventions and thus, tend “to assume that interventions will operate in the same way regardless of context” (p42).
   1. HOCs tend to underemphasise the role of “congruence” between a control and its target. For instance, behavioural changes rank low on HOC but may be impactful if designed to be highly congruent with the issues and backed by a “rigorous theory-of-change”.
3. HOCs may “have little to offer to our understanding of the social and organizational factors that contribute to the success or failure of safety interventions in healthcare” (p42).
4. It’s said controls are created via two intertwined elements: the content of the intervention (the thing itself thought to reduce the risk), and the supportive/facilitative factors that make it possible for the intervention to be implemented in specific contexts. HOCs seem to account for primarily the former and not the latter. Thus, HOCs “do not describe how interventions are carried out nor how they become embedded and sustained over time” (p42).

HOCs are said to “conceptualize risk controls in a direct and mechanistic way” and therefore may “fail to capture the non-linear, indirect and longer-term outcomes of ‘weaker’ forms of risk controls” (p42).

In concluding, they argue that some features of HOC approaches may be relevant for healthcare, such as a brainstorming technique, but the “straightforward application of this model” is said to add little to the value of developing effective risk controls in clinical settings and “lacks validity and usefulness” (p42).

They provide some thoughts for moving forward – including a revised model that categorises risk controls but doesn’t imply a hierarchy of stronger/weaker and instead, the strength of an action depends on the congruence between the local context and the intervention.

They say that assessing congruence should be part of risk control/HOC practices.

Finally, they state that “The complexity of healthcare organizations, and their inherent reliance on human behaviour, interactions and knowledge, indicates that the mere duplication of ideas from other fields—even when they are deemed ‘high reliability’— may turn out to be ineffective, if not harmful” (p42).

And “rigid or unreflective” application of the HOC could “mislead those seeking to improve patient safety” (p42).

Authors:  Liberati, E. G., Peerally, M. F., & Dixon-Woods, M. (2018). International Journal for Quality in Health Care, 30(1), 39-43.

Study link: https://doi.org/10.1093/intqhc/mzx163

Link to the LinkedIn article: https://www.linkedin.com/feed/update/urn:li:ugcPost:6942228764411973632?updateEntityUrn=urn%3Ali%3Afs_updateV2%3A%28urn%3Ali%3AugcPost%3A6942228764411973632%2CFEED_DETAIL%2CEMPTY%2CDEFAULT%2Cfalse%29