Table of Contents. Executive Summary Introduction and Background 1.1 Goals and Objectives. 2.0 Description of Current System

Transcription

1 Table of Contents Executive Summary 1.. Introduction and Background 1.1 Goals and Objectives 2. Description of Current System 2.1 Intake Process erification 2.2 Appointment Analysis 2.3 Telephone Management 2.4 Medical Assistance 2.5 Model Experimentation 3. Approach and Methodology 3.1 Intake Process erification 3.2 Appointment Analysis 3.3 Telephone Management 3.4 Medical Assistance 3.5 Model Experimentation 4. Findings and Conclusions 4.1 Intake Process erification 4.2 Appointment Analysis Pediatrics Continuity Clinic OB/GYN Clinic Family Practice Clinic 4.3 Telephone Management Telephone Memo Analysis Telephone Time Study 4.4 Medical Assistance 4.5 Model Experimentation. 5. Recommendations 5.1 alidation of Simulation Model 5.2 Appointment Analysis 5.3 Telephone Management System 5.4 Medical Assistance Analysis Appendices Appendix A: SimulationModel Flow Chart Appendix B: Check-In I Paperwork Flow Chart Appendix C: Telephone Management Flow Chart Appendix D: Model Experimentation Output Appendix E: Check-In / Paperwork Raw Data Appendix F: Appointment Analysis Raw Data Appendix G: Medical Assistant Time Study Form Appendix H: Sample of CLS Scheduling Data Appendix I: Phone Call Distribution Graphs Appendix. 3: Phone Call Time Study Data Appendix K: Medical Assistance Time Study Raw Data

2 Executive Summary The purpose of this project is to extend, refine, and modify the current baseline simulation model for the development of clinic operations in the East Medical Campus. Kodak has worked with the Program and Operations Analysis Department at the University of Michigan Medical Center to develop the current simulation model through the use of MedModel software. The data used in the current simulation was provided by the Program and Operations Analysis Department. Data was collected from the medical center s computer information system and time studies performed on various processes. Due to time constraints on obtaining data for use in the baseline simulation, estimates were used for some processes, while other processes were not incorporated into the model at all. In our efforts to refine, extend, and modify the current model, we collected and analyzed data on the processes of patient intake, MA assistance, and telephone management. To collect data on these processes, times studies were performed by the project team and clinical staff, data was tabulated from the clinic scheduling system, and clinic memos were reviewed. In addition, interviews were conducted to confirm the results of our analysis of the data collected. Upon completion of the data collection, a detailed statistical analysis was performed for each process. From our analysis we conclude the following: the distribution of the time associated with the patient intake process is not uniform unscheduled MA assistance occurs infrequently as a percentage of the number of appointments the time a nurse spends on the phone is small in comparison to the amount of time required for phone call related paperwork and research there are substantial differences between patient no show rates and appointment type distributions by clinic a change in the distribution of the intake function noticeably affects the wait time and length of stay of a patient, while having little impact on provider utilization The need for unscheduled MA assistance varied by clinic. If the utilization data is to be folded into the model, the differences between clinics should be included. However, because of the infrequent need for unscheduled MA assistance, including this utilization in the simulation model may not be necessary. A direct transfer approach to phone calls is a potential way to improve the current system. Instead of the clerk making a memo for the nurse to return pertinentcalls, it would be feasible for the nurse to answer the call directly. Access to patient information through computerization would decrease the time spent on paperwork and research. Modification of one process distribution in the simulation model noticeably affected the performance of the clinic. Work needs to be done to continue refining the process time distributions so that the model results adequately reflect real life. Differences between clinics need to be considered when simulating each clinic s operations. Our analysis of scheduling data showed a strong distinction between the percentages of appointment types for different providers in the clinics we examined. In addition, we found patient no show rates vary by clinic. If these differences have a strong impact on clinical operations, they need to be accounted for in the simulation model.

3 The main issues regarding the use of simulation are the accuracy with which the model pomays actual operations and the level of complexity used in creating the model. By collecting data in great detail on every process, operations can be represented with a high level of precision. This level of detail, however, becomes exnaordinarily me consuming. For practical purposes and because of the cost associated with such data collection efforts. limitations must be placed on the level of detail incorporated into the simulation while still maintaining an accurate representation of the current system. Furthermore, the output of the simulation should be checked to ensure that it accurately estimates the current results of the system. Once the simulation model accurately reflects operations. modifications can then be made to the simulation to investigate ways to improve operations before actually altering the current system.

4 1. Introduction The purpose of this project is to extend and refine the base simulation model for the East Medical Campus. Upon its completion, the East Medical Campus will be home to several University of Michigan Medical Center clinics. The computer model is a tool used to simulate the operations in these clinics involving patients, employees, and paperwork before they are opened. MedModel software allows the user to visually track the flow of these entities while running the program. Once the model is run, the software can output results of key variables that serve as measures of clinical efficiency and customer service times. The Program and Operations Analysis Department will use this computer model to evaluate and improve existing and future clinic operations. 1.1 Goals and Objectives In our project, we performed analyses and experiments to help move the simulation model forward for the inclusion of other clinics and necessary employee functions. The simulation model will help the Programs and Operations Analysis Department obtain three maingoals: improve internal processes of clinical operations reduce patient wait time and length of stay increase employee utilization During the process of completing our project we have accomplished the following objectives: erification of current process distributions for the intake process Collection and analysis of telephone management and medical assistance data for incorporation into the current model Model experimentation upon modification of the current model U i

5 medical advice the attending patient calls. the 2. Description of Current Systems Currently, the University of Michigan Hospital is using the simulation package, MedModel, as developed by Eastman Kodak Company. The Eastman Kodak Company and the Program and Operations Analysis Department devised a flow chart, shown in Appendix A, for the clinic operations of the U of M East Campus Internal Medicine Clinic. The time intervals for the different functions were either estimated or studied in Internal Medicine with the assumption that many process rates remain constant throughout the clinics. 2.1 Intake Process erification The only area of our study that is currently included in the simulation model is the intake process. One part of this function is the escort wait time or paper work time. The time interval that defines this function is the time a clerk begins a patient s registration and chart work until the clerk places the chart in a bin as ready for a medical assistant. Currently, one.clerk works at the check-in desk to greet patients and prepare their charts and paperwork for the medical assistants. At times when the check-in desk was quite busy, another clerk would put aside her work and assist with the check-in process. A flowchart of this process is located in Appendix B. The range of times used in the current model is a uniform distribution from 3.5 to 5. minutes. This interval is an estimate established by the.internal Medicine Clinic, not an actual time study. A time study was needed to modify the current distribution. Once we completed the time study analysis, the data can be used to modify the current simulation. Upon modification, the simulation results will be compared. 2.2 Appointment Analysis Distributions of appointment lengths and provider types of the current system needed to be developed for other clinics in the East Medical Campus. Distributions of appointment times and provider types are an important piece in the simulation logic. The distribution of this data is the eighth step in the simulation flow chart (see Appendix A) and is denoted by the percentage of doctors versus nurse practitioners that see patients for 15, 3, or 45 minute appointments. 2.3 Telephone Management One area of our project that are not currently included in the simulation model is the telephone management system used by nurses in the clinics. Prior to our study, statistical data did not exist on the length and frequency of calls to and from clinics. The data collected and analyzed on these processes will be used in further modifications of the current simulation. In general, clerks answer all incoming telephone calls to the clinic. At that time, the clerk is either able to answer the patient s question easily, such as a request for appointment time, or must fill out a form for a nurse s consultation or return call. In some cases, when doctors are available, they will attend to some of the necessary calls. Normally, two nurses are scheduled for this clinic each day. The nurses number one priority is attending to patient needs and their second priority is to When a nurse is available, they will return calls from phone memos in the Nurse Triage area. Nurses return patient calls according to a specified prioritization. For instance, nurses will return a patient call for before placing a call for request of immunization forms. See Appendix C for a detailed flowchart of the telephone management process.... 2

6 the MAs tasks. The frequency and duration of these instances needed to be collected. needed. A doctor may request an MA for the entire length or part of an exam for Currently, the doctors at the Northeast Ann Arbor can request an MA s assistance as model Times when an MA is requested for unscheduled assistance is not incorporated in.3 predicted, and actual performance. As stated previously, some of the current intervals are As the model is currently defined, a gap in clinic operations exists between desired, estimated rather than studied and two essential functions of the clinic, telephone management and MA assistance, are not included in the current model. After collecting data for the current processes described above, modification of the simulation and experimentation with the model will provide the Program and Operations Analysis Department with the ability to determine the value of applying simulation to existing and future clinic operations. 2.5 Model Experimentation assistance, as a chaperone, or to finish any procedure. Before our study there was no data that determined the length, frequency, or probability of such a request. The objective is to incorporate the utilization of MAs for assistance in the current simulation. A full description of medical assistants utilization is not included in the current simulation 2.4 Medical Assistance

7 (, 3. Approach and Methodology 3.1 Intake Process erification Teresza Poe, Patient Representative at Northeast Ann Arbor Clinic, gave our group a tour of the clinic and explained the check-in process in the registration are& Teresza clarified the steps a clerk takes to prepare a chart once the patient has arrived to the clinic. A data collection form was developed to measure the time a patient is greeted by a registration clerk and asked for insurance verification until the time the patient s chart is prepared and placed on a rack to be picked up by a medical assistant or nurse. According to Lisa Cayen, it is expected that the time interval being measured should be less than the 3.5 to 5 minutes designated in the current model. We decided that the time interval should be measured with an accuracy of plus or minus 5 seconds. Since patients continually arrive to the clinic, a continuous time study performed by team members was determined to be the most accurate way to collect data. We collected data for approximately 12 hours during which data was obtained for about 12 patients. The intake process data was analyzed by entering data into an Excel spreadsheet and calculating statistical measures such as the mean and standard deviation. All data was pooled together in the statistical analysis. Since team members collected data in 4 hour time blocks, statistical hypothesis tests were performed to verify that the means associated with each time block were not statistically different. Lisa Cayen and Teresza Poe both indicated that appointments are scheduled constantly through the day on Monday through Friday. Since patients are continually arriving in the clinic, we did not analyze the data by day of the week or time of the day. A histogram of the data was plotted to study the distribution of patient wait time during the check-in process. The cumulative distribution function of the check-in process was modified in the current simulation. Model experimentation was then performed and discussed in further detail later in this report. 3.2 Appointment Analysis Data regarding appointment length and provider type data was collected from the Medical Center s CLS scheduling system for the Northeast Ann Arbor Clinic. After meeting with Lisa Cayen and obtaining access to the CLS system, we printed reports for appointments occurring between July 17, 1995 to July 31, 1995 at Briarwood Family Practice, Pediatric Continuity Clinic, and OB/GYN. Appointments where patients arrived and appointments in which the patient did not show up (no shows) were tracked. The sum of these appointments equals the number of appointments scheduled over this time period. Reschedules and cancellations were not included because the appointment slots could be filled which would result in an appointment. The distribution of appointments lengths by provider type was determined after calculating the percentage of appointment intervals in each clinic. An example of the CLS system output is included in Appendix H. 3.3 Telephone Management Telephone memo data sheets were obtained from the Pediatric Continuity Clinic. If a clerk answered a telephone call in which assistance was needed by a doctor or nurse, then the clerk completed a telephone memo sheet so that the call could be returned later by the appropriate employee (an RN or MD). The data sheets were collected from June 1, 1995 to October 4, 1995, and Sue Nebring, Nurse Manager, estimated that the data sheets capture 9% of the total calls that resulted in a written memo. We met with Sue Nehring, Nurse Manager, to discuss appropriate categories by which to analyze the phone calls. From our observations and her knowledge of the telephone process, we chose the following categories for the phone call subject matter: 4

8 Medical Advice General medical advice Schedule an appointment Advice which resulted in sending patient to ER Referrals Prescriptions Forms (sent, faxed, or picked up by patient) Immunization forms Crippled Children s forms Other forms Calls from people other than parents (e.g. Homecare, schools) Calls on how to access telephone system Test Results With Ms. Nehring, we also discussed the procedure to be used to obtain data on the length and category of return telephone calls. A time study was determined to be appropriate. Data from approximately 2, telephone memo data sheets was entered into an Excel Spreadsheet. The data was analyzed by frequency and statistics in the following subgroups: Day of the week (e.g. Monday, Tuesday...) Time of the day by 1 hour intervals (e.g. 9: am to 1: am) Time of day by day of the week in 1 hour intervals Category of call (e.g. prescriptions, medical advice, referrals...) Number of return calls needed (e.g. phone busy, no answer Not all memos contained complete information so the sample sizes for each subgroup used in the analysis vary slightly. Additionally, unless written on the memo, it was assumed that only one call was need to reach the patient. A time study was performed in the Pediatric Continuity Clinic under the supervision of Sue Nehring to capture times involved with the return of patient calls. Because it would be very inconvenient for staff to collect phone call lengths, we collected the data ourselves. Phone calls are typically only a few minutes long, therefore accuracy is important. Due to the complexity of the process surrounding the response to a call, a time study did not provide all the necessary information. More importantly, the only time captured by this study was the actual time spent on the phone. After further investigation and suggestions from Sue Nehring, we learned that the response time of returning calls should also include the paperwork and research done both before and after a call is made and any additional calls necessary to respond to the initial call. A nurse may work on multiple memos at once or leave the reception area to work on memos, so it is very difficult to collect accurate measures of each activity associated with a single patient. Knowing how many memos were responded to and including all the time spent to complete the response process, we calculated the average total response time per memo. 3.4 Medical Assistance A time study of the MA assistance process at the North East Ann Arbor Clinic was conducted under the supervision of Nancy Eggerer, Nurse Manager. The data needed to analyze the time for medical assistance was collected by the medical assistants (MAs) themselves. We decided data could be collected more efficiently if the MAs complete a daily assistance form. TI we collected data by monitoring the MAs, then only one MA could be monitored at a time because they assist throughout all of the clinic rooms. The 5

9 MAs have other tasks to complete, so the majority of their day consists of activities other than providing assistance. Since the assistance time is expected to be greater than five minutes, accuracy down to the second is not necessary, so a MA recording the time to the nearest minute will be acceptable. Each MA attached a daily data form attached to a clipboard equipped with a stop watch. An example of the data collection form can be found in Appendix G. The following information was collected from the form: Time assistance begins Completion time of the assistance The clinic in which the assist was needed Whether a doctor was present during the assist When the doctor left if present A data sheet was given to every medical assistant each day on which they recorded this information daily for one week from November 15, 1995 to November 22, To determine the frequency of MA requests, the number of appointments per clinic during this time period was derived from the CLS scheduling system and then divided into the number of MA requests per clinic to give the frequency of MA requests by clinic. 3.5 Model Experimentation The simulation model is run on an IBM computer in the Program and Operations Analysis department. To familiarize ourselves with the MedModel software, we met with Roy More who walked us through the current simulation and suggested that we work with the tutorials available in MedModel. The MedModel tutorials introduced us to the details needed to build a model and output results. After completing the MedModel tutorials, we began working to understand the current model s objects and code. Joseph Kent, the programmer from Kodak who developed the model, was made available over the phone to answer any detailed questions we had about the model. As we began to understand how the model worked, we ran initial experiments to introduce ourselves to the various results the model can track and output. Working with the manuals to determine the location of defined distributions, we altered the intake process time distribution to reflect the data we found and saved the program as a new model. Now we could run the old model and the revised model and compare the results of each. After defining the variables that we wanted to track, the appropriate changes were made to both models so that statistics would be maintained for these values. We ran fifty replications of each model and tracked fifteen different variables related to patient wait time and employee utilization rates. The results of this experimentation are included in the Findings and Conclusions section, and the raw data can be found in Appendix D..6

10 for each data set are summarized in Table 1. The mean and standard deviation are shown for each individual trial and then for the data grouped as a whole. The overall mean was determined as 3. minutes. The standard deviation was quite large at minutes. The relatively large variance can be attributed to the variety of forms that needed to be processed for different appointment types. Table 1. Statistics of Escort Wait Time Study 4.1 Intake Process erification The escort wait time study determined the distribution of the length of time from a patient handing the clerk his insurance card until the patient s chart was completed for the MA. The raw data sheets for the escort wait time study are available in Appendix E. The results found that the means of the three sets of data were not significantly different. Once the data appeared to be statistically consistent, the histogram for the length of wait Table 2. Results of Mean Hypothesis Test was graphed as shown in Figure 1. This figure shows the distribution of paperwork process times. The distribution is skewed to the right with the average patient waiting 3. minutes. The minimum patient wait time was 4 seconds; while the maximum was 7 minutes and 45 seconds. The cumulative distribution function is also plotted in Figure 1. Table 3 outputs this cumulative distribution numerically. In the simulation model, this cumulative distribution function is how we defined the modified process time. 2&3 1.5 Accept I Accept I Accept, Data Sets to Result: of collection, a hypothesis test was conducted on the means. In performing the test, we did In order to assume that the data from the three trials were not dependent on the day or time a t-test using an alpha of.5. For the hypothesis that the means are equal to be rejected, the t-value must be greater than 2.. As shown by the summary of the test in Table 2, we ( total Data Set Count Mean(min) Std Deu(min) 4. Findings and Conclusions

11 Figure 1. Distribution Function of Paperwork Times The CLS data was tabulated and the results are displayed in the following tables. The 4.2 Appointment Analysis results are displayed in the following sections according to individual clinics. The raw number data can be found.in Appendix F. 4/os + t. q Aol -j- 8% +3% 18 t7 C nli I 16%. 9% T.5 % 1. 2% % 2. 31% % 3..6% % 4. 77% % 5. 88% % 6. 96% % 7. 98% % 8. 1% Table 3. Cumulative Distribution Function for Escort Wait Times Escort Wait Time (mm) Cumulative % o 6t Time (min:sec) o - - CM J C ) C ) IC) U) CO CD t- r. o a a a a a a o o a a o o..- CM CM C ) C ) IC) U) CO CO I- t. o C ) C ) C ) C ) C ) C ) C ) C ) % -I % 2%1-11j +1% t2 () E 8%t ±Ano, :I LI I -I- rio! I 1 T /e I I t5% Cu fr I r iml C) C) C) C) C) C) C) ) C) ) C) C) C) C) C) C) CM L() CM if) CM In CM if) CM IC) CM if) CM U) CM IC) 8

12 4.2.1 Pediatrics Continuity Clinic Results from the Pediatrics Continuity Clinic are based on 374 appointments for residents and 189 appointments for faculty. Table 4. Arrivals and No-Shows by Provider Type in Peds Continuity Clinic Provider Type Appointment Residents Faculty Length % Arrivals % No Shows % Arrivals % No Shows 15 mm 77% 23% 87% 13% 3 mm 8% 2% 85% 15% Table 4 suggests that residents and faculty experience similar arrival and no show rates for both 15 minute and 3 minute appointments. In all cases, the average percent of arrivals is about 82 percent. Table 5. Appointment Lengths by Provider Type in Peds Continuity Clinic Provider Type Residents Faculty %ofl5min appointments 86% 16% %of3min appointments 14% 84% Table 5 shows that residents have a much higher proportion of 15 minute appointments than faculty providers in the Pediatrics Continuity Clinic. Residents are over 5 times more likely to have a 15 minute appointment than faculty OB/GYN Clinic Results from the OB/GYN clinic are based on 132 appointments for nurse practitioners and 35 appointments for MDs. Table 6. Arrivals and No-Shows by Provider Type in OB/GYN Clinic Provider Type Appointment Nurse Prnctitioner MD Length % Arrivals % No Shows % Arrivals %No Shows 15 mm 93% 7% 98% 3% 3mm 1% % 93% 7% 45 mm 1% % 1% % The data for the OB/GYN clinic suggests that nurse practitioners and MDs experience similar arrival and no show rates for 15 minute, 3 minute, and 45 minute appointments. Table 6 shows a high percentage of arrivals for all appointment types. 9

13 Provider Type appointments 77% 69% 1 : %oflsmin %ofómin. % % % % of 45 mm %of3min 1% %of45min % of 3 mm %of2min NP MD Table 7. Appointment Lengths by Provider Type in OB/GYN Clinic C.. : appointments % 2% % appointments % 6% % appointments 51% % 22% appointments 49% 64% 78% % of 15 mm NP MD DO Table 9. Appointment Lengths by Provider Type in Family Practice Clinic Provider Type minute, and 6 minute appointments. percent of the time. MDs are the only provider type to offer 2 minute, 4 minute, 45 Table 8 shows a high rate of arrivals for all appointments lengths and provider types in the Family Practice clinic. In all cases, patients arrived for scheduled appointments over 97 6 mm 1% % 3 mm 98% 2% 97% 3% 1% % 2mm 98% 2% 45 mm 1% % 15 mm 1% % 98% 2% 1% % Length % Arrivals % No Show % Arrivals % No Show % Arrivals % No Show 4 mm 1% % Appointment Nurse Pmcthioner MD DO Table 8. Arrivals and No-Shows by Provider Type in Family Practice Clinic Provider Type 721 appointments for MDs, and 9 appointments for DOs Family Practice Clinic Results for the Family Practice clinic are based on 78 appointments for nurse practitioners, %of4min.... 1% % and MDs in the OB/GYN clinic. Table 7 suggests that all appointment types are divided evenly between nurse practitioners appointments 21% 28% appointments 2% 3%

14 It appears from Table 9 that there is an even distribution of 15 minute and 3 minute appointments for nurse practitioners. Both types of doctors, MDs and DOs, have a higher percentage of 15 minute appointments than 3 minute appointments. 4.3 Telephone Management Telephone Memo Analysis Figure 2 shows the distribution of telephone calls by day of the week. Incoming calls are most frequent on Mondays where the number of phone calls is about 5% higher than the rest of the weekdays. The number of incoming calls is quite constant from Tuesday to Friday. There are very few incoming calls on Saturday, since the clinic does not have regular hours on the weekend. (I) Cu C-) II 25% 28% 15% 18% 5% % S Ion Tue Wed Thu Fri Sat count= 1768 Figure 2. Incoming Calls by Day of the Week Day of the Week 18% 98% 8% 78% -6% -5% -48% -3% -2% 18% -% a) > 4-. E C.) The following table displays the average and standard deviation of the number of calls per weekday. About 18 weeks of data was available for analysis. In order to compensate for the estimated 1 percent of missing memo sheets, the last column in the table represents the approximate average number of calls per day assuming 1 percent of the data was available. Table 1. Summary of Weekda Statistics of Incoming Calls Day AG Calls/Day Standard Deviation Count Max Mm AG 1% Monday Tuesday Wednesday Thursday Friday

15 :: Figure 3 shows the distribution of incoming telephone calls by the time of the day. The clinic is open from 8: am to 6: pm. Incoming telephone calls are the most frequent between 8: and 9: am, and then tend to decrease in frequency as the day progresses. 25% -- 1% 9% 2% 8% - ioio 15% 6% 5% 1% 4% %E 3 Time (hr:min) count= 1571 Figure 3. Distribution of Incoming Calls by Time of Day Table 11 summarizes the statistics associated with the average number of calls per day for one hour intervals. Again, in the last column of the table we estimated the average number of calls per day if a complete set of data was available. Since the data is estimated to include 9 percent of all telephone memos passed on to doctors or nurses, we added 1 percent extra to the calculated average to account for the missing data.. Table 11. Phone Calls Per Day By One Hour Intervals Time of Day AG Calls/Day Standard Devianon Count Max Mm AG + 1 7:-7: :-8: :-9: :-1: :-11: :-12: :-13: :-14: :-15: :-16: :-17:

16 13 C of the following graphs shows a similar distribution, we can generalize the distribution of physician. count= ensure that the separate daily distributions are similar to the overall distribution. Since each calls by time of day for all weekdays by the overall distribution. Because this distribution distribution to reduce complexity in the model. Although the above five graphs are not time of day into each weekday. It is necessary to compare the disthbution of each day to the assumption that the distribution of telephone calls throughout the day does not depend will be added into the simulation at a later date, it is helpful to have an accurate general on the day of the week. identical, they each show a pattern of a high frequency of incoming calls in the early morning with a gradual decrease of calls throughout the rest of the day. Because of this Figure 4 shows the distribution of the number of return calls needed to get in contact with first return call. However, many times the doctor or nurse returning the calls must make multiple phone calls in order to reach the patient. In these cases, either an answering machine picked up, no one answered the telephone, the telephone was busy, or the patient similarity, we feel comfortable recommending that the simulation model be modified with was not available to speak. Figure 4. Distribution of the Number of Return Calls Needed to Reach the Patient Figure 5 shows the distribution of the content of the incoming calls. The most common reasons for an incoming call are general medical advise and making an appointment to see a # of Return Calls % 2%.1- the patient. About 9 percent of the time, the patient is home and answers the phone on the In Appendix I, there are figures which break down the distribution of incoming calls by 1% 1% 4% 2% C.) 3%.1-I 7%. I I I - I % 4 o 3% - 4% - C 5% - C a) 6% -.2! 7% 8% - 9% - 6%.? 5% ca T 1% + 9%

17 75th %ile 4:25 99th %ile 11: 25th%ile 1: max 11: mm :15 standard deviation 2:55 mean 3:1 Table 12. Statistics for Nurse Return Call Length (min:sec) 5% % E 15% 1% 3 25% 2% 35% 4% 4% 1% 2% 14 Figure 5. Distribution of Reasons for Incoming Calls variance is a result of the wide variety of phone calls the nurse returns. Some of the calls count = 1793 minutes) Telephone Time Study The time study at the Pediatric Continuity Clinic resulted in several sets of data. The first Table 12 lists the notable results for the time study. The mean was found to be 3 minutes set of data collected was the length of time the nurse spent actually talking on the phone. and 1 seconds, but the standard deviation is large at 2 minutes 55 seconds. This large were simply setting up an appointment for a patient (as shown by a minimum length of one minute). Other calls, however, were more involved such as giving advice for a sick child or explaining a diagnosis to a worried parent (as shown by the maximum length of 11 C...=, C C. C C C S % 1% 9% 8% 7% 6% 5% 3% c C. 3 E C.., S S E 3- o 3- w Q 3.- w 3- e5 (3 3-1 C Reason for Call -3 o C =.3.3. w S

18 The distribution of this data is shown in Figure 6. Over 5% of the calls lasted under two minutes, however, the disthbution has a long tail accounting for the large variance. The each memo (phone time plus preparation and/or ap up time). To find this data we cumulative distribution function is also shown in Figure J. This distribution follows the estimated the length of actual time the nurse spent on phone related activities and divided it data by rising quickly and then leveling off for the majority of the intervals. The second value evaluated by this time study was the length of time the nurse spends on Table 13. Results of Time Study of Phone Related Work 11 the four data sets are shown in Table 13. Data sets 1, 3, and 4 have times per memo that by the total number of memos completed during the period studied. The results for each of are extremely close in range. Data set 2 may be different due to differences in the rate of 1% 3% 25% 2% 4 4: 2: :54 Total 14: 8:6 47 1: : 2: :48 2 3:3 2: :3 = N = = = I( In I5 a a a %.8% 1.% 5% 15% 1.8% 9.% 8.8% 7.% 6.% 5.8% 3.% 2.% Figure 6. Length of Nurse Return Calls N a a i i i i I i 1 3:3 1:3 7 12:54 (Iu:min) Phone (hr:min). (min:sec) Data Set Length of Time Study Time Spent on # of Memos Avg. Time per Memo Time (min:sec). á a t4 Lfl Lfl In In In In In a a a a a a a a a a average length of time a nurse takes to work on a memo is around 1 - minutes. incoming calls or employee variation. Overall, in 14 hours of study, we found that the _ -. N I I I 4.% =

19 5% 15% 1% 35% 3% 4% 16 25% 2% C, C, C, WE 45%- Pediatric Clinic or Med-Peds Clinic. We were concerned about the small sample size of the Figure 7 shows the distribution of MA assists by clinic. Raw data for the MA time study is located in Appendix K. For the week data was collected, no assists were performed in the data collection because we expected a higher number of assists. After meeting with Nancy themselves. Eggerer, she felt confident that the MAs collected accurate data and recorded all assists during the week. She was not alarmed by the small sample set and confirmed that the data Figure 7. Distribution of Medical Assists by Clinic count = 18 Clinic O8/GN Family 1% 2% 3% = 9% 8% 4.4 Medical Assistance Practice Peds Med P eds 4% - represented actual needs for the week. During the week, providers were present with the MAs during 72 percent of the assists, while the MAs performed 28 percent of the assists by 5% g

20 Date - Practice - Family OB/GYN Total 17 Time for flssist (min:sec) C = I 1W 3% Assistance 15% 1%.% 5% 25% 2% 5.% 1.% 35%. 9.% 6.% 3.% 2.% E = 4.% I 4% IM 1.% 8.% 7.% mean =9 standard deviation =7 Figure 8. Distribution of Medical Assist Times count=18 %- H C, C, C, Figure 8 shows the disthbution of MA assist times. All of the assists are under 2 minutes except for the one assist time of 35 minutes. % of Appts. Requiring 2% 4% 6% 4% need for unscheduled MA assistance was low. utilized MAs for unscheduled assistance more than the other clinics. However, the, overall In Table 14 the daily frequency of each clinic s MA assists is shown. The OBIGYN clinic 2-Nov Nov Nov 1 Total Appts. for Week * LI, C. LI, La C LI, 17-Nov Nov Table 14. Freouency of Daily Medical Assists by Clinic C. LI, C LI, 1

21 Internal Medicine OB/GYN any number in the specified range had an equal likelihood of being picked as the process time. After gathering data and preparing a cumulative distribution function of intake program is to defined data distributions. The old model used a uniform distribution 4.5 Model Experimentation Overall Family Practice Clinic AG Time (mm) St. Dev. (mm) Count Max (miii) Mm (mit Table 15. Statistical Analysis of Assist Time by Clinic Experimenting with the modified simulation model allowed us to examine how sensitive the between 3.5 and 5 minutes to define the intake process function. With this disthbution, process times, we defined a new distribution and used it in the modified simulation model. Figure 9 shows the differences in average patient length of stay depending on appointment Figure 9. Comparison of Average Lengths of Stay by Appointment Types The new distribution had a higher likelihood of lower process time, so we would expect results of 5 replications (of 12 hour days) are shown below in Figures 9, 1, and 11. This new distribution was defined as a continuous function with a mean 3. minutes. The length times. Detailed results of the simulation dat4 can be found in Appendix D. lower wait times in the new model. The results of the model experimentation verify this Table 15 summarizes the average assist time for each clinic and the associated variance. hypothesis. Every appointment type experienced a shorter length of stay in the modified DOId Distribution Modified Distribution a ci) c Cl) 4 6 C) 2 -J 3 - Appointment Type 15 mm 3 mm 45 mm

22 19 The results of patient length of stay are very close regardless of appointment type. This concerned us and needs to be examined. After reviewing the simulation flow chart (see Appendix A) we found some possible reasons for the similarities of results. In the model, second encounter, respectively. These second encounters average.1 minutes for 15 minute appointments, and 28.1 minutes for 3 minute appointments. This likelihood and appointments that doesn t occur in 45 minute appointments. Also, while the mean of duration of second encounters adds additional time to 15 minute and 3 minute necessary provider time in 45 minute appointments is higher than the other appointments, likelihood for the provider process time to be very low. Figure 1 shows the statistical results for other variables concerning patient wait time. This figure shows a comparison of values for Average Wait for Provider, Average Waiting Room Time, Max Wait Time for Provider, and Max Waiting Room Time. distribution than in the old distribution. With the modified model, the Average Walt Time Figuie 1. Có pnof Average Waiting Times for Provider decreased 11.7 percent, the Average Waiting Room Time decreased 14.1 percent, the Max Wait Time for Provider decreased 1.8 percent, and the Max Waiting Room Time decreased 6.6 percent. Provider for Waiting Time Avg Wait Max. - 1 E3 C) percent model. The average decrease in length of stay for the three appointment types was about 45 minute appointment types never require a second encounter during their stay. 15 minute and 3 minute appointment types have an 8 percent and 6 percent likelihood of requiring a the distribution is skewed further to the left than the other distributions. This results in a Again, the average values of these variables are consistently lower in the modified c Old Distribution Modified Distribution Time 5 IRoom Avg Room Provider Waiting for Max Wait I I

23 same. The confidence intervals for these variables (see Appendix D) are quite similar for the two models providing evidence that the differences in results are not very significant. 1 L This figure shows the average percent utilization for each provider in the two models. The affect of the modified disiribution on employee utilization rates are displayed in Figure Percent utilization is defined as the amount of time that workers are attending to work. Figure 11. Comparison of Utilization of Providers utilization of Providers 1, 2, and 7 rose slightly, the percent utilization of Provider 4 decreased slightly, and those of the nurse practitioner and Provider 6 basically remained the It appears that the modified disthbution doesn t have as significant an effect on employee utilization rates as it does on variables considering patient wait times. While the percent z I. 2 [OId Distribution Modified Distribution % 4%- 2%- 6%- 7% 9% 86% 88% 1 % t/ I / = 3%-.2 5%- 8% as Provider 5, and Provider 3 is not scheduled for the day of the week we examined. Providers 1,2, 5, 6, and 7 are doctors. In the simulation, the nurse practitioner is defined > I.-. ci) > > c. J L. ci) > C Co ci) > I. ci)

24 21 Experiments with the MedModel software showed that the simulation was quite sensitive to data distributions in the model. We found that small changes in a single process time distribution had significant effects on several key variables. To validate the use of the simulation model, the model must first reflect the results experienced in real life. Then the model can be used to correctly estimate the impact of different changes. In order for the distributions are based simply on estimations. While these estimations may provide general adequately reflect real life. One reason we feel that the current results inadequately estimate real life was the small and disadvantages of overbooking must first be considered before implementing such a Once the model is an accurate simulation of actual clinical processes, it will be a useful tool to determine actual patient lengths of stay to compare real life values with the simulation model to approach the actual results experienced in different clinics, close attention must be paid to the data distributions used in the model. In the current model, some key guidelines for process times, they may result in output statistics that are unrealistic. In order for the simulation to be a valuable tool, it needs to adequately represent real life. Process time distributions need to be continuously refined so that the model results difference in patient length of stay for varying appointment types. Studies should be done estimations. If there is a large disparity between the sets of numbers, the model needs to be modified to account for the differences. utilization rates. Then steps could be taken in real life to adjust the process distribution that was changed in the model. Also, the model could be used to reflect the results of changes will only be a reliable and valuable tool if it can show estimates that currently are indicative The results of the scheduling data showed differences in scheduling policies and arrival sensitive to data distributions, assumptions that different clinics experience the same of real life results. More work is needed to get the model to that point. in staffing plans such as replacing a doctor with a nurse practitioner. However, the model 5.1 alidation of Simulation Model for improving current systems. For instance, if a current problem in a clinic was low 5.2 Appointment Analysis provider utilization, relevant distributions could be changed in the model to increase the rates in different clinics. TI these clinics are to be modeled with the simulation, it is with the implications of the assumptions that are made. important that these differences are reflected in the model. Since the model is quite process time distributions may lead to inaccurate result estimations. If the same distributions are to be used in different clinics, the simulation user must be comfortable The scheduling procedure for residents and faculty needs to be examined in the Pediatrics residents need to understand how they are being utilized. provider according to the results we tabulated from the scheduling data. The advantages Continuity Clinic. The results of this study should be compared with scheduling guidelines used in the clinic. There should be strong reasons for the disparity of appointment lengths by provider type if such a difference in scheduling rates is acceptable. Faculty and The rates of no shows that we determined could be incorporated in the scheduling system. For instance, the Pediatrics Continuity Clinic could overbook their appointment types by change. Other clinics, such as OB/GYN and Family Practice, should be careful in 5. Recommendations

25 The telephone management study at the Pediatrics Continuity Clinic showed that the and other functions before calling the next patient. A direct transfer system would require A direct transfer approach to telephone management may lead to some possible problems, the nurse to put aside what they are doing to respond to the patient s call immediately. This the calls they are required to return. Some current categories of calls don t require a in the Northeast Ann Arbor Clinic. Because of the infrequent need for unscheduled MA to make an informed decision. Another problem with this system would be the problem of 5.3 Telephone Management System incorporated in the simulation model arrival rates for the individual clinics. One way to potentially improve the current system would be to implement more of a direct overbooking because of the low rate of no shows. These rates of no shows should also be amount of time nurses spend on the phone is small compared to the paperwork time associated with a call. If the paperwork time could be reduced, the nurses could respond to patient s needs much more effectively. transfer approach to phone calls. Instead of the clerk making a memo for the nurse to return pertinent calls, it would be quite feasible for the nurse to answer the call directly. This is possible because the nurse s phone line is rarely in use. A direct transfer approach might be successful if the time required for nurses to document necessary information could be decreased. A possible solution is keeping patient information on-line (i.e., computerizing patient information)/ If records and forms were kept in a computer database, incoming phone call. Computerized forms would also decrease the time needed to record information, and would reduce excessive paperwork throughout the clinic. prior task. Once they have called the patient, the nurse completes the necessary paperwork nurses could quickly access patient information and respond to their needs during the initial however. It might be more difficult for nurses to respond to a patient s question since they have not studied the patient s medical history prior to the phone call. The nurse would have to quickly access the computer records to find and read medical histories if necessary could cause the nurse to lose concentration on tasks and potentially forget important The results of the MA utilization time study showed that unscheduled MA assistance is rare interruptions. Currently, the nurse responds to a patient call when they have completed a information to record on the patient s computer record. Hopefully, the advantages in transfer approach. Another way to reduce demands on the nurses in the clinic would be to reassign some of nurse s knowledge and could be handled by clerks. For instance, calls regarding immunizations forms, referrals, and other documents reassigned to the clerks. This would 5.4 Medical Assistance Analysis efficiency and utilization would outweigh the potential problems associated with the direct increase the amount of time nurses are able to attend to patient calls. assistance, including this utilization in the simulation model may not be necessary. If the data is to be folded into the model, the differences between clinics should be included. Different clinics in the Northeast Ann Arbor Clinic experienced different assistance needs according to our data, so different distributions are needed. While there is a need for unscheduled assistance from MAs, it does not appear that these tasks create an excessive demand on the MAs time. We would suggest that the model does not need to require the integration of these employee roles that occur on average less than 4% of the time in the 5 clinics we studied. Further study may be needed to verify this recommendation. 22

26 Appendices C

27 Iosf,, sopl, P. Ku 1 kndq G4)lIia,,y C, C C 2% to5m) IgoestoRmR oomp. after ecam Dr: 9% Nota 1: (mm; a1; 99% rnaic) S I Ii )WA )$ ç iwi

28 212mm Ea.tinan Pontts Joseph P. Kent p1 iuiip1l Internal MCdICI atlent Flow. 15 nm Appointment 3 mm Appointment 45 mm Appointment lime between is? limo botweon 1st limo between 1st and 2nd encounter and 2nd encounter and 2nd encounter (,1 7.8,4 ijmin (1 4.6,51?)mln 15 mm pporntm] 3 mm Appomnlm] 45 mm Appointment Provider 2nd Provider2nd Provider 2nd oncounterw/palient oncounterw!patient oncountorwpatient (j,8321?)min (4,l4.3,22?min N rpounenvsic. rcqulrod? Y9% Nursin Service RN: treatment and oducation pl mm N SocialWo,IcI lioniequt Ycl% Social Work/Nutxition Note: Since <1% Social Work! Social Worker or Nublliorr will not be Nutritiondoneth inckzdod inthe model. in MP counsel rm. PiChecksout Note t:frntn;avg;99%mac) Paqo2 I ll I O )(l 11 Ill IfljtI Ko1ik Comins

29 Appendix B. Patient Check-In Process Flow Chart Patient Check-In / Paperwork Process: Northeast Ann Arbor Health Center I Patient hands clerk insurance and hospital cards Clerk processes necessary paperwork

30 and signs memo notes on memo - RN writes and places call -. memo to RN memo and chart problem patient and finds No Clerk writes RN talks chart with memo RN resolves when not busy with RN returns to desk and attaches Clerk pull chart Yes to caller on RN s desk Clerk places Is an RN available? Yes needed to answer Clerk anwers End IsanRNorMD No phone call question? question transfers call Incoming Clerk memo Pediatric Continuity Clinic Start Telephone Call Flow: Appendix C. Telephone Management Flow Chart

31 rovider7 % In Use vider6 - % In Use wider 4 % In Use Provider 1 % In Use Provider 2 % In Use Provider 3 % In Use Pct_Pts_Seen_w_in_lojnin_ Avg alue NP % In Use Max_Wait_Time_ Avg alue Avg_Wait_Time_ - Avg alue Avg_Tinie_to_See_Prvd_ - Avg alue Max_Time_to_See_Prvd_ Avg alue Avg_LOS_3_ - Avg alue Avg_LS_15_ Avg alue IPLE. REPLICATION SUMMARY Avg_LOS_45_ - Avg alue C:medmodeIoutputum91 3b.mrs - 3b.mrs c:medmodei\output\um Provider 1 % In Use Provider 2 % In Use Provider 3 $ In Use 5 Provider 4 $ In Use Provider 6 % In Use Provider 7 % In Use Pct_Pts_Seen_w_in_1O_min_ Avg alue NP % In Use Max_Wait_Ti!ne_ Avg alue Max_Time_to_See_Prvd_ Avg alue Avg_Wait_Time_ Avg alue Avg_Time_to_See_Prvd Avg alue Avg_LOS_3_ Avg alue Avg_LOS_45_ Avg alue Avg_LS_3_ Avg alue Avg_LOS_iS_ Avg alue Data for: Sorted Data Avg_LOS_15_ Avg alue Avg_LOS_45_ Avg alue Statistics for: Reps Mean Median Std ev Std rr Skewness Appendix D.. MedModel Output Using Current Distribution Avg_Wait_Time_ Avg alue Avg_Time_to_See_Prvd_ Avg alue Confidence Intervals for: 9% 95% 99%

32 NP % In Use Max_Wait_Time_ Avg alue Pct_Pts_Seen_w_in_1_min_ Avg alue Provider 2 % In Use C:\medmodefloutputum91 3b..mrs - c:\medmodeioutputum91 3b..mrs : Provider 1 % In Use Max_Time to See Prvd. Avg alue Page Provider_4 % In Use Provider 3 % In Use

33 Page3 - C Provider 7 In Use Provider_6 % In Use C:medmodeIoutputum91 3b.mrs - c:medmodeioutputum91 3b.mrs :

34 95% Statistics for: Reps Median Std ev Mean C IPLE REPLICATION SARY C:medmode1outputum481.mrs - c:\medmode1outputum481.mrs ji % Std Err Skewness Sorted Data 9% Max_Time_to_See_Prvd_. Avg alue Avg_Wait_Time_ Avg alue Avg_Time_to_See_Prvd_ Avg alue Avg_LOS_4 5_ - Avg alue Avg_LOS_3_ - Avg alue Avg_LOS_15_ - Avg alue Avg Wait Time Avg alue Avg_Time_to_See Prvd_ Avg alue Avg_LOS_45_ Avg alue Avg_LOS_b_ Avg alue Avg_LOS_15_ Avg alue Data for: Provide r_7 $ In Use )vider6 $ In Use ( vjder4 % In Use ovider_3 % In Use Provider 1 $ In Use Provider 2 % In Use?ct_Pts_Seen_w_in_1_min_ Avg alue NP % In Use Max_Wait_Time Avg alue Avg_Wait_Time_ - Avg alue Avg_Tizne_to_See_Prvd_ Avg alue Max_Tirne_to_See_Prvd_ Avg alue Avg_LOS_3D_ Avg alue Avg_LOS_15_ - Avg alue Avg_LS_4 S - Avg alue Confidence Intervals for: Previder_7 % In Use Provider_4 % In Use Provider_S % In Use Provider_2 % In Use Provider 1 $ In Use Provider 3 $ In Use?ct_?ts_Seen_w_in_1_min_ Avg alue N? % In Use Max_Wait_Time_ - Avg alue Appendix D. MedModel Output Using Revised Distribution

35 Max_Wait_Time_ Avg alue Max me_to_see_prvd_ Avg alue Pct_Pts_Seen_w_in_1_min_ Avg alue NP % In Use Provider 2 In Use Provider 3 % In Use Provider 4 % In Use C:medmodeIoutputum481.mrs - c:medmodei\outputum481.mrs Page2.. O Provider 1 % In Use

36 Page Provider 6 % In Use Provider 7 % In Use C:medmodeIoutput\um481.mrs - c:medmodefloutputum48 1. mrs

37 Tijjie Patient Time Patient 8:14: 8:15:55 :1:4 -Oct 8:3: 8:33: :3: -Oct 8:31:55 8:34:4 :2:45 2SOct 8:39:5 8:39:5 ::45 8:51:45 :2:15 -O 8:53:35 8:54:45 :1:1 28 -Oct 8:54:15 8:55:5 :1:35 -Qct 8:55:45 9::5 :4:2 -Oct 8:57:15.9:2:2 :5:5 Qct 8:58:25 9:3:5 :4:4 -Oct 9:14:5 9:16:45 :1:55 -Oct 9:16:2 9:18:2 :2: 9:23:45 :3:5 2S-Qt 9:3: 9:31:1 :1:1 9:45:25 9:46:3 9:58:15 1::5 :2:35 1 :6:25 1:8:5 :1:4 -Oct 1:14:45 1:16:1 :1:25 Oct 1:25:5 1::2 :1:15 -Oct 1::1 1:3:4 :4:3 1:3:45 1:32:15 :1:3.O 1:4:5 1:44:1 :3:2 -Qct 1:44:45 1:47: :2:15 1:55: 1:58:35 :3:35 11:7:5 11:13:3 :6:25 -Qct 11:6:1 11:11:2 :5:1 11:7:5 11:13:3 :625 -Oct 11:15: 11:17:3 :2:3 Oct 11:19:5 11:21:45 :2:4 11:21:3 11:24: :2:3 11:22:15 11:23:5 :1:35 -Oct 11:3: 11:33:5 :3:5 -Oat 11:32:3 11:36:45 :4:15 -Oct 11:34:35 11:37:45 :3:1 -Oct 12:23:15 12:25:5 :1:5 12:39:4 12:41:5 :1:25 -O 12:46:5 12:47:55 :1:5 -Oct 12:55: 12:56:4 :1:4 25 -Oct 12:59:5 13:1 : :1:55 -Oct 13:7: 13:1: :3: 13:11:2 13:14:2 :3: 13:19:45 :3:25 13:16:45 13:21:2 :4:35 13:21:15 13:22:3 :1:15 13::4 13:29:4 :3: -Oct 13:27:5 13:33: O:O5:io 13:31:55 13:37:3 :5:35 13:3: 13:41:5 :6: -Qct 13:36:4 13:44:25 :7:45 -Oct 13:41:5 13:44:55 :3:5 13:45:15 13:5o.1 :5:Oo -Qct 13:47:2 13:53: :5:4 13:52:35 13:59:3 O:O5: Oct PP 13:53:15 13:59:2 :6:5 PP.13:54:55 14:2:1 :O7:i Appendix E. Raw Data of C1leck4 Process Time Study

38 16:68: :28:5 -Oct 14:39:3 14:42:25 :2:55 -Oct 14:54:2 14:58:46 :4:25 -Oct 15:4:25 15:6: :1:35 -Oct 14:32:35 14:35:3 :2:55 -Oct 15:2:25 16:3:4 :1:15 -Oct 15:2:45 15:4:15 :1:3 -Oct 15:5:3 15:7:2 :1:5 -Oct FP 14:8:15 14:12: :3:45 31-Oct 3W. 16:2:41 :4:6. 31-Oct JW 14:2:49 14:6:37 :3:48 31-Oct 3W 14:6:2 14:8:45 :2:25 31-Oct 3W 14:8:58 14:1:15 :1:17 31-Oct JW 14:23:15 14:25:45 :2:3 31-Oct 3W 14:25:56 14:28:54 :2:58 31-Oct JW 14:29:8 14:31:16 :2:8 31-Oct JW 14:4:12 14:41: :1:14 31-Oct 3W 14:56:16 14:57:57 :1:41 31-Oct JW 14:58:25 15::19 :1:54 31-Oct 3W 15:5:54 15:8: :2:6 31-Oct.JW 15:6:58 15:1: :3:28 31-Oct JW 15:22:5 15::12 :3:22 31-Oct JW 15:2:1 15:22:14 :2:4 31-Oct 3W 15:32:16 15:34:38 :2:22 31-Oct 3W 15:39:45 15:42:48 :3:3 31-Oct 3W 15:53: 15:56:1 :2: Oct.1W 15:56:34 15:58:23 o:oi: 31-Oct 3W 13:44:2 13:46:38 :2:18 31-Oct JW 13:46:32 13:49:58 :3: 31-Oct JW 14:37:47 14:4: :2:13 31-Oct JW 14:44:5 14:47:46 :3:41 31-Oct Jv 15:11:2 15:13:39 :2:37 31-Oct JW 15:27:11 15:29:2 :2:9 31-Oct 3W 15:29:4 15:31:2 :1:22 31-Oct 3W 15:44:34 15:47:9 :2:35 31-Oct 3W 13:23:58 13:27:24 :3: 31-Oct JW 13:28:35 13:33:52 :6:17 31-Oct JW 13:32:3 13:37:2 :4:32 31-Oct JW 13:4:35 13:44:14 :3:39 31-Oct 3W 13:19:2 13:2:56 :1:36 31-Oct 3W 13:14:21 13:18:16 :3:55 31-Oct JW 13:15:17 13:18:5 :2:48 31-Oct 3W 13:12:2 13:16:48 :4:28 31-Oct.JW 12:49:44 12:54:2 :4:36 31-Oct Jv 12:55:45 12:58:27 :2:42 31-Oct f 12:59:55 13:1:55 :2: 31-Oct JW 13:1:1 13:3:15 :2:14 31-Oct JiN 13:2:1 13:1)7:35 :5:25 31-Oct JW 12:48:16 12:52:16 :4: -Oct. 16:32:36 :4:3 31-Oct JW 12:42:2 12:43:25 :1:23 -Oct 15:23:45 15::35 :2:5 -Oct 15:27:5 15:29:15 :2:1 -Oct R 15:16:5 16:18:45 :1:55 -Oct 15:7:35 15:8:55 :1:2 -Oct 15:12:3 15:14:15 :1:45 -Oct 15:15:15 16:18:45 :3:3 -Oct 15:2: 16:24:1 :4:1 -Oct 14:18:45 14:21:2 :2:36 -Oct 14:17:3 14:19:45 :2:15 -Oct FP 14:25:5 14:27:1 :1:2 -Oct 14:29:4 14:33:35 :3:55

39 n

40 C Prouider Tgpe OB/GYN Appendix F. Raw CLS Data 3mm mm mm Length Arrival No Show Arrival No Show Appointment Nurse Practitioner MD 3 mm 41 I 1 I 135 I mm Length Arrival No Show Arrival No Show Appointment Residents Faculty Pro uider_type Pediatric Continuitg Clinic 6mm 6 45mm 5 Appointment 4 mm 1 7 Length of 3 mm mm mm Appointment Nurse Practitioner MD Lenath Arrival No Show Arrival No Show Arrival No Show Provider Type FamEig Practice

41 Date: MA Name: Program and Operations Analysis Assistance Completed o (YIN) present Time Room Cilnic (Check One) Begin Assistance E Present? Left, if Entered to Time Was Provider Time Provider Shift: University of Michigan Medical Centers Medical Assistant Reluest Data Collection Appendix G. Medical Assistance Time Study Data Form

42 . By :. Keg an X next to selected appointments then PF9: T R S fl 11/16/95. ye tu CD (Ti CD O (Ti (Ti CD CD : 1JJS 1 girioiiiieehi CLI[NC SCHED 116 SYSTEM 11//95 89 Reso.,ce HEIICIIB DR. CIIRISTIIHI II BUYE Date Time Status X Date FromTo CPI 8 Uisit Patient Mame Contact Phone p f a t 11/16/ H SAREIIULLAH, RU SF II 11/16/ H IITAL, BhllIHINtIE flu SF II 11/16/ H MILLER, KYLE A RU SF A H HAIR, IIE1II.T RU SF A 11/16/ H DONHAIl, JIISOH RU SF II 11/17/ H SCOTT, KUIIME 11/17/ N IIELIEHS, TREUOII RU SF II NP SF A 11/17/ THORNTON, STEP 11/17/ H KIll, YEONIIIEE RU SF II HP SF II 11/17/ N BICKLEY, flue RU SF II 11/17/ H KIRK, JEHHIFER 11/17/ KIRK, ADIGAIL HP HP SF II SF A 11/17/ H ESTIICID. MARIE RU SF 11 11/17/ N RICHISSIH, CAL RU SF A 11/17/ H CHrn1OELL, CAHD RU SF 11 IIACK=PF7 HEXT=PFO CLSRSDSP User Id 19? Termid 1114 Function CLDS

43 N C) 1 ( ( r. o ( R 1 F 5% 7% 25% 8% 3T 2%+ I 9% I 6% c l I ( I 1 o o a, O ) ) 16% 14% 6% 4% % % 2%. J :. 18% 12% 1% 2% 4% bl IIj ii i: i 6/o I%1 1% Time of Day 4% 1% Monday Incoming Calls by Time of Day Taubman Center- Pediatric Continuity Clinic Telephone Call Distribution T7% Time of Day 666 o q q c if.- c1 c,. Ifl ( o o o o o / i; 1% :x 4 i 1: 2% A 3% E1.. 4 C.) t 1 I,.l AI E I:cl Il tl j 1/ 5% : ri : : fl :4 J ll 7 9% Taubman Center - Pediatric Continuity Clinic Tuesday Incoming Calls by Time of Day Telephone Call Distribution O%Lr h H 2% 1/.L 1 \ fl I -.1 ci % I % o ti E % -- Appendix I. lncoming Calls by Day of Week and Time of Day

44 :: 3% Ci 1%.6 4% 2 o 15% 6% 7% 4/ 7% 25% 2% 2% Fl 8% 9% -x 1% 9% 8% 6% 4% 3% 2% 1% 5% C., E C.) a 15% Taubman Center - Continuity Clinic a a Pediatric Telephone Call Distribution Wednesday Incoming Calls by Time of Day Time of Day i I 5% Taubman Center- Pediatric Continuity Clinic -IfTlOO% Telephone Call Distribution Thursday Incoming Calls by Time of Day Time of Day %ITrij1I[:iI[ I 1 1. a > a

45 25% 1% Time of Day (hr:min) 1% A 2%,/% 8%. 9% Friday Incoming Calls by Time of Day Taubman Center- Pediatric Continuity C inic Telephone Call Distribution

46 : Call /95 12:13:15 12:13:5 :: referrals 1 1/ :39:55 12:41:4 :1: perscnptions 1 1/1 4/95 13:44:15 13:55: :1: immunization forms 1 1(1 4/95 1 4:4:5 14:8:5 :3: Crippled childrens forms /95 14:36:4 14:42: :5: other people (homecsre, Schools) 11/14/95 12:9:3 12:14:4 :5: /14/95 14:2:2 14:27:45 :7: Other forms Date Began Ended Tim. Elapsed. of Call I.1 general advice Tim. Call Tim. Call Content/Nature Categories. 11/ :3:4 12:9:1 :5: appointment 11/21/95 9:24:3 9:25: ::3 1 11/21/95 1:1 2:3 1:19: :6: /21/95 1:32: 1:34: :2: /21/95 11:15:45 11:21: :5: /21/95 11:23: 11:34: :11: /21/95 11:36: 11:37:3 :1: /27/95 8:56:1 8:58:1 :2: /27/95 8:59:25 9::24 :: /27/95 9:18:2 9:21:6 :2: /3/95 8:57: 8:58:5 :1: /3/95 9:9:25 9:1:2 :: /3/95 9:35:55 9:37:5 :1: /3/95 9:47:1 9:48:5 :: /3/95 9:55:5 1:4:1 :9: /3/95 1:24:15 1:3:15 :6: /3/95 1:59:55 11:2:2 :2: /3/95 11:14: 11:17:1 :3: /3/95 11:28:5 11:29:1 :1: /3/95 11:31:3 11:34:35 :3: /3195 1:6: 1:13:5 :7: /95 11:5:4 1 1:1 :1 :4: :56:25 1:57:2 :: /27/95 9:13:31 9:14:5 :: \ 11(21/95 11:41:15 11:51: :9: ( 11/21/95 11:39: 11:41: :2: /95 9:25: 9:27:15 :2: /95 9:44: 9:44:15 :: / :59: 1:1: :11: / :12:45 11:14:3 :1: /95 11:14:58 11:16:21 :3: /27/95 1:34:4 1:38:45 :4: /27(95 9:52:34 9:56:57 :4: /14/95 14:43:15 14:44:4 :1: How to access system 11/21/95 9:13:3 9:17:3 :4: test results 11/21/95 9:42:3 9:44: :1 : /21/95 9:45: 9:46: :1: /21/95 1:19: 1:27:3 :5: /21/95 1:37:15 1:4: :2: /21/95 1:41:15 1:42:3 :1: /21/95 1:43:3 1:44: :: /21/95 1:44:15 1:46: :1: /21/95 1:46:45 1:47:45 :1 : /21/95 1:48: 1:51: :3: /21/95 11:14:45 11:1 5:3 :: /21/95 11:37:45 11:38:3 :: /21/95 11:51:15 11:51:45 ::3 1 11/27/95 :1: /27/95 :1: /27/95 9:8:39 9:1:8 :1: /27/95 9:11:48 9:13:25 :1: /27/95 9:17: 9:23:15 :6: /27/95 :3: /27/95 9:43:45 9:45: :1: (27/95 1:28:17 1:3:4 :2: /3/95 9:4:15 9:6:5 :1: /3/95 9:37:4 9:38:35 :: /3/95 1:22:1 1:22:35 :: /3/95 1:31:2 1:4:3 :9: /3/95 1:53:5 1:54:2 :1: /3/95 11:18:1 11:19: ::5 1 5 Appendix J. Peds Continuity Clinic Phone Time Study

47 2-Nov 2:35 2:4 :5 4 A,C Y 2-Nov 2:5 2:55 :5 4 A,C Y 21-Nov 14: 14:8 :8 4 C Y 2-Nov 3:1 3:18 :8 4 A,C Y 21-Nov 1: 1:1 :1 4 C Y 2-Nov 16:45 16:55 :1 2 C 21-Nov 11:15 11:27 :12 4 C 2-Nov 15:25 16: :35 2 SIGNOID 16-Nov 15: 15:3 :3 1 PEAKFLQW N 18-Nov 1: 1:3 :3 1 PEAKFLOW N 18-Nov 1:8 1:11 :3 1 PEAKFLOW N 16-Nov 1:42 1:47 :5 2 C Y Date Begin Assist End Assist Elapsed time 5=Peds) C=chaperone) Present? Left 4OB/GNY, (A=assist, Provider Time Prov. 21-Nov 11: 11:8 :8 4 C 22-Nov 2:4 2:58 :18 2 A Y 16-Nov 14: 14:1 :1 2 PFT N 16-Nov 2:3 2:4 :1 2 SPIRCMECTY N 16-Nov 16: 16:1 :1 2 C Y 16-Nov 3:32 3:46 :14 2 A 3:45 3..Med-Ped, Reason Prac, 2.IM, Clinic (1=Fam Appendix K. Medical Assistant Raw Data