Time pressure in diagnosing written clinical cases: an experimental study on time constraints and perceived time pressure

Introduction

In most hospitals and general practices, physicians are allowed a fixed amount of time to see each patient. Several studies into working conditions among health care professionals have reported that physicians often complain about time pressure [1] or express the need for more time to provide patients with good quality of care [2]. Time pressure relates to the amount of time allotted for a task compared to the amount of time one considers necessary for the task. Furthermore, working under time pressure is associated with job dissatisfaction, stress, fatigue and burnout [1], [2], [3], [4], [5].

Unsurprisingly, time pressure can have an impact on a physician’s clinical work as well. A clinical study by Welker and colleagues [6] showed that reducing the ‘time to first antibiotic dose’ for patients with community-acquired pneumonia from 8 h to 4 h, had an effect on diagnostic accuracy. This measure is used as a core quality indicator and is publicly reported as a percentage of patients who are diagnosed and treated within that timeframe. A lower score could impact the hospital’s quality ratings, and thus could potentially create a sense of time pressure to diagnose a potential patient and give them the antibiotic treatment, even in the face of diagnostic uncertainty. The study showed that diagnostic accuracy decreased when the timeframe was shortened, despite the fact that the mean time to first antibiotic dose did not change significantly and was far below the 4 h time limit.

Experimental studies using clinical case vignettes have shown mixed results of the effect of time pressure on diagnostic accuracy. In two studies by AlQahtani [7], 8], resident participants were told, independent of their actual time, that they were either on or behind schedule with diagnosing their cases. Results showed diagnostic accuracy was negatively influenced in these group that thought they were behind schedule. In two studies, time pressure was implemented by allotting resident participants a fixed amount of time to diagnose 20–25 cases [9], 10]. Not all cases needed to be completed. Results showed that residents who were asked to diagnose 25 cases in 30 min had similar accuracy rates compared to residents that were asked to diagnose 20 cases in 40 min [10]. Furthermore, residents were more accurate when they had a lower time-to-diagnose [9].

In psychology, the effects of a time limit on performance have been studied using the Iowa Gambling Task, a psychological decision-making task [11]. In this task, participants choose cards from card decks that are either ‘good’ or ‘bad’, determined by their in-game money reward and/or punishment. The goal is to learn the rules of the decks and obtain as much in-game money as possible. Results showed that participants with a time limit had an impaired performance compared to participants without a time limit. In a follow-up study [12], the effect of the perception of time pressure was measured by manipulating the instructions for picking a card. Participants were instructed to select a card within 2 s and half of the participants were told that this provided time was insufficient. Results showed that participants who were informed that the time allotted was insufficient performed worse than those who were informed that they had sufficient time, despite the actual time allotted being the same.

While both time limits and instructions to induce subjective time pressure have been shown to affect diagnostic accuracy, the effects are not consistent and the combination of these factors has not been studied together. The current study aims to fill this gap by investigating the effects of perceived time pressure (sufficient vs. insufficient) in combination with a restricted or lenient time limit with regard to diagnostic accuracy, number of differential diagnoses, and confidence. We hypothesized that diagnostic accuracy, number of differential diagnoses, and confidence are lower when participants think they have insufficient time, regardless of actual time constraint. Furthermore, we hypothesized that the manipulations enhance each other when they ‘match’ (i.e. diagnostic accuracy, number of differential diagnoses, and confidence are highest in cases where participants think they have enough time and the time limit is lenient, and lowest in cases where participants think they do not have enough time and the time limit is restricted).

Materials and methods

Procedure

This study was conducted in accordance to the Declaration of Helsinki (as revised in 2013) and approved by the Medical Ethics Board of the Erasmus Medical Centre.

Before participating, participants were asked to read the information letter and sign an informed consent form. They were not fully informed of the aim of the study beforehand, but were told that the goal of the study was to measure diagnostic performance under time pressure.

The experiment was divided in two blocks (see Figure 1) which were shown in random order. After reading the instruction, participants would diagnose six cases: two matching filler cases, two matching experimental cases, and two mismatching experimental cases, in that order. We chose to have the first four cases always be ‘matching’ (i.e. condition A or D) so participants would not grow suspicious of the instruction. The participant either clicked forward to the next page themselves, or were forwarded automatically to the next page by the timer. They were then asked for their most likely diagnosis and for other diagnoses they had considered. Participants were not restricted in their time to enter the diagnosis, although time spend on the page was recorded.

Figure 1:

Order of the cases within the two blocks.

The cases within the blocks were counterbalanced in eight versions to prevent order effects. Participants were randomly assigned to one of the versions. After participants finished both blocks, they were asked how confident they were of their diagnosis for each case on a scale of 0–100. At the end of each block and at the end of the survey, participants were asked several questions about the time pressure they felt (see Supplementary Material, Appendix 2). After completing the survey, they were given the link to the correct answers of the cases.

Results

Forty-two residents participated in this study (see Table 2). They were mostly female (54.8 %), internal medicine residents (90.5 %) at University of Pittsburgh Medical School (85.7 %). Most residents were in Year 1 (38.1 %) or Year 3 (28.6 %) of their residency.

Discussion

This study investigated the effects of perceived time pressure and actual time constraints on diagnostic accuracy, number of differential diagnoses, and confidence in an online experiment with written clinical cases. In summary, the current study shows an effect of a restricted time limit on diagnostic accuracy. Although participants reported feeling more time pressure in the perceived-insufficient conditions, no effect of perceived time pressure was found and no interaction effect was present. This means that, contrary to the expectations and previous research [7], 8], the perception of time pressure did not influence diagnostic accuracy. In addition to the most likely diagnosis, participants were asked to name other diagnoses that they considered. It was hypothesized that the number of differential diagnoses that were considered would be higher when participants perceived to have enough time. However, the number of submitted differential diagnoses did not differ between conditions. Lastly, participants were asked about their confidence in the most likely diagnosis on a scale of 0–100. Results show that participants were more confident both when they had a lenient time limit and also when they received the instruction that they would have sufficient time.

It seems that most of the effect on diagnostic accuracy in this study stems from the actual time constraint. This means that participants performed worse in cases were the time limit was restricted time compared to the lenient time limit cases. Participants were not aware of the exact time limit for each case, but they were more likely to reach the end of the timer while reading the case, and so were more likely to be automatically forwarded to the next page to fill in their most likely diagnosis and differential diagnosis, possibly disrupting their thought process. In addition, they were possibly aware of the passing of time due to the time bar that was present at the top of the page, which logically moved faster when there was a shorter time limit.

These results for the effect of the time limit on diagnostic accuracy seem to be different from the study by Norman and colleagues [10], in which residents had a time limit for a total of 20–25 cases and participants who were allotted less time did not have a lower diagnostic accuracy than those who were allotted more time. However, in their experiment they chose a larger range of case difficulty that included easier cases where the participants were likely able to use pattern-recognition to diagnose them correctly. The current study used cases that were developed to be challenging. This means that participants might have needed more time than they were allotted, because they could not rely on pattern-recognition. The restricted time limit, visible through the live time bar, therefore could have resulted in lower accuracies.

It is somewhat surprising that no main effect of perceived time pressure on diagnostic accuracy was found, as this effect has previously been established [7], 8]. This discrepancy could in part be explained by a difference in method between the current study and the studies by AlQahtani and colleagues [7], 8]. In their studies, participants were shown a message after diagnosing each case telling the participant that they were behind schedule and they needed to hurry. The current study relied on the message introducing each block of six cases. It is possible that this message did not have enough effect, or was forgotten after a couple of cases, diminishing the effect of the perceived-insufficient condition.

Interestingly, results showed an effect of both perceived time pressure and time constraint on confidence, indicating that participants felt more confident about their most likely diagnosis when they thought they would have enough time or when the time limit was more lenient. The reasons for this could be that the lenient time limit and low perceived time pressure allowed for double-checking of the answers.

Additionally, there are some indications that the time bar has inadvertently added a layer of time pressure to all the conditions. When participants were asked if they thought the time bar contributed to their feeling of time pressure, most of them answered ‘yes’. They also described that they were feeling rushed by the time bar moving and changing color from green to red, and that it was distracting them from reading the case. Because the time bar was present during all conditions, it is possible that there was some form of perceived time pressure in all conditions induced by the time bar.

Conclusions

In conclusion, this study we investigated the difference between perceived time pressure via instruction and time constraint via a lenient or restricted time limit. Results showed that a restricted time limit has an effect on diagnostic accuracy. Although participants reported feeling more time pressure when they received the perceived-insufficient time instruction and felt less confident on cases in these conditions, no effects of the instruction of perceived time pressure were found for diagnostic accuracy. More research is needed to investigate the robustness of the effects found in this study. In addition, different forms of (perceived) time pressure should be investigated in various settings to gain a better understanding of the effects on diagnostic accuracy.