In this article, we show that teamwork can contribute to a small but significant decrease in wasted time for patients in the ED, in terms of shorter time to physician and shorter total visit time. This effect of multi-professional teamwork on time to physician and total time in the ED is in line with a recent systematic review of interventions aiming at improving processes and patient flow in the ED. Excessive total time in the ED has been linked to poor outcome, and has been suggested as an important quality indicator. Time to physician, as well, has been described as an indicator of quality and safety in the ED, since seeing a decision-maker is a necessity in order to get a prompt assessment and, subsequently, adequate treatment. Hence, our results indicate that teamwork may have a positive impact on these important quality indicators in the ED.
Furthermore, we show that the 4-hour target was met for a larger group of the patients in the ED at the last follow-up. This confirms the findings reported by Mason, who also showed that the introduction of a time target, in itself, was effective in reducing the proportion of patients staying in the ED longer than four hours. However, the 4-hour target has also been criticized. It is emphasized that lead times in the ED are not fully within the control of the ED. For example, ED processes are reliant on other diagnostic facilities at the hospital such as radiology and laboratory, and not least the availability of hospital beds and the communication between wards and the ED. Hence, in the context of both ED development and improved patient flows, processes and safety, it is necessary to view ED together with and in interaction with the whole hospital. This interdependence between the ED and actors outside the ED may be one reason why the effect of teamwork on lead times and the 4-hour target was limited in this study, and did not reach the goal set by the politicians.
It is important to note that the effect on lead times in this study was only evident at the last follow-up, almost 1.5 years after the introduction of teamwork. One interpretation may be that the chosen intervention was ineffective in achieving teamwork; or that it was effective, but that the resulting teamwork was not sufficiently linked to the outcomes. Another interpretation, which is consistent with other studies of improvement processes, is that change takes time. This is particularly true when it is a multi-faceted change in a complex environment with many different staff members involved. The implementation of change is difficult for several reasons: it is hard to reach staff with information, changes interfere with the hierarchies within hospitals, and achieving behavioural changes is difficult when there is a large and heterogenic group of staff who all need to change their way of working and do not always clearly see the benefits of the change right away[25, 26]. Because of the physicians’ variety in number of on calls, we may also assume that time to adjust to the new work process would take longer than in contexts only dealing with permanent staff members. Also, over a period of time, changes in context often occur that may interfere with the implementation. In this particular case, organizational change and changes in managerial positions took place during the first follow-up (phase C) and were followed by a high turnover rate among RNs. This highlights that things outside the scope of the intervention may affect the interpretation of the results. It also indicates the usefulness of several follow-up periods. It is a challenge to investigate changes in clinical practice, but it is necessary in order to find efficient work procedures that lead to quality and safety improvements for patients and staff.
A wider question is whether the statistical significant effect on lead times also has clinical significance. In terms of patient safety, this is complicated since it is likely that the clinical consequences of long lead times differ between patients, depending on the severity of their medical condition. An endpoint to study would then be to pair lead times with triage levels. However, beyond patient safety issues, waiting times are also highly relevant from a patient satisfaction perspective in that long waiting times negatively influence patient satisfaction[27, 28]. Thus, shortening lead times may nevertheless be important from this perspective.
This study was set in the ED’s section of internal medicine. Therefore, generalization to other sectors and specialties in healthcare may be limited to those parts that involve a large staff, physicians who are on call rather than part of the regular staff, and high dependability on other facilities at the hospital. However, the outcome in terms of lead times may be relevant for emergency care as the time the care process takes not only affects internal efficiency but also clinical outcome and patient safety in EDs.
Since the data are based on a number of two-week periods a number of months apart, it cannot be dismissed that the differences between the control phase and the follow-up phases were related to things other than teamwork. One factor that may contribute to a false effect of an intervention is the fact that when investigating change in clinical practice, the staff is not blinded to the fact that their work is being evaluated. Hence, it cannot be dismissed that staff may have behaved differently because they knew they were under study. However, it is unlikely that this would result in better results at the last follow-up. Another factor could be the variable levels of competence among the physicians, although there was no systematic difference in this respect between data collection periods. Furthermore, the large turnover rate among RNs means that, to some extent, there were different individuals on staff between the follow-up periods. Based on the importance of work experience in working in the ED, it is unlikely that having less experienced staff would result in better lead times. Rather, the fact that a new work routine and shorter lead times were achieved despite high turnover implies that focusing on changing work processes may be a way to implement teamwork.
In our study we used registry data whereby the time is manually registered when a patient physically leaves the ED. However, when a patient is admitted to the hospital there is often a delay from the moment when the admission decision is made and the time point at which there is a hospital bed available. This means that the lead times in this study are inflated. On the other hand, there is no evidence to suggest that there were any systematic differences between the measurement phases in this respect, and therefore the comparison over time is not likely to be affected. For the same reason the data regarding arrival mode had different missing values within the different data collection periods. There could be a possibility that patients arriving by ambulance would have been taken cared of in a shorter time, however, standard triage assessment were conducted and therefore the arrival mode unlikely could affect our main findings. Additionally, the findings from the adjustment for confounders should therefore be considered carefully, as it could be a result of limited data rather than effective adjustment.