Simulation Based Approach for Improving Outpatient Clinic Operations

The aim of this study is to suggest the optimum number and schedule of doctors at the OPD (Out-Patient Department) of Gastrology of a hospital in Pakistan. In order to achieve this aim, the discrete event simulation model is developed to minimize waiting time of patients. Data is collected for one week from the OPD; Data collection variables are arrival and service rate of patients, their salaries/income, patient‘s OPD fee, doctor’s charges/patient, service time of patients at each of service channel i.e. reception, triage and doctors’ cabin. Stop watch is used for recording the service time of patients. Input analyzer is used to reveal the distribution of the data. Rockwell arena software version 14.5 is used to model and simulate the queuing system of the outpatient department. Scenario analysis is conducted in four scenarios; in each of the scenario doctors were assumed to be seated for one additional hour. During the period of data collection, it is observed that most of the patients are coming with an appointment of doctors therefore, it is not justified to suggest the hiring of new doctor; especially when patients are coming for the particular doctor; therefore, already available doctors are suggested to be seated longer in the OPD; that is the way to serve the maximum number of patients in the virtual queue of patients that has been kept waiting for having an appointment and for their turn to see the doctor.


INTRODUCTION
P rovision of the good medical care has always been expected from the healthcare providers so that the population can be served effectively and efficiently. Healthcare service improvement is one of the important and urgent issue for Pakistan [1]. One of the essential role of healthcare systems is the improvement of health of the citizens of country [2]. It is essential for the people to have good health for the development and improvement of economic state of the country. In the comparison of service deliveries, healthcare care delivery is a kind of delivery in which the customer is more involved in the process of consumption. The patient can be harmed by the provision of bad service delivery and may lead to loss of life. In this regard, satisfaction of patients is necessary to be investigated for the sake of improvement in healthcare delivery system [3]. The side of patient satisfaction has been targeted attentively in the health care recently [4]. Patients' satisfaction is more important Authors E-Mail: (12in10@student.muet.edu.pk, sonia.irshad@faculty.muet.edu.pk, saad.memon@faculty.muet.edu.pk, anwaruddin.tanwari@faculty.muet.edu.pk, muetanian05in04@hotmail.com) * Department of Industrical Engineering & Management, Mehran University of Engineering & Technology, Jamshoro, Pakistan. to produce retaining customers/patients [5]. Patient satisfaction and the service quality are in a close association, good service delivery yields the retaining customers [6]. According to Suryadana [6], service quality is the difference between actual and perceived service delivery quality by the customer.
Outpatient and Emergency Departments are the most visited departments by the patients and they are also the initial contact of patients with the hospital's staff [7].
Literature indicates that in the past decade of developing counties, emergency departments are stressed on congestion and its influence on service time; furthermore, hospitals were paid attention in terms of their ability to meet the emergency needs [8]. Because of the lacking on the side of control over the customer services, the capacity planning can be complicated by service demands in the particular department. Research shows that Pakistan is also facing shortage of doctors which may be the reason of long waiting lines [9]. Nowadays, Congestion of patients in OPDs, ICUs (Intensive Care Unit), emergency departments and waiting areas is the big problem.
Because of the congestion, planned queuing system do not work effectively. In daily life, queue is basically the common occurrence [10] e.g. at Hospitals, superstores, petrol pumps and CNG (Compressed Natural Gas) filling stations. When the available resources are less than entities then the queue is formed. The term delay is defined as the difference in demand of service and the associated available capacity of meeting the demand [11].
Waiting cost is associated with queues, when the patients wait in the system for getting service [12]. Waiting for long the queue often has negative influence on the satisfaction of the patient [13]. Because of the long waiting lines doctors are put to stress and due to the congestion they try to get the patients free without any deep examination, which yields customers dissatisfaction [2,[13][14]. Dissatisfaction incurs cost to the organizations i.e. cost of customer dissatisfaction.
In chain of health care services, hospitals are considered as one of the most important links which have impact on lives of the people. Hospitals have major influence on the prevention of diseases, early diagnosis of the diseases, management and treatment of patients [8]. The major area of focus for the hospitals is the occupancy and discharge of patients; so that the executive capacity of the system can be managed. A number of research studies are available on the subject of reforms in healthcare but an explicit frame work is required in which the performance can be judged and quantified against the performance management system [15]. Operation research and the management sciences are the subjects by the help of which managers can plan and manage the resources effectively [16].Queuing theory is the tool of operations research which is widely used for simplifying the problems of waiting lines. Mathematical approach such as queuing theory is only possible to be used when the queuing system already exists, then after the collection of data and calculation of performance measures the optimum queuing can be suggested. In contrast to the mathematical approach, simulation is more preferable. Most of the decisions are made on the results of simulation. It has been considered as more effective and more precise as compared to the mathematical model based calculations; because, in complex scenarios assumptions can be changed and behavior of the model can be observed in simulative analysis [8].
This study is aimed to suggest the optimum number/ schedule of doctors at the OPD of Gastrologyat hospital of Karachi, Pakistan.

LITERATURE REVIEW
Several studies have been conducted to improve the hospital performance. Most of the studies focus on simulation optimization methods to reduce waiting time.
Agyei et. al. [10] developed discrete event simulation Arena software was used for simulation and analysis.
They indicated that 179 beds should be there instead of 81. Furthermore, the patients having high degree of illness were given the priority. Table 1 summarize the literature.
In this study not only waiting time of patients was focused but the increased percentage of served patients was computed along with the service cost, revenue and the profit so that the operational managers have enough information to improve the queuing system at the OPD.

Data Collection
OPD of Gastrology was selected for the data collection.
Duration for data collection activity was 7 days. Data collection variables included arrival rate, service rate, service time taken by each resource (Doctor, Nurse, and Receptionist) per patient, income of doctor per patient, fee paid by the patient for getting served by the doctor.

Arrival of Patients
The arrival distribution of patients was not known, the inter arrival time of patients changed independently; therefore, the arrival time of patients was got down on the paper at the study area. The arrival time of patients was necessary to be noted in order to reveal the distribution that patients' arrival followed.

Patients' Service Time
The time taken by the doctor to serve the patient is called as service time. Since, resources at the OPD were receptionist = 1, nurse = 2, doctors = 6; so the service time of each resource was noted with the help of stop watch.

Input Analysis
Analysis of the input variables i.e. arrival and service times was conducted into input analyzer of the Rockwell arena software version 14.5. After the input analysis, the arrival and service distributions of the data were revealed which were put into the several processes in the simulation model.

DEVELOPMENT OF SIMULATION MODEL
After the input analysis of the collected data, the queuing system of the OPD was modelled from the reception to the service by the doctors (Figs. 10-11).

Cost Analysis
Cost analysis was conducted in which, cost of hospital, earning of doctors, waiting cost of patients, total revenue and hospital's profit were calculated for each scenario.

N = Number of Working Hours of the OPD n = Number of Patients
Total Amount charged per patient by the hospital =

Rs.300/-
Rs.300/-was the amount which was supposed to be paid by the patient at the OPD in order to see the doctor.
Service cost of Doctors/patient= Rs.70/-Rs.70 was amount which was supposed to be paid to the doctor for consulting one patient.

Receptionist Service Cost:
Monthly salary i.e. Rs. 12000 of receptionist was collected from the OPD and the service cost per minute was calculated: because it was the part of analysis to calculate the cost of receptionist per patient and the time taken by receptionist to serve the patients was in minutes.
Nurse Service Cost: Similarly, to the receptionist's cost, nurse's cost was calculated in the same way. The salary of the nurse (Rs.10000/-) was collected from the OPD.
Since the customer had to pass from the triage therefore the cost of nurse per minute was necessary to be calculated.
Patient's total system cost = Total System time x Patient's waiting cost/minute (6) Total revenue = n x amount charged by the hospital/ patient (7) Hospital's profit=Total revenue-(total service cost) (8)

ASSUMPTIONS OF THE RESEARCH
Assumptions play a vital role in bridging the real world to mathematical world. Mathematical models are good only because of the good assumptions. In the case of incorrect assumptions or not explicitly described; results would be difficult to assess and misleading as well [22]. Following assumptions were made during the development of the model:  All of the six doctors were there at the OPD for five hours.
 Breaks e.g. tea and prayer were not kept into the consideration.
 Occasional non-availability of resources was not considered in this research.


Only the waiting cost of patient was included in the study; waiting cost of those people who came along was not considered.
 Until the patient was in the OPD, his/her waiting cost was counted.
 Service cost was calculated by summing up the salaries of resources.

Arrival Distribution of Patients at the OPD
As discussed earlier, input analysis was conducted in the input analyzer of Rockwell arena. Arrival distribution of patients was indicated to be poison distribution (POIS (19.9)) with the square error=0.002798 which can be seen in Fig. 1.

Patients' Service Distribution at Reception
Patients' service distribution at reception counter came out to be triangular distribution (TRIA (1.8, 3.15, 3.18)) with the square error=0.002798 which can be seen in Fig. 2.

Patients' Service Distribution by Nurse
Distribution of service of patients by nurse was taken out to be normal distribution (NORM (2.14, 0.442)) with the square error=0.002798 which can be seen in Fig. 3.

Patients' Service Distribution by Doctor1
Distribution of patients' service by doctor1 was taken out to be beta distribution (1+4 x beta (1.57, 1.81)) with the square error=0.002798 which can be seen in Fig. 4.

Patients' Service Distribution by Doctor2
Service distribution of patients in case of doctor1 came out to be triangular (TRIA (3, 9.13, 14)) with the square error=0.002798 which can be seen in Fig. 5.

Patients' Service Distribution by Doctor3
Service distribution of patients in case of doctor3 came out to be normal distribution (NORM (12.2, 1.66)) with the square error=0.002798 which can be seen in Fig. 6.

Patients' Service Distribution by Doctor4
Service distribution of patients in case of doctor4 came out to be uniform distribution (UNIF (3.29, 10.9)) with the square error=0.002798 which can be seen in Fig. 7.

Patients' Service Distribution by Doctor5
Service distribution of patients in case of doctor5 came out to be triangular distribution (TRIA (7.18, 11, 13.6)) with the square error=0.002798 which can be seen in Fig. 8.

Patients' Service Distribution by Doctor6
Service distribution of patients in case of doctor6 came out to be beta distribution (11+7xBeta (1.45, 1.47)) with the square error=0.002798 which can be seen in Fig. 9.

Modelling of the Queuing System of OPD
In the development of queuing system model, patients were considered as entities, OPD was considered as system, receptionist, nurses and doctors were considered as resources.

Simulation of Model for the OPD
Queuing system of the OPD was modelled which can be seen in Fig. 10. In creating module (Patient Arrival) patients were created in the system. As figured out from the collected data that arrival distribution of patients at the

Sub Model
Since the patients were coming for the specific doctors that's why it was necessary to send them to the doctor of their interest. Sub model (Fig. 11) was initiated to model this mechanism of queuing system. Decide module (Nway by chance) was used for sending the patients to the specific doctors. Percentage of patients was calculated for each of the doctor (Doctor1 = 37%, Doctor2 = 16%, Doctor3 = 11%, Doctor4 =15%, Doctor5 =9%, Doctor6 =12%).

Simulation Based Approach for Improving Outpatient Clinic Operations
After deciding upon the patients to be sent to the specific doctor, the time of patients was noted by using the assign modules as shown in Fig. 11.

Replication Parameters
It is required to set replication parameters as given in

Results of Existing Queuing System
Results which were obtained after simulating the model are presented in the below given headings.

Number of patients Served by Each Doctor
In the duration of 4 hours, 71 patients were served by the doctor1 and it was the maximum number of patients and on the contrary 17 patients were served by the doctor5 as shown in Table 3. The number of patients served by each doctor depended on the service time of patients and the minimum service time was doctor1.

Awaiting Patients in the Waiting Line
In the queuing system of OPD, number of queues was eight. The number of awaiting patients in the various queues is presented in Table 4. It can be seen in the

Waiting Time of Patients in the Queues
Waiting time of patients in the various queues is presented in Table 5. On an average, the waiting time patients in different queues is quite less but in the queue of reception it was simulated to be 50.69 minutes.
Maximum time of patients in the reception queue was 233 minutes. Secondly, the waiting time of patients in the queue of doctor 6 was 69.840 minutes; this time was more because of more service time (22.368 minutes (Table   8)) of doctor6. Furthermore, waiting time of patients in the remaining queues was comparatively lesser as shown in Table 5.

Instantaneous Utilization of Resources
In Table 6

Recorded Time of Patients
In Table 7 the recorded service time of doctors to serve the patients is presented. On the same time, total system time of patients coming to six different doctors was recorded separately as shown in Table 7.

Scenario Analysis
Analysis was conducted in four scenarios; in each scenario the working time of resources was increased by one hour. Scenario analysis was conducted because of the existence of two queues. First was the queue of

Simulation Based Approach for Improving Outpatient Clinic Operations
patients which is in its way to see the doctor and another queue was virtual queue which was on hold to have an appointment. Therefore, second queue was required to be minimized so that more and more patients could be served.

Replication Parameters
Replication parameters were set before simulation. In scenario 1, simulation time period was 0.5 year and the time of the OPD was 4 hours. Replication length was kept constant but the duration of OPD was changed by one hour as shown in Table 8.

Patients Served by Specific Doctor in Different Scenarios
Total number of served patients per day was calculated for each of the doctor across all scenarios. The number of served patients kept on increasing with the increasing working time of the OPD. The degree of increment can be seen in Table 9.

Cost Analysis across all Scenarios
Cost analysis was also conducted across all the scenarios.
It included, service cost per day, total revenue of the hospital and total profit of the hospital as shown in

Hospital's Cost on Each Patient
The cost of the hospital on each of the arriving patient at the OPD was taken out.

Patients' Waiting Time and cost in Different Scenarios
During the simulation, some expression were formulated so that the total system time of patients could be recorded as shown in Table 6. Since, they wait in the system that's why cost is involved in it and that cost was termed as waiting cost/opportunity cost.

CONCLUSION
It was not possible to increase the number of doctors at the OPD because the patients were coming for particular doctors at the OPD; since it was aimed to put the suggestion by which the maximum number of patients could be served, therefore, currently available doctors were suggested to sit for longer. If the doctors are supposed to sit for 7 hours in the OPD then 105% more patients would be served as compared to the existing service rate of the OPD.

FUTURE WORK
In this research, total system time of patients at the OPD was considered for the waiting cost. Time spent by the patients at the pharmacy was not included in this study therefore, it should be considered in the future studies.
On the same time patients also spent the time on their way to hospital and the way back to their homes. This time was not also considered for the calculation of waiting cost of patients. (iii) Whole data should also be collected two to three times a year in order to produce the justified results.

MANAGERIAL INSIGHTS
This study helps the hospital administration to improve the healthcare performance. It was revealed during this study that a few numbers of patients may shift to other doctors if they will properly guide by hospital administration. These patients are usually come first time to OPD. By doing this, patients waiting time may reduce as they may be shifted to doctors with shorter queues.