Final paper sent to printer
- 1. 1 Faculty Advisor: Prof. Greg Dobson Institutional Liaison: Susan Powell Thursday, December 5, 2013 Recommendations to Improve New Patient Visit Wait Times for the NeuroMedicine Pain Management Program Project Team 1: Addie Bardin Christopher Gallati Raquel Martinez-Calleri Melitta Mendonca Holly Smock
- 2. 2 TableofContents 1. ExecutiveSummary 2 2. Key Definitions and Abbreviations 3 3. Objective 5 4. Background 5 5. ProjectScope 5 6. Market Assessment 7 7. Primary Data Source 11 8. New Patient Visits (NPVs) 11 a. Wait Times 12 b. Demand 13 c. Capacity 15 d. Demand vs. Capacity 18 e. Queuing Model 19 9. Follow Up Appointments (FUAs) 21 a. Utilization of FUAs 21 b. Length of Time Spent in Practice(LOS) 24 10. Relationship of FUAs and NPVs 27 11. Financial Analysis 29 a. Revenue 29 b. Expenses 32 c. Profitand Loss Statements 34 12. Recommendations 34 13. Appendices Attached a. Current Queue Model b. Forecasted Queue Model foradding an MD performing procedures c. Forecasted Queue Model forNP, PA or MD Non-Proceduralist d. Reimbursement Rates by CPTand PayerMix e. 2013 MD PerformingProcedures Reimbursement by CPT f. 2013 MD Non-ProceduralistReimbursement by CPT g. 2013 APP Reimbursement by CPT h. Profitand Loss Statement – MD Performing Procedures i. Profitand Loss Statement – MD Non-Proceduralist j. Profitand Loss Statement – NP k. Profitand Loss Statement – PA l. Procedures Referred by Additional APP m. Procedures Referred by Additional MD Non-Proceduralist
- 3. 3 1. ExecutiveSummary The University of Rochester Medical Center’s (URMC) NeuroMedicine Pain Management Program (NMPMP)must reduce new patient wait times, 80% of new patient visits (NPV) scheduled within 14 days of initial request, to meet the institutional standard. The NMPMP is a healthcare clinic within the Department of Neurosurgery that provides comprehensive pain care. NMPMP is currently experiencing wait times of 30 days. They suspect that demand exceeds their capacity and are considering hiring an additional provider in order to meet this wait time standard. The group assessed the regional market to determine geographic location,services offered,and wait times. We analyzed the clinic scheduling and billing data to determine capacity and utilization. The group forecasted the financial impact of adding a provider to the clinic. The regional market assessment revealed 40 competing pain providers in Monroe County and the 15 surrounding counties. The services offeredvaried by clinic withfew offeringthe same comprehensive care as the NMPMP. Only seven out of forty competing clinicshad a wait time of 14 days or less. Analyzing the scheduling data, NMPMP had 51% of NPVs withwait times of ≤14 days, which is significantly below the desired 80% URMFGstandard. There are 40 NPV requests per weekand only 29.9 NPVs seen. A queuing model confirmed that demand exceeds current capacity and showed whichprovider typeyielded the greatest decrease in NPVwait times. Billing data shows the length of stay for patients and how much clinic time they use is not the driving factor fornew patient wait times. NMPMP is currently not filling their follow-up appointment (FUA) capacity utilizing the current 20% NPV / 80% FUAscheduling model. Financial projections of profit and loss statements were generated for a Nurse Practitioner,a PhysicianAssistant, MD non-proceduralist, and MD performing procedures. We determined that the best option to support their mission, decrease NPVwait times, while still being financially viable is to add capacity by hiring a Nurse Practitioner. Additional recommendations include: blockscheduling, not allowing new patients to request a specific provider, multiple provider / single queue model, charging a cancellation fee, adjusting the 20% NPV/ 80%FUA ratio forappointments, recording actual time spent with patients, tracking patients who choose not to acceptan appointment due to wait time, recording when patients are discharged from the practice.
- 4. 4 2. Key definitionsandabbreviations: APP Advanced PracticeProvider– includes NP and PA Arrival rate Rate of patients calling and booking a NPVappointment Capacity Amount of resource available CPT code Current Procedural Terminology codes used to report medical procedures and services under public and private health insurance programs Data Set Data from January 1, 2011 through June 30, 2013 FUA Follow Up Appointment I STOP – legislation Legislation passed by NYS requiring the provider to record any controlled substance prescriptions on a centralized database by patient Length of stay (LOS) Time between patient’s first and last appointment Market Monroe County and the 15 surrounding counties MD Non-Proceduralist MD whois a pain specialist but does not perform interventional procedures, typically a neurologist or anesthesiologist MD performing procedures MD whois a pain specialist and performs interventional procedures, typically a neurologist or anesthesiologist NCQA An independent, non-profit organization that certifies physician organizations, and accredits managed care organizations and preferred provider organizations NMPMP NeuroMedicine Pain Management Program NMPMP mission “The URMC NeuroMedicine Pain Management Center was established withthe goal to provide the most comprehensive and optimal care in the region by bringing interventional, medical, rehabilitative and psychological approaches to pain management under one roof.” From URMC Website NP Nurse Practitioner-midlevelprovider limited in NPV,85% reimbursement rate NPV New Patient Visit NPVqueue Number of patients who have requested an NPVbut have not had their first visit
- 5. 5 NPVwait times Differencebetween the date of first requesting a NPVand the date of the appointment booked PA PhysicianAssistant-midlevel provider limited in NPV, 85% reimbursement rate, requires close monitoring by MD Procedures Interventional treatments forpain Requests Phone calls received by officefrom new patients seeking a new patient visit Scheduled Patients booked by officestaff for an appointment with a specific date, time and duration Seen Patients arriving fortheir scheduled appointment URMC University of Rochester Medical Center URMFG University of Rochester Medical Faculty Group Utilization Actual use of an available resource
- 6. 6 3. Objective Reduce new patient wait times, the difference between the date of first requesting a NPVand the date of the appointment booked, in the URMC NMPMP and determine the effectof adding an additional provider on financial performance and wait times of new patients. 4. Background The URMC NMPMP is a healthcare clinic within the Department of Neurosurgery that provides comprehensive pain care. The services include interventional, medical, rehabilitative and psychologicalapproaches to pain management. NMPMP provides a unique and comprehensive approach to pain management that is superior to its competitors. NMPMP was named a 2013 Clinical Center of Excellencefor pain management by the American Pain Society, one of only two centers nationwide. NMPMP was founded in 2008 and has grown substantially. As the clinic has grown it has not kept up with demand. The University of Rochester Medical Faculty Group (URMFG)has set standards for waittimes for new patients. This group is responsible for credentialing physicians, negotiating payment rates with third-party payers and is certified by the National Committee forQuality Assurance (NCQA). URMFG has adopted the NCQA’s standard wait time fornew patients: 80% of NPV scheduled within 14 days of initial request. The NMPMP has recently experiencing wait times of 20.1 days and is currently quoting 30 days. The NMPMP is seeking to reduce its wait time to meet this standard. It is assumed that the NMPMP is rapidly growing and experiencing high demand, they suspect that demand exceeds their capacity and are considering hiring an additional provider in order to meet this wait time standard. 5. ProjectScope In order to analyze the NPVwait time problem weassessed the operations and productivity of the clinic, the regional market, and value-added of a new provider. We began by assessing the market to determine who the competitors are, their locations, services and associated wait times. We then obtained scheduling and billing data from the NMPMP. Next, we attempted to analyze the operational efficiency and productivity of the NMPMP.Because there is no standard waittime for follow-upappointments (FUAs), wereduced this focus to only include the NPV queue. Additionally, as the clinic visits are scheduled separately from procedures wehave assumed procedures have no direct impact on the NPVqueue. Our analysis is limited tothe scheduling model and does not include the flow of patients through the clinic. We willdiscuss the ratios and relationships of NPVs to FUAs and procedures. I-STOP legislation was anticipated to increase the demand forNPVs and
- 7. 7 LOS. However,I-STOP has only been in effectforone month at the time of this analysis. Finally, we will determine the financial viability of adding a provider and their impact upon NPVwait times. 6. MarketAssessment Patients choose a physician each time they seek medical care. There are many contributing factors to how the patients make this choice. Marketing, referrals, word of mouth, reputation of provider, insurance coverage, services offered,appointment availability,and geographic locationall play significant roles in this decision process. A market assessment will show whothe competition is, where they are located,the services they offerand the wait times for a NPV. The new patients choosing the NMPMP are primarily referred by their PCP,so for the patient, the significance of direct patient marketing, wordof mouth and reputation of provider are diminished. The marketing efforts of the department are focusedon the PCP;however,the URMC employs a marketing initiative “medicine of the highest order” whichhelps build the reputation of the clinic by its association with the medical center. The clinic accepts all insurances so localand regional patients are coveredby in-networkco-pay and co-insurance rates. It is important to note that the implementation of the AffordableCare Act and the establishment of Accountable Care Networks may change the in-networkavailability of the clinic in the near future. It is believed that if the waittimes for NPVs are too long, the patient willbe referred to a competing practice. Patients will also consider the services offered and how far they are willing to travel. The focusof the market analysis, therefore, willbe on services offered, appointment availability, and geographic location. The region has been defined as Monroe County and the 15 surrounding counties, as determined by the URMC’s director’s office.A search of this region has shown that there are 40 other pain treatment centers or physician officesthe patient, by PCP referral, can choose. The search was confined to practitioners that offermedical management and at least one additional qualifying service to be considered a competitor. The criteria eliminated the holistic practitioners, chiropractors, acupuncturists, physical therapists, massage therapists and other non-traditional practitioners fromthe comparison as they were not considered direct competition but substitutes to the services offeredby NMPMP.A list of clinics and providers whichmet the search criteria were selected fromweb pages, marketing material publically available, and the list provided by the
- 8. 8 NMPMP administration. For those practices without webpages or available marketing material, a phone survey of the competitors was conducted regarding their available services. The highest concentration of competing clinics is in the Rochester area withthe second highest in the Buffalo area. These findings are consistent with the region surrounding these twometropolitan areas being highly rural with less dense populations. Competitors and their services offered, and wait times forNPVs are shown in the followingchart (Chart1): Chart 1 Legend:X= service offered,R=service by referral, *=not available,blank= no service Center MedicalManagement Epidural NerveBlock SpinalDecompression-Surgical SpinalCordStimulatorimplant PhysicalTherapy Acupuncture Counseling-Psychologist Massage ChiropracticCare Hypnotherapy Lifestyle/NutritionCounseling Biofeedback WaitTimeinDays Interventional Pain Mgmt x x x x x x * Finger Lakes Pain Mgmt x x 14 AMS Pain Management x x x x x x 37 Upstate Pain Clinic x x x x x * Highland Pain Mgmt Center x x x x x x x x 21 URMC Spine Center x x x x x x 83 URMC Pain Treatment Cntr x x x x x x x x x x 21 Rochester Brain and Spine x x x x x x x x x x 7 Genesee Valley Pain Center Neuromedicine Pain Mgmt Ctr x x x x x 30 Pain and Symptom Mgmt Ctr x x x r r r r 70 Private Practice x 7 Center for Pain Mgmt x x x r r r r Maxwell Boev Clinic x x x x x 28 Unity Spine Center x x x x x x Pain Interventions x x x x x 21 Rochester Pain Management x x x x 7 Unity Spine Center x x x x x x Finger lakes Spine Center x x x x x 21 Pain Treatment Medicine x x x x x x x 21 Schuyler Pain Management x x * Guthrie Interventional Pain Mgmt x x x 34 Dansville Anestesia and Pain Cntr x x x x x * Unity Spine Center x x x x x x * Jones Memorial Hospital Pain Mgmt Center x x x x * Chautauqua Pain Medicine x x x x x 10 Olean General Hospital x x x x 60 Omni Pain & Wellness Centers LLC x x x x x x x 60 Erie County Medical Center x x x x x x x x x x * Pain Rehab Center of Western New York x x x x x x 180 Gosy and Associates Pain and Neurology Center x x x x x 60 H. Koritz Pain Management x x x x 90 United Memorial Pain Center x x x x x 30 Private Practice x x 1 Advanced Pain & Wellness Institute x x x x x x x x x * Pain Management and Headache x x * Mount St. Mary's Hospital- Pain Management x x * Spine and Sports Medicine x x x 2 Mount St. Mary's Hospital * Buffalo General Medical Center- Pain Mgmt Centerx x x *
- 9. 9 Wait times were collected in October, utilizing a mystery shopper method. The mystery shopper presented as a woman with lower back pain lasting 8 weeks, having Excellus insurance, seeking an appointment with the first available physician in the practice. The caller asked if a referral from their PCP was required and the date of the first appointment available. The wait times between the initial call for an appointment and the date of the actual appointment are plotted in the graph below (Graph 1). Graph 1 Only 7 of the clinics surveyed report a wait time of 14 days or less, they have been categorized ‘blue’ in the comparison chart (Chart 1). Clinics not meeting the URMFG standard of ≤14 days are categorized ‘red’ in the comparison chart. This category includes eight clinics, including the NMPMP, between 15 and 30 days and nine clinics greater than 30 days with the highest being 180 days. There are 9 clinics which are categorized as having scheduling difficulties and they have been categorized ‘orange’ in the comparison chart. Four of these clinics operate without telephone or reception staff. The clinics required the patient to leave a message on their voice mail with a promise of a return call for scheduling the appointment. The remaining 6 clinics had rigorous pre-screening making it difficult to enter their practice. These 0 2 4 6 8 10 7 8 9 9 7 NumberofClinics NPV Wait Times from Initial Call to Actual Visit Clinic Wait times
- 10. 10 screenings ranged from a telephone assessment, which the doctor would review and then return the call if the patient was a candidate for their clinic, to full medical history including MRI, CT, X-Ray and written assessment by the PCP, before an appointment could be scheduled. There are 7 remaining clinics in the comparison chart were discontinued and categorized ‘black’. These clinics are either out of business or have been consolidated with another practice. One of these clinics reported that their doctor left the practice greater than six months prior, and they are having difficulty finding a pain specialist to take his place. A map has been constructed showing the geographic location of the competing providers. The pins are color coded, following the coding pattern above, based on the reported wait times. The map shows that the clinics with shorter wait times are all in metropolitan areas with the rural areas requiring longer waits. It is reasonable to assume that patients living in rural areas would be willing to travel to the metropolitan areas to be seen more quickly.
- 11. 11 Due to data limitations, we cannot determine how many patients are refusing an appointment due to extended wait times. This data is not currently collected in the NMPMP; however, it is the suggestion of this team that the clinic consider the importance of this data in estimating demand, and in their considerations to hire an additional provider. The NMPMP assumes their marketing personnel can generate sufficient demand to meet additional capacity created by adding a provider, by steering PCP referrals back to this practice. In this analysis of the competition, we found that there are only 7 practices in the market that have a wait time of 14 days or less. Of these, only 3 offer the comprehensive range of services that the NMPMP offers. It is reasonable to assume the practice can gain some market share by improving their NPV wait times. 7. Primary Data Source We obtained raw de-identified scheduling and CPT billing data from the NMPMP’s manger of data integrity and analysis. Data was provided from January 1, 2008 through June 30, 2013. Prior to 2011 the clinic was staffed with one MD and one NP. In January 2011 the NMPMP added a second MD, and in March 2011 it added a second NP. The second MD left in August and was replaced in the same month by another MD. The current clinic provider complement, 2 MDs and 2 NPs, began in 2011. Additionally, the clinic did not perform procedures before August 2009, and the clinic moved to its current location in October 2009. Due to these changes before 2011, we analyzed data from January 1, 2011 through June 30, 2013 as this represents the clinic in its current format. 8. New Patient Visits(NPVs) NPVwait times, demand, and capacity of the NMPMP will be determined based on clinic scheduling data. The relationship of demand to capacity was also analyzed. Finally,a queuing model was created and current and forecasted queue lengths and waittimes were determined.
- 12. 12 a. Wait Times Historicaland current waittimes willbe determinedinthis section. NMPMP’smain concern and goal is to improve NPVwait times; therefore, we began withcalculating NPVcurrent and historical waittimes. NPV wait times were defined as the difference between the date of first requesting a NPV and the date of the appointment booked. The average wait time has been steadily increasing since 2011 from14.5 to 20.1 days currently (Chart 2). Chart 2 We subsequently determined the percentage of NPV waittimes of ≤14 days over3 years (Chart 3). We found that from 2011-2013 the percent of NPVs with wait times ≤7 days decreased from 49% to 35% and wait times ≤14 days decreased from and 66% to 51%.1 The 51% of NPVs with wait times of ≤14 days is significantly below the desired 80% URMFGstandard. (Note:2013 is only a half-year, but the trend is similar even when comparing the 1st twoquarters of each year).Also of note, the 51% of NPVwait times ≤14 days we report does match the URMFG’s report, thus our methods of calculation appear to be consistent. 1 Although the URMFG measures their wait time standards by percent of patients with ≤14 days wait, they also report the percent of patients with ≤7 days wait. As this seems to be important to the URMFG and possibly a future standard compliance measure, we have reported this value as well. 10 12 14 16 18 20 22 CY 2011 CY 2012 CY 2013 Averagewaittime (days) Year Avgerage NPV Wait Time
- 13. 13 Wait times have been increasing and quarters 1 and 2 for CY 2013 show an average wait time of 20.1 days. Additionally, only 51% of NPVsare seen within 14 days. Chart 3 *note 2012 and 2011 are full CY, 2013 is a half year b. Demand Historicaland current demandwillbedeterminedin this section. In order to determine demand for NPVs,we used the arrival rate, defined as patients calling the NMPMP and booking a NPVappointment. This arrival rate of NPV requests is not the true demand as the clinic does not currently keep track of patients whorequest an appointment but decide not to schedule an appointment, i.e., customers lost. Including these lost patients and those scheduling appointments wouldyield the true demand. The demand does not truly match that of the arrival rate, yetit appears to approximate it closely,at least with the current wait times. Based on this assumption and limited data, weused the booking of NPVs as the arrival rate of NPVs and calculated the average demand for years 2011, 2012 and 2013. This was calculated as follows: The number of NPVappointments scheduled was determined foreach weekand averaged for each year. 0% 10% 20% 30% 40% 50% 60% 70% %ofNPVrequests CY 2011 CY 2012 CY 2013 % NPV wait times ≤ 14 days
- 14. 14 We found that weekly arrival rates have steadily increased from31 NPVrequests in 2011 to 40 in 2013 (Chart4). Chart 4 We also examined the daily arrival rate of NPV requests for quarter 1 and 2 of 2013. The arrival rate ranged from 1 to 29 per day with a standard deviation of 4.4 (Chart5). 0 5 10 15 20 25 30 35 40 45 CY 2011 CY 2012 CY 2013 NPVrequestsperweek Year Average NPV requests per week
- 15. 15 Chart 5 Chart4 shows that the demand forNPVs is growing overtime and Chart5 shows that there is significant variation fromday to day. In the future, werecommend the clinic record all patients requesting an appointment, not just those booking appointments. The data willprovide the true demand forNPVs whichwill enable a more accurate projection using the queuing model discussed later. Demand has been increasing and currently stands at 40 NPVrequests per week. There is significant variation in NPVrequests from day to day. c. Capacity Capacityof the clinic wasmeasuredto determinewhetherit couldmeet the demandforNPVs. We first began with defining the schedule of the clinic. We consulted withthe NMPMP office manager who provided us withthe hours that each provider is scheduled to see patients foroffice visits and procedures. Table1was constructed and shows each provider and the number of hours they are scheduled to see patients each day of the week and whether they were seeing patients for officevisits or procedures. 0 5 10 15 20 25 30 January-13 February-13 March-13 April-13 May-13 June-13 NPVrequests Date NPV appointment requests per day, Q1-2 2013
- 16. 16 Table 1 Clinic schedule, office visits (hours) M T W T F Total MD 1 6.25 6.25 2.5 15.0 MD 2 6.25 6.25 12.5 MD totals 6.25 6.25 6.25 6.25 2.5 27.5 MD average 13.75 NP 1 4.0 6.25 4.75 4.25 5.5 24.75 NP 2 6.25 6.25 6.25 3.75 22.5 NP totals 10.25 12.5 4.75 10.5 9.25 47.25 NP average 23.63 All providers 74.75 Procedure schedule, patient procedures (hours) M T W T F Total MD 1 7.25 7.5 14.75 MD 2 7.5 5.5 13.0 MD 7.25 7.5 5.5 7.5 27.75 *MD = Medical Doctor,NP = NursePractitioner **Timerepresentedinthe abovetablerepresents onlytimescheduledto seepatients,i.e. all breaks, such aslunch are accountedfor and notincluded. In the clinic each NP workswith a dedicated MD, i.e., they workas dedicated pairs or teams. The twoMD providers have separate clinic and procedure schedules whilethe twoNPs have only clinic schedules. One MD/NP pair shares a common clinic schedule and has separate schedules at times. The other MD/NP pair never share a common clinic schedule. As the focusof the project was to examine and improve wait times fornew patients wefocused on the clinic portion rather than the procedure portion of scheduling. This is reasonable as these officetime and procedure time for providers are “blocked” separately. There are twotypes of clinic visits, NPVs and FUAs. All NPVsare booked for30 minutes and all FUAs are booked for15 minutes. The officedoes not keep a record of the actual time spent with patients. However,this information should be recorded. Knowing the real time spent witha patient allows formore accurate scheduling of appointments and would decrease scheduling variability
- 17. 17 and in-clinic wait and processing times. We first needed to calculate the actual volumes of NPVs and FUAs and the proportion of NPVs to FUAs to determine the clinic capacity for NPVsassuming a stable ratio. Chart 6shows the volumes and their relative ratios of NPVsand FUAs actually seen in the clinic for2011 and 2012. Chart 6 This same approximate ratio of 20 NPVs : 80 FUAs existed for the first twoquarters of 2013 as well (Chart7). Going forwardwe will assume this same 20:80 ratio of NPVs to FUAs for capacity and utilization calculations. Chart 7 1121 1195 4129 4540 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% CY 2011 CY 2012 %oftotalclinicvisits Volume of Visits by Type FUA NPV 539 592 667 1908 2048 2721 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% CY 2011 CY 2012 CY 2013 %oftotalclinicvisits Volume of Visits by Type, Q1-2 FUA NPV
- 18. 18 *Note that the volumes of both NPVs and FUAs are steadily increasing, thus confirming the NMPMP belief of increased demand. Using our 20:80 ratio, and the hours From Table1,we estimated the capacity of each provider (Table2)and the subsequent utilization rates based on the actual number (average) of NPVs each day of the weekin 2013 determined from scheduling data. Table 2 NPVcapacity and utilization for2013 M T W T F Total % NPV 20% 20% 20% 20% 20% Actual NPVs/day (avg) 7.1 6.9 3.8 6.6 3.7 28.1 Total Clinic Capacity (hours) 16.5 18.8 11.0 16.8 11.8 Total Clinic Capacity(# visits/day) 6.6 7.5 4.4 6.7 4.7 29.9 Utilization 108% 92% 86% 99% 78% 93% The findings in general demonstrated that the utilization rates of providers for NPVswere very high, ranging from 78%-108% per day, with Monday being the highest and Friday the lowest.The weekly average utilization was 93%. The weekly clinic capacity is 29.9 NPVs per week given a historically stable 20% NPV/ 80% FUA ratio. The average utilization rate is 93%. d. Demand vs. Capacity Thedemandandcapacity calculated arecompared. Given that the demand is 40 NPV appointment requests per weekand the clinic has the capacity to see 29.9 NPVs per weekgiven the current ratio of NPVs to FUAs, it is clear that demand farexceeds
- 19. 19 capacity. This couldcertainly lead to prolonged waittimes for NPVs to be seen and adding a provider would certainly add capacity to help better meet the demand. e. Queuing Model A queuingmodelisdescribedandsimulationsforcurrent and forecastedmodels were performedto determinethe queuelengthandwait times for NPV. Model description We assumed a Poisson distribution forarrival rate of NPVs.We assumed the probabilities that patients wouldaccept a NPV appointment based on the quoted wait times and constructed a probability table using this assumption. Weassumed a NPVprocessing rate to clear the queue. We determined the length of the queue (number of patients) that have booked a NPVand have not yet been seen. We then simulated each day as follows: 1) The total number of days to process the NPVqueue is determined by dividing the length of the queue by the average number of NPVs seen each day. 2) The number of days is then rounded to the nearest wholenumber and the probability that a patient willtake the offeredappointment based on quoted wait time is determined fromthe probability table. 3) The number of arrivals of patients requesting a NPVis then randomly generated from the probability distribution. 4) The number of patients whoactually booka NPVcan then be simulated and a resulting new queue length and waittimes can be estimated using the assumed NPVprocessing rate. Current model We assumed a mean arrival rate of 8 NPV requests per day calculated as described in the ‘Demand’ section. We assumed the probabilities that patients would accept the appointment based on the quoted wait times to be nearly 100%. We assumed that 29.9 NPVs are processed per weekas determined in the ‘Capacity’ section. We used the length of the queue (number of patients) that have booked a NPVand have
- 20. 20 not yet been seen. The length of queue for 2013 was calculated in the following manner: I. The difference between the first day of clinic that year (1/2/2013) and the date of all NPVrequests was calculated foreach scheduled visit. These were then added sequentially to determine the number of departures. The same was done for the date the NPV was requested to determine the number of arrivals. The “VLOOKUP” functionwasthen utilized in excel to calculate the number of arrivals and departures for each day. The difference between these twocalculated the length of the queue, whichwas then plotted in Chart8. Chart 8 II. The length of the queue ranged from 96-141 patients and averaged about 120. The tight clustering withlittle variation suggests demand exceeds current capacity.If the NMPMP’sdemand wasn’t met or was less than their appointments available, one wouldsee greater variability in the length of the queue. We previously discussed that given the demand exceeds capacity by 10 NPVs, wewould expect to see an exponential growth in the NPVqueue. However,as seen here, it is stable. This may be due to cancellation or no-shows. This is feasible as there is a 17% combined late cancellation and no-show rate (discussed further below). 80 90 100 110 120 130 140 150 0 20 40 60 80 100 120 Patientsinqueue Day of the year Length of Queue, Q1-2 2013
- 21. 21 Currently, the NMPMP has a wait time of approximately 4 weeks and simulations typically projected wait times of 5-7 weeks at 2 months into the future. (See AppendixA) Forecasted model We assumed the same Poisson distribution and mean arrival rate of 8 NPV requests as in the current model. We used the same probabilities that patients wouldaccept appointments fromthe current model. Weused assumed NPVs processing rates based on the addition of another provider. For a MD performing procedures, we assumed 5.5 additional NPVs per week in additional capacity. For a NP, PA,and non-procedural MD, we assumed an additional 9.5 NPVsper week in in added capacity.We used the average length of the queue (120 patients) as described previously as our starting point. Finally, running we ran the simulation the same as we did forthe current model we found that a MD performing procedures would decrease the forecasted wait time to 2-5 weeks at 2 months into the future and did not significantly decrease after that time. (AppendixB) However,the NP, PA or non-proceduralist MD decreased the forecasted wait time to 1-3 weeks at 2 months into the future and it continued to decrease significantly after that time. (AppendixC) A queuing model is useful to project waittimes and queue length in the future. The current model confirmsincreasing wait times. The forecastmodel shows that a NP,PA, or non- procedural MD is preferred overthe MD performing procedures in terms of reducing wait times. 9. FollowUpAppointment(FUA) a. Utilization of FUAs Capacityand utilizationof FUAswas examined. The utilization of providers for FUAs was low,ranging from 39%-53% per day, averaging 45% (Table3).
- 22. 22 Table 3 FUA capacity and utilization for 2013 M T W T F Total % FUA 80% 80% 80% 80% 80% Actual FUAs/day (avg) 27.8 31.2 13.8 21.0 16.2 110.0 Total Clinic Capacity (hours) 16.5 18.8 11.0 16.8 11.8 Total Clinic Capacity(# visits/day) 52.8 60.0 35.2 53.6 37.6 239.2 Utilization 53% 52% 39% 39% 43% 45% The overall utilization rates of provider hours in the clinic forall patient types ranged from 49-64% per day and averaged 55%. These utilization rates may be considered reasonable as the focusof the NMPMP is quality and they spend significant amounts of time workingwith patients. If weassumed the excess unused capacity of FUAs from Table3could be converted to NPVs, the clinic might be able to reduce the wait times and length of the queue. The excess capacity appears to be nearly 130 FUAs or 65 NPVs.Running the queue simulation this wouldeliminate the queue in under 3 weeks time. It is unlikely that low utilization is due to slow process clinic time, because even if patient visits ran longer than scheduled they would all be seen by the end of the day, i.e. if the clinic is scheduled for6 hours and it takes 7 hours to see all of the patients this would not affectthe utilization as we calculated it. There is variability due to cancellations <48 hours and no-shows whichgenerate unused capacity as there is no time to fill these appointments as their patient population has trouble obtaining transportation on short notice (Table4).
- 23. 23 Table 4 CY 2011 CY 2012 CY 2013 % FUAs keeping their original appointment 56% 57% 59% % NPVs keeping their original appointment 64% 65% 61% % FUAs NOS 5% 6% 6% % NPV NOS 6% 6% 7% % FUAs CAN 36% 34% 30% % NPVs CAN 30% 29% 29% % FUAs NOS and CAN <48hrs 17% 18% 17% % NPVs NOS and CAN <48hrs 17% 16% 17% *NOS = no-shows, CAN = cancellations One suggestion to discourage no-showsand late cancellations is to charge patients a fee for doing so. NMPMP believed that NPVs were more likely to “no-show” and FUAs were more likely to cancel. This is not the case, as the rates are nearly identical. Since FUA utilization is low,more of their capacity could be shifted to NPVsto match the NPVdemand that is exceeding capacity and therefore decreased NPV waittimes. However,if trying to match the demand forNPVs in this manner one couldpotentially significantly increase wait times forFUAs and cause patient dissatisfaction. This represents the ability to game the system by defining a standard forNPV wait times and not forFUAs. FUAs are different from NPVs and difficultto standardize as they have variable treatment plans. NMPNP believes there is a shift to more FUAs as a result of extended time patients spend as a member of the practice. One would expect a larger rate of growth of FUAs than NPVs and this is supported by Chart9.
- 24. 24 Chart 9 *Data represents Quarters 1 and 2 of each calendar year. Full 2013 CY data is not available FUA utilization is generally low whichmay be explained by high cancellation and no-show rates. FUAs are increasing at a faster rate than NPVs. b. Length of Stay (LOS) Historicaland current LOS wascalculatedbasedon billingdatato determine the FUA demand of the clinic. Our initial hypothesis was that as NPVsare added to the clinic, and stay in the practicelonger, the availability of appointment slots decreases and the wait time fora NPVincreases. To test this hypothesis the LOS was calculated, using only data from 2011-2013 for the average number of visits patient had per year, and the average time in weeks they spend in the practice. We calculated the number of patient visits per year as follows: The average number of billed visits per patient foreach year and total average visits per patient over their length of stay was obtained from billing data and calculated (Chart 10). 9.8% 12.7% 7.3% 32.9% 0.0% 5.0% 10.0% 15.0% 20.0% 25.0% 30.0% 35.0% CY 2011-2012 CY 2012-2013 %growth Growth by Visit Type NPV rate of increase FUA rate of increase
- 25. 25 Chart10 The average number of visits forpatients appears to rise slightly from 2011-2013 at 2.63, 2.74, and 2.74, withstandard deviations of 1.91, 2.04, and 2.11, respectively. These findings are not conclusiveas some patients have a LOS of more than one year; so calculating the average visits per year is not representative of how many visits patients have. Thus the average number of visits per patient was calculated (represented as “total”) to be 3.81. Due to large variations in the number of visits per patient, the average number of visits does not actually tell us how much FUAdemand is on the clinic or space/time patients are taking up in the clinic, as the average is skewed. Therefore, the amount of scheduled time for each NPVand FUA was calculatedand plotted to determine FUA demand. The distribution of time was calculatedby counting the number of days between each patient’s first and last appointment, dividing by 7 days to show LOS in weeks (Chart 11). 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 averagenumberofvisitsper patient fiscal year '11 '12 '13 total visits Average LOS
- 26. 26 Chart 11 *Notethat the vertical scalehas beenadjusted.Thenumberofpatients at ‘0’ weeks in thepractice is 1349. The range of LOS in weeks is 0-155, 0 represents one NPVthen exiting practice this distribution shows the variation in LOS. While interesting to observe that many patients only stay for one visit, and the majority stay less than 30 weeks, this chart also does not translate to the amount of space/time patients are taking up in the clinic,again not leading us to the FUA demand. Our next attempt to determine the actual amount of space/time patients are taking up in the clinic, each billed patient visit was convertedinto the amount of scheduled time, 15 minutes for FUA and 30 minutes forNPV, then sorted by patient ID,summed foreach patient and plotted as a distribution (Chart12). 0 20 40 60 80 100 120 140 160 180 200 0 6 12 18 24 30 36 42 48 54 60 66 72 78 84 90 96 102 108 114 120 126 132 138 144 150 Numberofpatients Number of weeks in practice Distributionof LOS in weeks
- 27. 27 Chart 12 The chart shows the amount of time, in minutes, that patients spend in the clinic is quite variable. Most patients take up 90 minutes or less of time in the clinic schedule, indicating that the amount of time FUAs are taking up in the system, or rather the FUA demand is not currently the main driving factorfor NPVwait time. The impact of I-STOP,may later affectthis distribution, if patients start spending more time in the clinic to manage their controlled substance prescriptions. The average LOS is 3.8 visits and the distribution is quite variable. The amount of time FUAs are taking up in the system, or rather the FUA demand is not currently the main driving factorfor NPV wait time. 10. RelationshipofFUAsandNPVs Having analyzed NPVs and FUAs demand independently we then examined the relationship between the two. Knowing that the clinic schedules approximately 80 FUAs to every 20 NPVs we needed to determine how many FUAs per weekwere required to sustain the given NPVrate. To determine this calculation welooked at a sample of weeks fromJanuary 1, 2011 to June 14, 2013, where patients already in the practice(had their NPV previous to X week, and their last visit on or after X week)then calculated the percentage of these patients whoactually had an appointment in the given weeks (Chart13). 0 200 400 600 800 1000 1200 1400 0 30 60 90 120 150 180 210 240 270 300 330 360 390 420 480 Numberofpatients Time in mins Total time patients spend in clinic
- 28. 28 Chart 13 Chart13 displays the follow up patients, or patients currently in the practice typically take up approximately twopercent of the appointment schedule. We excluded data after April 13, 2013 as the volumes of patients appear to significantly decrease as we approach the end of the data set and this cause the percentage of FUAs to be skewed. Weplotted the actual count of FUAs against the total number of existing patients in the practice. 0.00% 0.50% 1.00% 1.50% 2.00% 2.50% 3.00% 3.50% % of follow up patients that had an appointment
- 29. 29 Chart 14 *Note thatthe volumesareincreasingin 2011,suggestinggrowthinthe clinic. The number patients having an appointment seems to remain fairly constant, again indicating that the number of or demand of FUAs does not appear to be the main contributing factorforlength of NPVwait times. It is important to consider this percentage when adding new patients to the clinic, as they willneed to increase the clinic time by 2% to accountforthe needed FUAs. This couldcome from some unused capacity,but would eventually require adding overallcapacity. 11. Financial Analysis a. Revenue An analysisofthe revenuegeneratedbyeachphysiciantype will showexpectedgainfrom hiringan additionalprovider. Therevenueis a compilationofreimbursementfromoffice visitsand procedures,gainsharingandcontractedincome. The NMPMP utilizes a blended billing style. The reimbursement rates are different forphysicians and NPs. The NPs are only reimbursed 85% of the reimbursement rate forservices billed under their license. Since they workdirectly with the physicians, they are able to bill 75% of their visits under the physician code,yielding a 100% reimbursement rate for those visits. Some of the payers will not reimburse any amount fora NP to see the patient fora NPV. These visits are either booked to the physician, or billed under the physician. 0 1000 2000 3000 4000 5000 6000 7000 8000 9000 10000 Numberofpatients Number of patients in practice number of patients that had an appt existing patients in system
- 30. 30 Reimbursement rates are set by contractwith the individual payers and the providing institution on an annual basis. The URMC rates are on the SharePoint site http://sharepoint.mc.rochester.edu//sites/MFGBO/referencesandresources/FeeSchedule/default. aspx Under direction of the clinic financial director, the Master Fee Schedules (ie. Office) wasused to determine reimbursement rates. Each individual Current Procedural Terminology (CPT) codewas looked-up and the rates for each payer recorded on a spreadsheet. The rates from 2012 and 2013 were compared for percent change to forecastthe future reimbursements. The data set is limited to twopoints, as available on the site, and thus any change greater than 10% was considered to be an outlier and adjusted to the average percent change for each individual payer. There are seven main payers that were considered (Table5): Table 5 Payer PayerMix Medicare 14% Medicaid 27% MVP 10% Aetna 2% Blue Shield 17% Blue Choice 5% WCMVA 24% Self-Pay 1% 100% MVP is a commercial insurance with a moderate reimbursement rate. The payer mix assigned to MVP is inclusive of all other miscellaneous commercial payers. Self-Pay is uninsured patients. There is no expectation of reimbursement for any care provided to self-pay patients, they are considered charity care. WCMVAis worker’s comp and motor vehicleinsurance combined into one category. They are very good payers forprocedures, but pay very little forofficevisits and consults. The reimbursement rate forthese insurances is set by RVU units. Each CPTcode is assigned an RVU value. This RVU is then multiplied by the rate for officevisits, $8.84, or procedures, $184.12, to determine the reimbursement forthat CPT.See appendixD forbreakdown.
- 31. 31 These reimbursement rates were then multiplied by the volume billed foreach CPT code by provider type. This gave us the anticipated income foreach new provider: MD performing procedures, non-proceduralist, NP/PA. (AppendicesE-G) A profit and loss statement has been prepared foreach provider type, to compare revenue and expense. The goal of the NMPMP is to at least break even on the added provider within 3 years. The profit and loss statement has been constructed utilizing the calculated revenue by provider type and CPTcode volume. It is assumed that if operating at capacity,the volume will remain constant provided the demand is maintained. New providers do not start billing immediately. The proceduralist and non-proceduralist start billing in their third month and the NP/PAlevel providers start billing in the fifthmonth. The revenues forthe first year were adjusted at 83% revenue for physician and 67% revenue forthe NP/PAto accountfor this delayed billing. The pain provider participates in a contracted gain sharing whichyields additional $200,000 revenue per year for the pain provider. The clinic also has a contractwith a provider in Auburn, NY and they receivea monthly ProfessionalService Fee in the amount of $10,000. This is for one proceduralist and one nurse practitioner holding clinic in Auburn twodays a month. The revenue is split $8200 additional revenue forthe physician and $1800 additional revenue for the NP. A similar contractmay be attempted with an additional provider. The non-proceduralist and NP/PAwillindirectly generate additional revenue for procedures. (Appendices H-I) They are unable to perform procedures, but the officevisits will generate more procedure volume. There is excess capacity in the procedure booking blocksand the additional volume can be absorbed by the pain providers by adding additional hours on Wednesdays and Fridays (see Table1). This will yield a higher volume of procedures which reimburse at a higher rate. In 2013 29% of the CPT codes billed are procedures. Applying this to the additional visits created by adding a NP or PA, the procedure volume willincrease by 40%. The additional visits created by the non-proceduralist will increase the procedure volume to 51% of all CPTcodes. Revenue - Average Year MD Proc MD Non-Proc NP PA Provider Revenue 421,570.13$ 248,483.41$ 156,442.44$ 156,442.44$ Increase in Procedure Revenue by % of CPT -$ 157,366.85$ 123,424.98$ 123,424.98$ Professional Service Fees 98,400.00$ -$ 21,600.00$ 21,600.00$ Gain Sharing 200,000.00$ -$ -$ -$ Total Revenue 719,970.13$ 405,850.26$ 301,467.42$ 301,467.42$
- 32. 32 The MD performing Procedures yields the highest total revenue. This is due to the procedures performed. The reimbursement rates forprocedures are much higher than officevisits and the proceduralist is the only provider eligible for the gain sharing. b. Expenses Expenseswerecalculatedusingdatareceivedfrom theNMPMP andindependentresearch. Theexpenseswill showsalaryandbenefits for each providertype,startupcosts,licensingfees andassessments. The salary for each provider was obtained from Salary.com and verified with the department. The average national salary was selected for each provider. This was accurate to the department estimates withthe exception of the Pain Providerwith Procedures. The NERVES (Neurosurgery Executives’ Resource Value and Education Society a society conducting surveys and analysis for neurosurgery practice managers and administrators) survey was used to verify the salary forthe pain provider and the lower 25% salary was determined to be too high. The salary fromSalary.com and NERVESwas averaged to reach the $380,310.00 estimate. It is assumed that the providers will each receive a 2% salary increase per year. The URMC assigns a sliding scale to the cost of benefits for the providers. The benefit rate forthe Pain Provider– with procedures is 18.25% of base salary. The benefit rate forthe Non- Proceduralist is 27.77% of salary and the benefit rate for the NP or PAare 37.3% of salary. Startup costs are comprised of recruiter fees, travel expenses for interviews, moving expenses and officeset up including furniture and computers. The recruiter fee fora physician is between $25K and $50K, a mean value of $37.5K has been assumed. For the NP / PAthe recruiting fee is between $8K and $10K, a mean value of $9K has been assumed. The interview travel expenses fora physician has been calculated at $20K, allowing for 2 trips each for5 candidates. The interview travel expenses forNP / PA $250.00, allowing for 5 candidates one trip each. The NP /PA candidates are recruited from the local region so there is limited expense involvedin the interview travel. Moving expenses are allotted to physicians depending on the distance of the move. The range is $8K to $30K, a mean value of $19K was assumed. No moving expenses are allowed for NP / PA providers.
- 33. 33 Malpractice insurance rates have been estimated by taking the current year rate for one of the procedural MDs. URMC malpractice is self-insured with four peer academic medical centers paying into a resource pool forthe payment of claims. There is an expected 5% increase in the rate per year. This rate has been applied to the MD performing procedures and MD non-proceduralist. The URMC purchases a bulk malpractice policy forthe NP and PA staff. There is no cost at the department level for this coverage. Dues wereassumed at one professional organization per provider. The fees vary by organization; however,an average fee has been suggested by the clinic financial director. Licensing fees assessed every three years and set by the license agency and are provider level specific. Travel expenses are provider for professional conferences. The rate was determined assuming 3 trips per year for a physician and 1 trip per year for the NP / PA at a rate of $2,500 per trip. The department has set a rate of $5,000 per provider for phones, transcription services and miscellaneous expenses. The assessments and fees are set by the URMC and are based on a percent of revenue. The MD performing Procedures incurs the highest expenses. The PA incurs the lowest expenses. These must be compared against revenue to determine the most desirable provider. Start Up Costs MD APP Recruitment 57,500.00$ 9,250.00$ Office Set Up 20,250.00$ 10,250.00$ Moving Expense 19,000.00$ -$ Total Start Up Costs 96,750.00$ 19,500.00$ Expenses - Average Year MD Proc MD Non-Proc NP PA Salary & Benefits 467,885.12$ 285,779.86$ 139,775.71$ 131,552.01$ Malpractice Insurance 8,699.83$ 8,699.83$ -$ -$ Liscences / Dues / Travel 9,545.00$ 9,545.00$ 3,545.00$ 3,545.00$ Fees & Assessments 64,324.03$ 48,630.30$ 55,096.27$ 55,096.27$ Total Expenses 550,453.98$ 352,654.98$ 198,416.98$ 190,193.28$
- 34. 34 c. Profitand Loss Statements Therevenuesandexpenseshavebeenappliedto aprofit and loss statement,byprovidertype (Appendices J-M). The analysis shows that the MD – Non-proceduralist is the least profitable provider option, not showing a positive accountbalance until year 5 showing a total gain of $55,026.39. The MD performing procedures is the most profitable showing a positive accountbalance in year 1 and a total gain of $624,031.21 at the end of year 5. The NP and PA both show positive accountbalance in year 1 and a total gain of $435,491.83 and $476,626.44 respectively. Although the MD performing procedures yields the higher gain, hiring a provider at this level is riskier. The time to recruit and hire the provider is longer, the upfrontcosts are higher and if an MD is hired, additional administrative staff will be required. The NP does not require additional administrative staff and the upfront costs are $77,250 less than the MD. 12. Recommendations The NMPMP is currently exceeding capacity in requests for NPVs. This is causing a queue to form and the clinic to exceed the goal of 80% seen in ≤ 14 days. It is clear that clinic capacity must be increased to meet the demand and the goal. The followingrecommendations are made based on the analyses described in this paper. Immediate changes that are possible based on evidence FUA utilization is low and more of capacity couldbe shifted to NPVs, therefore decreasing NPVwait times. However, if trying to match the demand for NPVs in this manner could significantly increase waittimes for FUAs and cause patient dissatisfaction. This represents the ability to game the system by defining a standard forNPV waittimes and not for FUAs. The clinic does not block time for NPVs or FUAs; they schedule them on a “first-come” basis. Scheduling FUAs 15 minutes apart takes away slots for30 minute NPVs, thus NPVs are more difficultto fit in. By setting aside a specific amount of time foreach type of appointment, we can reduce variability. This is commonly referred to as blockscheduling. Gain Year 5 MD Proc MD Non-Proc NP PA Revenue 3,534,102.08$ 1,890,690.88$ 1,431,001.80$ 1,431,001.80$ Expenses 2,910,070.88$ 1,835,664.49$ 995,509.97$ 954,375.36$ Gain (Revenue over Expenses) 624,031.21$ 55,026.39$ 435,491.83$ 476,626.44$
- 35. 35 Given the average 20% NPV/ 80% FUA wewould recommend using this to blockratio for the clinic Attempting to schedule NPVon a first provider available basis wouldreduce waittimes and allow better predictability of the queuing from the simulation rather than allowing patients to request specific providers. By doing this, patients’ wait times will be shorter as there is a single queue with multiple providers. This can be done at the time of booking the appointment and again when they arrive to the clinic for both NPVsand FUAs, by seeing the next available provider. This coulddissatisfy follow-uppatients whoprefer their existing provider. Discourage no-shows and cancellations with <48 hour notice by charging the patient a no- show fee, a widely acceptedpractice in the medical field. Data to collectfor future decisions Record the actual time spent with patients. Knowing the real time spent with a patient allows formore accurate scheduling of appointments and would decrease scheduling variability and in clinic wait and processing times. Record the number of patients who callto request an NPVand decline based on the wait time. This will give the true demand of the arrival rate and show the expected increase in booked appointments if the waittime issue is resolved. Record when the patient is discharged from the practice. NMPMP believes patients are staying longer and suspect that I-STOP willincrease LOS. Keeping this data will enable them to monitor LOS and demand forFUAs. Long Term Solution Expand capacity in the clinic by hiring an APP.I-STOP legislation is changing the nature of the visit. It is reported by one of the NP providers that there is an increase in the number of patients who believe they need medical management. The I-STOP database is giving the providers more information about the regional treatment of the patient and is enabling them to avoidprescribing unneeded drugs. The NP is spending extended visit times with these patients to offercounseling and direction fortheir treatment. It is better use of clinic resources foran APP to filter out these drug seeking patients.
- 36. 36 The clinic model yields a high volumeof medical management and drug monitoring. The volume of procedures is reasonable, according to the clinic director. This model is not the best financial model for a pain treatment clinic;however, it meets the mission of the URMC. Adding an APP to the practice allows forreferrals to the MD performing procedures forthe patients requiring interventional treatment. This adds medical management cases to the clinic and increases the procedures, which yield a higher reimbursement rate an estimated $99K in the first year, without the higher expenses associated with an MD. The current staffing model is paired teams of one MD and one NP workingas a team. This willbe unbalanced by the addition of an APP,howeverthe clinic has structured their schedule so that an MD is in clinic at all times. The added APP willworkunder both MD’s and the attending will be the MD in clinic on the given day. Since an NP yields similar gain to the practice and requires less direct supervision than a PA,and it is recommended that an NP be added to the clinic.