According to our qualitative study, Finnish physicians and nurses found the CDSS and CMR useful in primary care settings if diagnoses were documented and use of medication was updated in the EHR. Based on this study the use of the CDSS is facilitated by beneficial reminders, safety checks and summaries that users find useful in clinical situations. To enable a well-functioning system, medication and diagnoses must be updated by the users. There should be common practices within organizations for documenting structured data. CDSS software should facilitate ease of documentation and produce focused and graded information at the point of care for the end-user.
In our study, the health professionals, who recently started using a CDSS system, found the lack of diagnosis codes to be the most significant barrier to using the system. Similarly, the fact that medication lists were outdated in the EHR prevented users from fully benefitting from the CDSS. This made the system incomplete and less trustworthy. Apparently, these observations turned up to be ‘control beliefs’ working against willingness to adopt the system. According to the Theory of Planned Behavior [TPB] [25, 26], such beliefs could limit the adoption of a system. In further testing of the CDSS, more effort must be made to make the system more complete with adequate with drug information data to reduce such control beliefs.
Another link to the Theory of Planned Behavior was that apparently the CDSS users were more confident of the system when it linked information to action. According to the TPB [25, 26], this is an example of a behavioural belief, which produces a favourable attitude toward the behaviour: starting to apply decision support more often in the own clinical practice.
A significant reason for the missing information related to their organization’s practices and lack of leadership regarding this. There were no common instructions to document permanent diagnoses in the study health center. Moreover, it appeared that in some units the diagnoses were documented in the free text, preventing their use in decision support.
Busy clinicians felt that updating the diagnoses in the EHR was complicated and time-consuming. This sets a challenge for the CDSS system providers since the CDSS system should be designed to support this transition.
In order to fully benefit from the potential of the CDSS and CMR systems, the healthcare documentation practices and organizational setting should be updated to exploit the utilization of the recorded patient information instead of simply recording the information. Patient information is required in an accurate structured format in the EHR system. One conclusion of our study is the need for specific management for the patient record information practices.
In our focus groups, it appeared that the lack of correct information undermined trust in the CDSS. The missing diagnoses were considered a problem. It seems that the change in practices needs support from management, e.g. instructions, as well as license for the nurses to copy the diagnoses from hospital information records to the health center EHR.
On the other hand, if diagnoses were documented and medication was up to date, the CDSS and CMR were found to be useful for both physicians and nurses in clinical work in a primary care setting. The CDSS and CMR systems have the potential to function as a tool that would enhance the culture for inter-professional collaboration in primary and home care. Both nurses and physicians found that nurses could use this as a screening tool for important medical issues. This would optimally lead to supervised medication treatment decisions by the nurses in conventional situations, while providing the physician with CDSS assistance in more complex medication situations.
Limitations and strengths
A limitation of the current study is that participants were recruited from a single, albeit large health center; they used a single CDSS and their experience as decision support users was as yet brief. The physicians and the nurses had voluntarily participated in the focus groups. It is therefore likely that they are more interested and are more reflective about the CDSS than the average clinicians. A further important limitation is the well recognised problem that individuals do not always do what they say they do.
Two of the four researchers were CDSS developers. In the qualitative study, with subjective approach, this could have caused bias.
On the other hand a strength of the study is that it was aimed successfully at obtaining new information on the user perspective regarding the use of a CDSS system with CMR. Nurses were included in addition to physicians in the study group, facilitating a broader understanding of the different end-users of the system as well as on the system as a tool in an inter-professional collaboration in health care. In the recent review Piscotty and Kalisch pointed out that little is known how, when, and why nurses use CDSS [19]. Physician interviewees were both young and experienced [Table 2]. We also interviewed nurses and physicians in separate focus groups in order to avoid the dominance of either group in the discussion.
The data reflect the perspectives and experiences of the participants. A qualitative study in this context is important for generating hypotheses. The generalizability of the findings is limited, but this is never the intention in qualitative studies, the main aim of which is to contribute to increased understanding.
Comparison with existing literature
Clinicians often fail to adopt CDSS recommendations [13, 14]. The results of this study emphasize some possible reasons. In the previous study of Patterson et al., the primary reason for not paying attention to clinical reminders was extensive workload. Physicians reported using clinical reminders only when they had additional time [17]. In our study clinicians also mentioned lack of time as a significant reason for not updating the diagnoses and medication. In a systematic review, users reported clinical situations in which inappropriate reminders are annoying [14]. The high frequency of reminders has been perceived as irritating in the context of a consultation and as a consequence, users have felt they may become desensitized to alerts [14, 21]. Further, the simplicity and visibility of messages were considered as key drivers of use. There have been suggestions that alerts should be graded by severity [14]. Our study underlines the same observation. Too much non-graded information on the screen could be a barrier to the beneficial use of CDSS. In addition our study highlights that reminders were perceived as inapplicable due to missing diagnostic codes or outdated medication. False alarms create a lack of trust in the CDSS and according to TPB this may lead to behavior which is a function of attitudes towards the behavior in question [25, 26]