How doctors diagnose diseases and prescribe treatments: an fMRI study of diagnostic salience

Understanding the brain mechanisms involved in diagnostic reasoning may contribute to the development of methods that reduce errors in medical practice. In this study we identified similar brain systems for diagnosing diseases, prescribing treatments, and naming animals and objects using written information as stimuli. Employing time resolved modeling of blood oxygen level dependent (BOLD) responses enabled time resolved (400 milliseconds epochs) analyses. With this approach it was possible to study neural processes during successive stages of decision making. Our results showed that highly diagnostic information, reducing uncertainty about the diagnosis, decreased monitoring activity in the frontoparietal attentional network and may contribute to premature diagnostic closure, an important cause of diagnostic errors. We observed an unexpected and remarkable switch of BOLD activity within a right lateralized set of brain regions related to awareness and auditory monitoring at the point of responding. We propose that this neurophysiological response is the neural substrate of awareness of one’s own (verbal) response. Our results highlight the intimate relation between attentional mechanisms, uncertainty, and decision making and may assist the advance of approaches to prevent premature diagnostic closure.


Participants
The recruitment criteria of participants were: 1-working part time or full time in primary care -general internal medicine or family medicine -and completed medical residence; 2-having Portuguese as first language; 3-right handedness, assessed with a modified version of the Edinburgh Handedness Inventory ; 4-absence of neurological and/or psychiatric disorders in the present and not in use of psychopharmacologic or neurological medication.
Most physicians were affiliated with the medical facilities of the Faculty of

Medicine of the University of São Paulo and the Albert Einstein Israelite
Hospital.
Four participants had their data excluded for the following reasons: 1-intense headache during the data collection; 2-distortions in fMRI images provoked by dental amalgam; 3-complaint of intense jitteriness during data collection; 4-intense somnolence with several micro naps during data collection.

Experiment 1
Altogether, 79 sequences with the diagnosis of different diseases as a target, 36 sequences for naming different animals, and 40 sequences for naming different objects were assessed in the pilot tests. The final selection is detailed in item 4.
The organization of the sets obeyed the following criteria: 1-Order of the sequences, including sequences in the training, was pseudorandomized in such a way to avoid more than two events of the same type occurring in succession; 2-The presence of a high or low diagnosticity stimuli in the beginning of the sequence was counterbalanced so that half of the sequences began with one of the two types of stimuli; 3-The order of the sequences in relation to the diagnosticity of the first stimulus was also pseudorandomized avoiding more than three sequences with the same type of first stimulus; 4-Two versions were created for each of the seven sets of sequences, with the inversion of the initial and the end stimuli; one set beginning with a low diagnosticity and the other with a high diagnosticity stimulus, totaling 14 sets.

Experiment 2
Two lists of stimuli were created and assessed in the pilot tests: 1-112 diagnostic information selected from the sequences used in Exp.1; 2-78 names of diseases. The final selection is detailed in item 5.
The order of the stimuli was pseudorandomized; with no more than three tasks of the same type in succession.

Training
The training order of the two experiments followed the order of the experiments in the scanner. Participants received written instructions for the tasks. They were trained to minimize movements while vocalizing the responses.
The training sequences of 16 events and eight null events for each experiment were presented twice to the participants in a notebook computer using E-Prime 2.0. In the first run, the presentation pace was controlled by the researcher for an initial familarization. In the second run, the pace followed the temporal structure of the experiment.

Assessment and management of performance anxiety
We used an assessment scale presented verbally in which participants were asked to rate their grade of tension: zero represented 'complete relaxation' and 10 'high tension'. The use of the term anxiety in this assessment was purposefully avoided. The assessment was carried out during the training, before the beginning of the data collection inside the scanner and in the intervals between fMRI sessions.
After reporting tension greater than 3, 18 participants (58,2%) were oriented to do abdominal breathing. For 11 (35.5%), the relaxation was conducted before data collection. For seven (22.7%), it was necessary to realize the relaxation between, at most, two functional sessions. All 18 participants presented reduction of tension to ≤3 after brief periods of relaxation exercises supervised by the researcher in charge of the data collection.

Data collection
The stimuli in size 50 Arial font were retro-projected using a Sanyo ProtraX Multiverse projector (Sanyo Electric Co.) onto a translucent screen and viewed through a mirror in the head coil. The resulting visual angles ranged from 6.0º to 26.0º in width and from 1.4º to 4.9º in height, depending on the number and extend of the words in each stimulus.

3-Supplementary Tables
Supplementary Table S1-Types of responses (in percentage)* * There were no significant interactions between tasks, diagnosis and naming, and diagnosticity of the first stimulus in each of the response types discriminated. For this reason the results are reported for the main effect of task type.

4-Set of stimuli for Experiment 1
To help the visualization of the task, stimuli are placed in the order in which they were presented to participants using one of the sets employed in the experimental protocol, including training stimuli. In the first column of the

5-Set of stimuli for Experiment 2
To help the visualization of the task, stimuli are placed in the order in which they were presented to participants using one of the sets employed in the experimental protocol, including training stimuli. In the first column of the