Transcript Slide 1
Medina Healthcare System: Centralized Scheduling Center 1 Inbound Calls Patient calls 30 sec. Automated message Scheduler available? Schedule apt.? Yes No Yes No Placed at the back of queue Wait ≤7min? Take message? Yes Rolled to registration No Call picked up? No Scheduler greets patient Transfer call to hospital department No Get up and find MA Yes Yes Physician to Physician Get up and find physician Pharmacy to Physician Patient calling MA to return call 2 Process Flow Chart: Scheduling Center Inbound Calls Inbound Calls Cont. New patient? Yes Verify Insurance Problem w/insurance No No Input info Re-schedule/ Schedule followup Schedule patient Input message Yes Immediate doc. contact? Yes Patient on hold Call insurance company/v erify Call doc No Forward doc message/ printer Call patient back Process Flow Chart: Scheduling Center Inbound Calls (Continued) 3 Data Analysis June 1, 2012 – May 30, 2013 4 • Average queue time per 15 minute time interval Data Limitations • Average service time per 15 minute time interval • Reduces variation; data in the tails • Does not include the time the patient spends on “make busy” • Calls interflowed that return to the queue are then tracked as if they are new calls 5 7 8 9 Data Analysis June 2013 10 June - Record of changes made 04/2013: Added phone tree. Issues: Incorrect routing, Took too long to get through the phone tree. Interflow noted as a problem 06/01/2013: Took out phone tree, routing to physician line removed Issues: Interflow a continued problem 11 12 13 14 15 Data Analysis July 1 – July 23rd 16 July - Record of changes made 06/29/2013: Added physician phone number to beginning of automated message. 7/01/2013: Removed interflow completely. Began using referral center staff during peak times (Mornings) 7/02/2013: Radiology dept. staff began giving patients in need of follow-up appointments the radiology scheduling line while reminding them not to call the clinic scheduling line for these appointments Clinic scheduling center staff began sending emails to employers requesting information needed to process workman’s comp claims Started using Recondo. Avoiding the rework of pasting information into templates previously entered – Faster insurance verification 17 18 Process Capability of Converted USL P rocess Data LSL * Target * U SL 0.004 Sample M ean 0.00312725 Sample N 36 StDev (O v erall) 0.000354939 O v erall C apability Z.Bench 2.46 Z.LSL * Z.U SL 2.46 P pk 0.82 C pm * 0.0024 O bserv ed P erformance % < LSL * % > U SL 0.00 % Total 0.00 0.0028 Exp. O v erall P erformance % < LSL * % > U SL 0.70 % Total 0.70 0.0032 0.0036 0.0040 Mean shifted from 5:00 in June to 4:30 in July 20 Mean shifted from 5:00 in June to 4:30 in July 21 Hypothesis • Ho: June and July cycle times were essentially the same • Ha: June and July cycle times were significantly different 22 Hypothesis Testing Two-sample T test for Cycle Time vs Cycle Time1 N Mean StDev SE Mean Cycle Time 36 0.003477 0.000304 0.000051 Cycle Time1 36 0.003127 0.000355 0.000059 Difference = mu (Cycle Time) - mu (Cycle Time1) Estimate for difference: 0.000349 95% CI for difference: (0.000194, 0.000505) T-Test of difference = 0 (vs not =): T-Value = 4.49 P-Value = 0.000 DF = 68 23 24 Comparison of Cycle Times 25 Other important numbers… Average Delay to Abandon 2012-2013: 1:45 June 2013: 2:03 July 1-23rd 2013: 2:01 Calls abandoned 2012-2013: 21.47% June 2013: 23.73% July 1-23rd 2013: 17.59% 26 2nd Hypothesis: Ho: Regardless of the call type, data is essentially the same Ha: Call cycle time is significantly different between call types 27 • Manually tracked • Not all calls fit cleanly into a category Data Limitations • Time includes service time but not time spent waiting • Time does however include time spent on “make busy” • Data sampled from 3 different time periods on 6 different days • No interflow point 28 Call Type/Location/Time Interval Data – July 29 Call Type Pa 140 120 100 80 60 40 20 0 100 80 60 40 20 t t r ) ll t. le n y s rs e an a p u i e a e e h c i a -r l th ed ys ic A qu Ot at X e h e O p M r c h , nc s p b g w s a ' c pt ne s (l a r nin nt or p f e a le ti g lt tu s u a p u e r -u e m ed es tp t r h a w v r en sc lo i a o l nt t f e o e l a ll F ti p a a to C p g e n l l li du a c e t ch n s tie Re Count Percent Cum % 54 25 15 14 8 7 6 5 2 39.7 18.4 11.0 10.3 5.9 5.1 4.4 3.7 1.5 39.7 58.1 69.1 79.4 85.3 90.4 94.9 98.5 100.0 0 Percent Count Pareto Chart of Call Type 31 32 33 Take Aways It is vital that systems are in place to accurately report on performance of the call center. Times need to include “make busy” and at least include a range or standard deviation for the wait time and service time per time interval Seemingly minor changes in the processing of calls makes a significant impact – Interflow point removal – Recondo – eliminate rework – Workman’s comp calls Adjusting staffing for peak times will reduce time spent waiting – Referral coordinators used during peak times Potential improvements: Designate a staff member to focus only on insurance verification during peak times Redesign how messages for physicians are processed 18% of calls in July were from patients who were calling to leave a message for their physician 34 The Path Forward 1. Develop reports that track call center data. Enabling us to analyze performance and continually make improvements 2. Create weekly and monthly goals for scheduling center staff 1. Improving patients satisfaction through decreased wait times and a more consistent scheduling process 3. Ensure changes and improvements are sustained through monitoring the process using statistical process control charts and re-evaluating every quarter 35