gms | German Medical Science

Introduction: Patients with pulmonary surgery (e.g. partial removal of pulmonary lobes) usually receive a thoracic drainage which also serves to control pulmonary fistulas. For the termination of the thoracic drainage it is crucial to weigh the risks associated with early termination of the drainage (e.g. collapse of the lung due to fistulas) and the risks associated with unnecessary long length of drainage (e.g. infection risks). Therefore, the aim is to terminate the thoracic drainage at the earliest possible time point where it is ensured that the patient is no more exposed to unreasonable risks due to premature termination. It turns out that the standard of care regarding the termination of thoracic drainages varies widely between different clinics resulting in different durations of the thoracic drainages. Therefore, the goal of our considerations is to develop a new standard for the termination of thoracic drainages that takes the trade-off between the above mentioned risks into account and can be employed in the clinical routine.

Methods: We analyzed Flow-data of 112 patients with digital thoracic drainage pumps (Thopaz®) collected in 10-minute intervals over the period of the thoracic drainage, resulting in approximately 90,000 data points.

The general methodical task is to find an objective stopping criterion for a time series measuring treatment induced progress which is so far terminated by subjective clinical decisions. In order to find this criterion a trade-off between potential risks of premature termination (of treatment) and the benefit from earlier termination has to be made. As the prolongation of treatment is typically associated with diverse problems the benefit of earlier termination may be defined as the time difference between the actual termination of the time series by subjective clinical decision and the virtual time point of termination according to the objective criterion. In the framework considered, there is an upper bound for the data points whose transgression is considered potentially harmful for the patient. Thus, the risk associated with premature termination is defined as the mean excess of the bound in the time between the fulfillment of the objective criterion and the actual termination of the time series. To assess the magnitude of the risk associated with premature termination we defined a reference risk, which is the risk during that period of time which may be used for the subjective decision about termination.. In the thoracic drainage setup this time period was chosen to be 24 hours, since the physicians decision whether or not to terminate the drainage is usually based on the data of the previous day. Thus, the risk in this period is assumed to be acceptable for the termination and is therefore used as a reference.

In the application to the thoracic drainage data, we defined several competing criteria for the termination of thoracic drainages and compared them in terms of the risk associated with premature termination. The comparison yields criteria which result in risks which are comparable to the risk the patient is exposed to in a 24 hour period before the actual termination of the drainage. Additionally, we examined the possible influence of other variables on the actual time until termination of the drainage as well as on the virtual time derived from the newly defined criteria using linear regression models.

Results: The new, objective criteria for the termination of thoracic drainages lead to shorter drainage durations without increasing the risk a patient is exposed to. The mean reduction in the duration of the drainage lies, depending on the criterion applied, between 106 and 114 hours, leading to a substantial benefit for the patient. Furthermore, it could be shown that the actual duration of the drainage is driven by factors explaining the state of health of the patient (e.g. smoker status, pre-existing illnesses), the type of surgery and the clinic. The first-order kinetics of the Flow-data did not have a significant influence. On the other hand, when regarding the virtual (defined by the new criteria) termination of the drainage we observe that it is only driven by the first order kinetics of the Flow-data and therefore very objective. Direct influences of other variables could not be shown. However, the other variables had significant influence on the Flow-kinetics.

Discussion: The termination of thoracic drainages seems not only to depend on the medical need but also on the subjective assessment of the responsible persons. The virtual termination of the drainages, which are based on objective criteria do not depend on personal judgment and lead to a substantial reduction of the duration of the drainages without increasing the risk associated with premature termination. It is planned to validate these results by applying the methods derived here to a second data set which is independent from the data used in the present considerations.