Reducing Bottlenecks to Improve the Efficiency of the Lung Cancer Care Delivery Process: A Process Engineering Modeling Approach to Patient-Centered Care

Abstract

The process of lung cancer care from initial lesion detection to treatment is complex, involving multiple steps, each introducing the potential for substantial delays. Identifying the steps with the greatest delays enables a focused effort to improve the timeliness of care-delivery, without sacrificing quality. We retrospectively reviewed clinical events from initial detection, through histologic diagnosis, radiologic and invasive staging, and medical clearance, to surgery for all patients who had an attempted resection of a suspected lung cancer in a community healthcare system. We used a computer process modeling approach to evaluate delays in care delivery, in order to identify potential ‘bottlenecks’ in waiting time, the reduction of which could produce greater care efficiency. We also conducted ‘what-if’ analyses to predict the relative impact of simulated changes in the care delivery process to determine the most efficient pathways to surgery. The waiting time between radiologic lesion detection and diagnostic biopsy, and the waiting time from radiologic staging to surgery were the two most critical bottlenecks impeding efficient care delivery (more than 3 times larger compared to reducing other waiting times). Additionally, instituting surgical consultation prior to cardiac consultation for medical clearance and decreasing the waiting time between CT scans and diagnostic biopsies, were potentially the most impactful measures to reduce care delays before surgery. Rigorous computer simulation modeling, using clinical data, can provide useful information to identify areas for improving the efficiency of care delivery by process engineering, for patients who receive surgery for lung cancer.

Appendix: Calculation Formulas

To evaluate the total process time, the following steps are followed:

• Step 1: We decompose the complex process in Fig. 1 into a set of serial routes, each representing a possible trajectory of a patient. Then for a particular route i, the mean diagnosis time T _i can be calculated as follows:

If each stage j in route i with M steps takes an average waiting time τ _ij , then the mean time T _i of diagnosis route i will be:

$$ {T}_i=\sum \limits_{j=1}^M{\tau}_{ij}. $$

• Step 2: We calculate the product of all probabilities going to the next test after each one in this route, denoted as p _i in Fig. 2. Then we obtain the probability p _i of a patient taking such a route. This will be the weight of route i.

• Step 3: We denote p _i T _i as the weighted mean time. Summing up all the possible routes, we obtain the average mean time of overall diagnosis process.

$$ T=\sum \limits_i{p}_i{T}_i. $$

Using this result, we introduce the so-called bottleneck waiting time, which is the one whose reduction will lead to the largest reduction in overall diagnosis-to-treatment time, comparing with reducing all other waiting times.

Specifically, let T(τ _i -δτ _i ) denote the resulting overall time when waiting time τ _i is decreased by a small percentage δ (e.g., 10%). Then such a time τ _i is the bottleneck waiting time if it is the smallest comparing with all other T(τ _j -δτ _j ), where j is different from i.

$$ T\left({\tau}_i-\updelta {\tau}_i\right)<T\left({\tau}_j-\updelta {\tau}_j\right),\kern2.25em \forall j\ne i. $$

The above method views the diagnosis-to-surgery process as a complex network and assumes that both the trajectory routing probabilities and the waiting times are independent. In practice, the correlations between the probabilities and between the waiting times are difficult to measure since the next steps may not be carried out immediately or could be performed by independent providers. Note that even if the subsequent test is carried out immediately, this implies its waiting time is very short, and its impact on the overall waiting time will be insignificant or very small. Moreover, the scope of this analysis only considers the average performance in total delay time, thus we only need the mean waiting time.