Decision Support in Psychiatry - a comparison between the diagnostic outcomes using a computerized decision support system versus manual diagnosis
Decision Support in Psychiatry – a comparison between the diagnostic outcomes using a computerized decision support system versus manual diagnosis.
Bergman LG, Fors UG.
New methods are always being sought to improve the diagnostic procedures in medicine. Studies have shown that Clinical Decision Support Systems (CDSS) can be beneficial to medical specialties, in which physical examination and laboratory results are used. The question is whether psychiatry, which relies heavily on the patient interview process, can also benefit from computerized decision support. As far back as 1968, the DIAGNO computer program was used for psychiatric diagnosis. Other systems have been developed over the years with variable successes. A new promising system CB-SCID1, is a computerized decision support system that is “designed to support and facilitate axis 1 DSM-IV diagnosis in psychiatry.”
The objective of this study was to investigate the psychiatric decision support system CB-SCID1, establish its level of success, and its ability to save time. For this purpose it was compared to the traditional “paper and pencil” method SCID1.
The most important variables of the study were time, correct diagnosis and possible incorrect diagnoses. CB-SCID1 automatically recorded the times and a chronometer was used for the “paper and pencil” method. Two real cases were used from the DSM-IV Case Book. Specifically, one was a complex case with three diagnoses, and the other was simple with one diagnosis. The participants of the study were 63 volunteer clinicians, who were given a description of the cases in text format, and a four-hour instruction on the use of CB-SCID1. A randomization scheme was used on the two conditions (computerized vs. “paper and pencil”) and the degree of difficulty of the case. A variety of statistical techniques were applied to the distribution of the cases.
- For the easy case (one diagnosis), no significant differences were found for total time and diagnosis between the computerized and “paper and pencil” methods.
- For the difficult case, the total time of the “paper and pencil” process was significantly shorter.
- For the difficult case, the correct diagnosis “Depression” was “significantly more often found” by the “paper and pencil” process.
- For the difficult case, the correct diagnoses “Alcohol” and “Somatization” were more easily found with the computerized method, but not significantly.
In the follow-up interview, 40 out of the 63 participants found CB-SCID1 supportive and easy to use. However, it was slower than the “paper and pencil” process and did not give as many correct diagnoses.
The CB-SCID1 system is not fully automatic and requires a thinking process. This may explain the longer processing time for the difficult case. Its missing of the “Depression” diagnosis may be due to “automation bias, errors of omission, or merely applying sequential but no global thinking.” Other limitations may be the fact that real live cases were not used, the clinicians were not familiar with the CB-SCID1 system, and the software may have design flaws or other errors.
Although there were some limitations of the study, the main of which is the participants’ lack of knowledge of the CDSS, the general conclusion is that there is no major difference in diagnostic outcome between the “paper and pencil “ method and the decision support system CB-SCID1. Whenever a difference was detected, it was in favor of the “paper and pencil method.”
Submitted by Nicolas Thireos