How to manage patients who never show up

How to manage patients who never show up

Executive Summary

 

Appointments in the Out-patient department is the first port of call for every person requiring healthcare services. Surprisingly enough, instances of patients not showing up for pre-scheduled appointments are high and this imposes a huge cost on the system by leading to redundancy, lengthening of queues and low satisfaction scores for patients who were deprived of the service when they needed it. Worst still, it makes the job of the doctor and administrator difficult as it compounds the problems associated with high variability if demand.

 

myCOL has created a dynamic demand scheduler that uses proprietary algorithms that have been created using one of the largest databases in India. This demand scheduler helps hospitals and administrators to manage OPD appointments in such a manner that potential no-shows are estimated day-wise at the level of the disease so that appropriate number of overbookings can be done to ensure optimization of patient wait times and over-burden on doctors. The scheduler effectively replaces the manual effort and estimation which is presently expended at hospitals in scheduling appointments and managing doctor calendars and patient flows.

 

Analysis

 

Health care providers (HCPs) deliver services through the Out-patient Department (OPD) and the Inpatient Department (IPD). OPD involves consultation provided to patients by the doctors without getting admitted in the hospital. Each doctor allocates certain hours of the day to attend to outpatient cases. The hours are sub-divided into slots called appointment slots equaling the expected length of the consultation.

 

The process followed is that patients block appointment slots with doctors in advance to ensure that they are seen and also to ensure that wait times are reasonable. Compare this to the typical post office where people walk-in randomly and are queued into a first come first serve line. The appointments so mentioned are  done either telephonically  or online using either the native online appointment portal of the HCP or via third party appointment portals.

 

Since reneging on appointments have little cost, it has been observed that quite often out-patients do not turn up for the appointment and neither do they inform the HCP about the change in their plans. This is termed as ‘no show’. Since these slots are perishable they go waste or in other words doctor is unproductive for the length of the slot, which is a major loss to the healthcare economy.

 

As one can imagine, the HCPs do not have a mechanism to predict the ‘no-show’ and therefore are unable to optimize the doctor’s time. However, what some HCPs do is to overbook i.e setting up appointments in excess of the capacity as using some back of the envelope calculation. This procedure of overbooking is unscientific and very often forces the doctor to attend to more patient than planned, thus leading to either reduction in the per patient time spent (affecting quality) or forcing to doctor to commit more time than planned on OPD. Such inefficiency imposes at least 2 additional costs on the HCP, namely, Patient wait time and doctor overtime.

 

myCOL has created an analytics product that uses proprietary algorithms to extract information from the existing HCP data to compute accurately the optimal number of patients that should be booked in advance to ensure that

  1. Doctor overtime/overbooking is minimized. Doctors therefore can better manage and plan patient care in both IPD and OPD.
  2. Patient wait time is minimized. Therefore providing an enhanced patient experience.
  3. OPD profitability is maximized while maintaining the quality of service.

 

The algorithm takes into account patient, disease, department doctor, day of week and month and certain other temporal characteristics to determine the optimal appointment booking slots. The illustration below highlights the nuanced approach that has been adopted by myCOL in this matter.

 

Use Case : OPD Payoff Optimizer

 

 

No of appointments slots (say) 200
Probability of no-show (from data) 0.5
   
Current scenario
No. of slots available (capacity) 200
Bookings made 400
Payoff @1 unit per patient with assumption of wait-time penalty=1 and overtime penalty=2 XYZ
   
myCOL OPD Booking Optimization Algorithm
     
Capacity 200
Booking suggested 377
Payoff @1 unit per patient with assumption of wait-time penalty=1 and overtime penalty=2 184.86
 
myCOL OPD Booking + Slot Optimization Algorithm
 
Non myCOL Approach
If there are 3 doctors in the department and 377 slots are equally spaced out  
Capacity 200
Booking suggested 126/126/ 125
Payoff @1 unit per patient with assumption of wait-time penalty=1 and overtime penalty=2 179.59
myCOL Level 2 intervention – optimal booking distribution across slots
Capacity 200
Booking suggested 129/125/ 118
Payoff @1 unit per patient with assumption of wait-time penalty=1 and overtime penalty=2 180.41

 

The OPD scheduler was tested with a high degree of success at a large public hospital in India. myCOL spent considerable time with the management team and the administrators to define in pin-point fashion the objective function of the hospital keeping in mind capacities, costs due to doctor overtime and extended patient wait times. The Hospital identified the tolerance around the capacity (how many more and how many fewer patients) that it wanted to achieve. Post testing it was observed from the empirical data that in all instances, the myCOL booking slot number was always more accurate than the number booked by the hospital using heuristic models. Better still, myCOL was able to help the hospital ensure that it stayed within the tolerance limits over and above capacity in ~90% cases.