ERROR MANAGEMENT

Error management is an ancient practice. Most hazardous endeavours developed tools for minimising the likelihood of error. Victorian railway signalmen, for example, inserted flagsticks into the spring catches of signal levers on line sections that were blocked by a stationary train to help them remember that a train is waiting to proceed and to avoid pulling off the signal prematurely. Pilots routinely use checklists prior to take off, descent, and landing to minimise the possibility of omitting important procedural steps. Scrub nurses count swabs and instruments before and after surgery to help ensure that none has been left in the patient.

Although of proven value, these evolved techniques tend to be piecemeal rather than planned, ad hoc rather than theoretically driven, and fail to take account of the developments that have occurred in understanding the nature, varieties, and affordances of human error. As a result, they focus upon the personal rather than the systemic causal factors; they rely heavily on exhortations and disciplinary sanctions; and they often fail to distinguish between the random and systematic error shaping factors. Perhaps the most serious barrier to progress is the essentially moral belief that responsible professionals should not make errors. There are two corollaries to such a view. Firstly, the errors of professionals will be rare but sufficient to cause adverse events and, secondly, errors with bad consequences must be negligent or even reckless and so deserve deterrent sanctions. The reality, however, is that responsible and highly trained professionals, being human, make frequent errors,^{3, 4} most of which are either detected and recovered or are simply inconsequential. It is also the case that such errors are only occasionally necessary to add the final ingredients to event scenarios that have often been simmering for lengthy periods within the system.

Error management has two components: (1) error reduction (measures designed to limit the occurrence of errors) and (2) error containment (measures designed to enhance the detection and recovery of errors, as well as seeking to minimise their adverse consequences). A broad spectrum of error management techniques has been discussed at length elsewhere⁵; our present concern is with the use of task analysis and reminders to combat dangerous omissions.

OMISSIONS AND THEIR AFFORDANCES

Omissions make a particularly worthwhile target since the failure to carry out necessary steps in the performance of a task is probably the single most common human error.⁶ Affordances in this context are situational factors—as distinct from personal factors—that promote the likelihood of an omission error. The predominance of omissions arises in large part from the variety of mental processes that are implicated in their occurrence. Action control involves at least four main stages—planning, intention storage, execution, and monitoring—and a glitch in any one or more of these processes can lead to an omission. These possible failures are shown in more detail in table 1. It is often very difficult to pinpoint the precise cognitive processes that were involved in omitting a crucial task step—even the error makers themselves find it difficult to identify the cause of the failure.

A more promising way forward is to focus not so much upon the underlying mental processes but upon those task characteristics most likely to afford omissions (termed “affordances”). Recent psychological research has identified a number of task properties that are likely to increase the probability that a particular step will be omitted. Some of the more important of these features are as follows:

• The greater the informational loading of a particular task step—that is, the higher the demands imposed upon short term memory—the more likely it is that items within that step will be omitted.⁷

• Procedural steps that are functionally isolated—that is, ones that are not obviously cued by preceding actions nor follow in a direct linear succession from them—are more likely to be left out.

• Recursive or repeated procedural steps are particularly prone to omission. In the case where two similar steps are required to achieve a particular goal, it is the second of these two steps that is most likely to be neglected.⁸

• Necessary steps that follow the achievement of the main goal of a task are likely to be omitted. This is an instance of a general principle: steps located near the end of a task sequence are more prone to omission. Such “premature exits” are due in part to the actor's preoccupation with the next task, particularly when the current activity involves largely routine actions.²

• Steps in which the item to be acted upon is concealed or lacking in conspicuity are liable to omission.

• Steps following unexpected interruptions are especially prone to omission. This can occur because the person loses her place in the action sequence and believes herself to be further along than she actually is, or because some unrelated action is unconsciously “counted in” as part of the task sequence.²

• Tasks that involve planned departures from standard operating procedures or from habitual action sequences are liable to strong habit intrusions in which the currently intended actions are supplanted by a more frequently used routine in that context, and thus omitted.

• Actions that are triggered by weak, noisy or ambiguous signals are likely to be omitted.

A number of these omission provoking properties can combine in a single task step. When this occurs, the effects are additive and the result is a recurrent error trap that predictably snares a large number of people. The everyday task of using a simple desk photocopier of the kind shown in fig 1 will both support this claim and illustrate the basic error management principles.

A STUDY OF PHOTOCOPYING ERRORS

Using a simple desk photocopier provides a convenient test bed for investigating the combinatorial potency of the omission provoking task features outlined above. It is commonplace, relatively stress free (at least while the machine is working), and is associated with a variety of possible errors.

A questionnaire was devised that asked 95 undergraduates, researchers, and academics how often they committed 15 distinct error types (table 2) while using such a photocopier.⁶ The respondents were required to rate the frequency with which they committed each error type on a 7 point scale ranging from 0 = never to 6 = nearly all the time. The main prediction was that failure to remove the last page of the original would be the most frequent omission error because this step possesses the largest number of omission affording characteristics, as listed below:

• The emergence of the last copy page from the machine gives a strong but false completion signal. The main goal of the activity (copying) is achieved before all the necessary steps are complete.

• This false completion signal gains extra power from its closeness to the presumed end of the activity. As the end of this tedious task approaches, attention consuming preoccupation with the subsequent task increases.

• The emergence of the last sheet also indicates that it is no longer necessary to put in another original. This leaves the removal of the last sheet as a functionally isolated act. Throughout the copying process the removal of the preceding sheet has been cued by the need to place the next original sheet on the platen. To do this, it is first necessary to remove the preceding original. In the case of the last sheet this cueing is absent.

• The closed lid conceals the last sheet of the original, so no visible reminder of the need to remove is available.

Table 2 shows the mean ratings for the 15 error types set out in rank order of frequency. The omission errors are marked with an asterisk. As predicted, leaving the last page of the original was the most common omission. The main purpose of this exercise was to demonstrate that error traps likely to provoke frequent omissions can be established in advance by first breaking the task down into steps (task analysis), and then assessing the degree to which each step possesses error affording properties.

In contrast to higher order errors involving wrong diagnoses and bad decisions, the countermeasures for minimising the occurrence of omissions are fairly obvious. The provision of timely and suitably located reminders will make a substantial dent in the number of future omissions. A more effective solution would be to insert forcing functions—these are mechanical or electronic devices that block onward action until all of the prior steps have been satisfactorily completed. The problem with forcing functions, however, is that they usually involve the redesign of equipment and are thus likely to be both expensive and remote solutions. Reminders, on the other hand, are cheap and quick to apply. But what makes a good reminder?

THE FEATURES OF A GOOD REMINDER

The criteria for a good reminder were derived partly from the recent memory aid literature^{9, 10} and partly from a consideration of the factors promoting omissions (listed above). They fall into two groups:

• Universal criteria: characteristics that should apply to all reminders regardless of their form (see table 3).

• Secondary criteria: characteristics that could be useful in particular instances although are not applicable in all situations (see table 4).

A SURVEY OF EVERYDAY MEMORY AIDS

A first step in assessing the validity of these criteria was to survey the use of everyday memory aids.⁶ One hundred and forty seven psychology undergraduates (128 women, 19 men) were asked to list the strategies they used to help them remember to carry out necessary tasks or activities. The strategies so obtained were grouped into 12 main categories listed below in order of popularity (numbers in parentheses indicate the percentage of subjects citing their use):

• Notes and post-its (65.1%): pieces of paper (with or without adhesive) upon which one or two actions are jotted down and either carried around or else attached to walls, doors, or pin boards in a readily visible location.

• Diaries (57.3%): including filofaxes, notebooks, or any single volume that could be carried around and in which items to be remembered are written down in some relatively organised fashion.

• Lists (55.5%): pieces of paper on which items to be remembered for that day are written down and checked off when done. These lists are either carried around or displayed in some prominent position.

• Writing on hand (43.8%): an item to be remembered is written on the hand with a ballpoint pen.

• Object positioning (41.8%): this entails locating objects requiring action in a prominent position within the person's living space. Sometimes they are deliberately placed where they might block an exit unless they are moved (a forcing function).

• Getting others to remind them (34.9%): friends or relatives are asked to remind the individual to do something.

• Calendars and timetables (31.5%): these are markings upon displayed charts organised by time.

• Mental checking (8.2%): this entails a routine that takes place either before sleeping or on waking in which the person runs through a mental list of the day's tasks.

• Mental rehearsal (6.8%): this involves the individual saying over and over to herself the action that has to be done. This was an exclusively female strategy in this sample.

• Forming associations (6.2%): this approximates to the method of loci in which the person links items to be remembered to images of familiar places or objects.

• Visualising (4.1%): the person simply visualises the performance of some act to be remembered.

• Clocks and watch alarms (3.4%): this involves setting an alarm clock or watch alarm to ring at the time when some action item has to be performed.

Although women far outnumbered men in this sample, there were a number of significant sex differences. On average, men listed only 3.4 strategies compared with 4.3 strategies by women (p<0.01). Women cited the use of diaries, lists, and mental rehearsal significantly more often than men (p<0.01). There were no significant sex differences in either self-rated memory ability or in the mean rated effectiveness of their employment of these various strategies.

HOW WELL DID THE STRATEGIES SATISFY THE “GOOD REMINDER” CRITERIA?

In a further study⁶ 10 academic psychologists were presented with a list of the 12 main memory aid strategies and asked to rate how well each one satisfied the five universal criteria for a good reminder (table 4). They were required to use one of three ratings for each judgement: 0 = the strategy does not satisfy the criterion at all (or hardly at all); 1 = the strategy partially satisfies the criterion, at least under some circumstances; 2 = the strategy satisfies the criterion completely, or nearly completely. The results are summarised in table 5.

The Spearman's rank order correlation coefficient between percentage usage and mean criterion rating was 0.95 (p<0.01). This close correspondence confers a convincing degree of ecological validity upon the five universal criteria. Had criteria satisfaction not correlated highly with the relative popularity of strategy usage, there would have been little or no grounds for claiming that these characteristics were necessary features of good reminders. It seems reasonable to assume that people would not persist with these strategies if they failed to reduce the likelihood of everyday omissions. It must be noted, however, that undergraduates report relatively higher rates of omission slips and lapses than other groups.² We must conclude from this that the use of memory aid strategies does not guarantee the performance of all intended or necessary actions. However, we can only guess at how much worse they would be if they used no strategies at all.

SUMMARISING THE MAIN STEPS IN MANAGING OMISSIONS

We began by identifying omissions as a suitable case for treatment. Not only can they arise from failures at many cognitive stages, they have also been shown to constitute the largest class of human performance problems in various hazardous operations, particularly aviation and nuclear power generation.^{11, 12} They are especially evident in maintenance related activities—hands on, complex, time pressured tasks that have much in common with a wide range of healthcare procedures.

While it is not always possible to identify which mental process failed in omitting a necessary step from a task, we can predict with some confidence which task elements are most likely to provoke such omissions. It has also been shown—using the photocopier example—that the likelihood of an omission is related to the number of omission provoking features associated with a particular task step. The more features present, the more probable the omission. How can we apply these observations in a healthcare context? There are three distinct stages: (1) task analysis, (2) assessing omission likelihood, and (3) choosing and attaching a reminder.

Task analysis

In order to identify these steps in advance it is necessary to carry out a task analysis—a process that decomposes an activity or procedure into a meaningful number of discrete steps. This is not particularly difficult but it can be labour intensive (an example of a hierarchical task analysis for the simple photocopier is shown in box 1). It is therefore necessary to be selective in choosing the procedures for omission management. The most obvious basis for selection is safety criticality. Would the omission of particular steps in the task have an injurious effect upon the patient? Should the consequences of such omissions be especially dangerous, this alone would warrant the use of suitable reminders—that is, skipping to stage 3 directly. If this were the case, then it is possible that useful data exist in the system's incident recording system. It may also be the case that the task already has a written stepwise protocol that would remove the need for task analysis.

Assessing omission likelihood

Because omission provoking features are not always intuitively obvious, it is necessary to review each task step for its omission affordances. A useful tool in this regard is a grid on A3 paper. Each task step could be briefly described on a row, and each of the eight omission provoking features (listed earlier) could occupy a column. The analyst could then mark the columns in which an omission provoking characteristic exists for each task step. At the end of this exercise it is easy to identify the most omission prone steps by summing the ticked features across the columns. Any step that possesses two or more features is a candidate for a reminder, although such judgements should also take into account both the safety criticality of the step and the ease or difficulty with which its omission could be detected before the task is complete. Again it pays to be selective since a great many reminders dotted around the task space are likely to be counterproductive.

Choosing and attaching a reminder

Experience in other domains indicates that those intimately engaged in the task are often the best people to design and apply a reminder. The study described above suggests that post-its or tie-on labels possess most of the characteristics of a good reminder. It must be recognised that all reminders have limits to their utility. Sooner or later they are likely to merge into the background. Reminders need to be renewed regularly and should be designed from the outset with that in mind.

CONCLUSIONS

The omission management programme described here addresses both aspects of error management—reduction and containment. The appropriate and sparing use of good reminders will achieve some reduction of safety critical errors. Understanding the omission affording features of a task will enhance the likelihood of error detection even when omissions still occur, since an expected slip is more likely to be spotted and corrected. Neither process, however, will be wholly successful. There is no single “magic bullet” solution for omissions. An essential prerequisite for effective safety management is the expectation that errors will always occur. Chronic unease along with continuous vigilance and adjustment are still the main weapons in the error management armoury.

Finally, think kaizen. Kaizen is a Japanese word meaning small continuing improvements achieved through a process of close monitoring and refinement. Error management is not like some electronic gadget that can be plugged in, switched on, and left to run on its own. Like religion—in which there are many prayers but few miracles—the process of managing error is as important as the product. Error will never be eliminated, but we can hope to improve the conditions under which people work so as to eliminate the more dangerous affordances for error and to increase their chances of detecting and recovering those errors that will inevitably still occur.