Virtually every health system has launched various process excellence initiatives over the years. The exact methods range from a series of Kaizen events, Value Stream Mapping, or a top-down Six Sigma or Lean Initiative. The transformation efforts could have been guided by internal process excellence teams or external consultants.
These initiatives often succeed in demonstrating a tangible impact. For example, a distinct improvement in the accuracy and presence of orders coming into the Infusion center from providers, or a noticeable reduction in the waste level at the pharmacy. However, senior leaders often grapple with two uncomfortable truths regarding many of these improvement initiatives which are:
- The point improvement in a specific activity did not translate into a meaningful improvement in the overall end-to-end performance objective (e.g. the 30% improvement in order accuracy resulted in only a 3% improvement in the wait times experienced by patients).
- The results are often short-lived. Repeating the measurement 9-12 months later invariably shows that everything has gone back to the way it used to work and the improvement that was celebrated in the conference room is no longer being experienced by the front line.
Why does this happen?
We believe that the initial framing of the improvement initiative is often flawed since it does not establish the correct end-to-end metric as the key objective to be improved. Any end-to-end service process can be characterized by three (and only three) overarching metrics: cost, quality, and time.
Cost: The cost of any service process is driven by labor which usually represents 70-75% of the overall cost. Note that the “units of service” could be anything—a patient, an order for a drug to be mixed, a room to be cleaned.
Quality: The easiest way to think about quality in a service process is the error rate expressed as a percentage. Hence, the percent of incomplete orders or the percent of orders with an error in them are indicators of quality.
Time: This is intended to represent the end-to-end time experienced by the unit of service (arrival to departure for a patient or the interval between placing an order and receiving the drugs in hand). Working on a subset of the process (e.g. arrival to check-in) will not result in improving the overall performance.
Most service improvement initiatives start with either the cost or the quality metric—this is a flawed approach. Cost is predominantly driven by labor, so improving it requires a reduction in the overtime level or the headcount providing that service. As a result, skills and/or capacity are depleted, resulting in a degradation of quality and/or an extension of the end-to-end time taken to complete the process. Hence, the cost improvement that was achieved will likely result in a degradation of the other two metrics. Similarly, quality improvements cannot be mandated without a fundamental redesign of the process. Adding layers of inspection, error detection, and correction will result in additional headcount and extend the timeframe of the service process. Therefore, an improvement in quality also resulted in the degradation of the other two metrics.
However, focusing on the end-to-end process time forces a simultaneous improvement in the other two metrics. A bank that improves turnaround of a loan application from 7-days to 24-hours cannot have a dozen people reviewing each application or being forced to repeatedly fix errors. Hence, the design of the overall service process has to minimize the labor content and to prevent errors from entering into the process in order to achieve the desired improvement in the end-to-end cycle time.
In order to understand the power of time as a metric, it is important to understand the portion of the end-to-end time that is truly value-adding versus the portion of time that is simply waiting. As an example, a patient may spend an hour at a clinic going through the various motions of checking in, filling out forms, getting placed in a room, and so on. The true value-added time may have only been 6 of those minutes (4 minutes with the physician and 2 minutes with the nurse) for a value-added ratio of 10%.
Time in a service process is analogous to inventory in a manufacturing process. With an overstocked inventory, the manufacturing process can be slow and error-prone but can be hidden from the customer since their orders are being fulfilled from the inventory on hand. Similarly, if a service process takes a long time to complete, the errors and delays along the way can easily be masked. Hence, focusing on lifting the value-added percentage of a service process from 10% to 40% or 50% can result in a dramatic improvement in performance.
This requires a shift in mindset – instead of focusing on making the value-adding steps go faster (e.g., trying to get the physician to get the work done in 3 minutes instead of 4 minutes), focus on eliminating or reducing the non-value-added time (i.e., the 54 minutes the patient is spending in the clinic that are not value-added). To do so requires an understanding of “Takt time”—the natural heartbeat of the specific service process.
Takt time is not determined by the number of steps or the workload of each step – it is set by the external demand volume and the number of available hours within which to service that demand volume. Hence, if a pharmacy needs to fulfill 80 orders during a 4-hour period, the Takt time is 4*60/80 = 3 minutes. This means that every step of the end-to-end process from entering the order to receiving the drugs must advance to the next step every 3 minutes as if a bell were being run every 3 minutes and the order was expected to have advanced to the next step. If a particular step takes 6 minutes to get done, you must either improve it through automation or redesign so that it will only take 3 minutes or you must invest in cloning it so that 2 stations are working in parallel and advancing 2 units to the next step every 6 minutes which is equivalent to remaining consistent with the Takt time of 3 minutes. If a particular step only takes 1 minute to complete, consider combining it with other steps to get the overall process in balance.
The easiest way to think about this is to imagine if all of the people involved in each step of the process were running on the same treadmill analogous to the “people mover” treadmill at airports. There would need to be perfect agreement on the speed of the treadmill belt (otherwise one or more people would fall off the treadmill). That is the notion of Takt time.
Of course, no two patients are identical and the mix of patients undergoing a specific process at any time will vary from hour to hour. Therefore, the core concept of the “Takt time heartbeat” of the process needs to be augmented by continuous extraction of timestamp data and sophisticated data science algorithms to classify units of service into comparable clusters and to dynamically determine the appropriate Takt time in order to keep the processes moving as smoothly as possible throughout the day.
In subsequent blog posts, we will demonstrate the manner in which this concept can be applied to several service processes that health systems routinely struggle to optimize including phlebotomy labs, imaging appointments, radiation oncology, etc.