Target Variables For Study Of Project

Definition

 

• Project Justification: This entire study is to be conducted in order to reduce long stays of patients at the ED.Now for this we will focus our study on mainly few variables like departure time, arrival time, triage category, visit type, mode of separation. We are required to calculate number of hours spent at the emergency department by finding difference between departure time, arrival time and find the impact of each of the variables i.e.triage category, visit type, mode of separation on number of hours spent and justify it statistically in a scientific manner with right outcomes of our study. We will consider null hypothesis to study the problem. So at the end we can conclude if these factors lead to long stays or not.
• Scope: The main objective of this project is “to reduce long stay of patients at the ED.”

This includes a consistent description of data –

The target variables which are shortlisted as part of study of our project are: 

• Departure_Date –

Definition:

For the Admitted patient:

The time the person is either:

1. When patient is transferred to the ward or separate unit.

2. When patient departs from ED to some other Hospital.

For the Non-Admitted Patient:

This is the time when Patient has already undergone the initial phase of the treatment.

• Departure Time – This involves time only

3. Actual Departure Time-This involves time only

4. Arrival Date –

Definition

Date and time on which the patient is present for the service

5. Arrival Time—same as No.4

6. Mode of Separation-

 This gives us the status of the patient at separation from the Emergency Department and it also details

 about the place where the patient is shifted. So mode of eparation gives information about of pace of

 shifting and status.

 7. Ed Visit Type- The reason for which patient is present at the Emergency Department

Type of Visits: 

• Return Visit – Planned – the visit is planned and as a result of a prior emergency visit or return visit. It is because for planned follow-up treatment or due to the test results showing a necessity of further treatment.
• Unplanned Return Visit for Continuing Condition – the person unexpectedly returns due to same condition to the same Emergency Department. The visit may be due to the last treatment condition.
• Outpatient Presentation – visit that happens due to a planned visit to a formal clinic .

Pre-Arranged Admission:  Excludes ED check-up. This refers to where the patient has pre-organized the admission. Patient may or may not be triaged but is not inspected by ED Medical staff. 

Pre-Arranged Admission: Includes ED check-up – where the patient has pre-organized the admission process. Now the patient is triaged, undergoes the official admission and is inspected by ED Medical staff.

Person in Transit – the ED takes the responsibility for attending and treating the patient, who waits for an ambulance services to a different institution.

Dead On Arrival – A patient who is declared dead by Medical Staff and who receives no medical treatment by hospital staff. Mode of Separation of Dead On Arrival needs to be mentioned

Disaster – The visit is a result of a disaster or nature’s attack, where a disaster plan is followed.

Telehealth Presentation – In this process the patient’s treatment is done via audiovisual devices. 

a) Summary of dataset exploration (consolidation of individual results), including data quality

This Dataset involves one Response variable “Stay_Time_Hours ” which can be calculated by finding difference between Arrival time and Departure Time.

To agree with the objective of our study “To reduce long stay of patients at the ED” I have introduced

new variable for analysis : Stay_Time_Hours.This is derived from current dataset given to us for analysis by finding difference between actual_departure_time and actual_arrival_time.

The calculation to arrive at new variable:

Arrival Date

Stay_Time_Hours  = (actual_departure_time)  –  (actual_arrival_time) .

This is the number of hours spent by the patient at the emergency department i.e. time duration

for which he/she stays at the emergency department.

So as a part of our analysis we have just focussed on four variables as given below in the table: 

Variables	Type of Variables
Stay_Time_Hours	Continuous
mode_of_separation	Nominal(Categorical)
ed_visit_type	Nominal(Categorical)
triage_category	Ordinal(Categorical)

 b) Analysis Plan:

Step1: To analyse this complete data which involves 620999 number of observations as a part of study, we will first try to find the minimum, maximum, outliers and extreme values in the existing datasets for variable: Stay_Time_Hours.     

Step2: Then we will remove the outliers and clean the data set and replace the outliers with minimum and maximum values.

Step3: After doing outlier treatment we will do the general frequency analysis to get percentage of Patients falling under different categories like : mode_of_separation, ed_visit_type, triage_category.

Step4: Performing one –way ANOVA(Analysis of Variance) for variable :” Stay_Time_Hours “  separately with each of the different variables : mode_of_separation, ed_visit_type, triage_category.

Step5: Find the significant value to arrive at the correct model for our study.

Stpe6: Assume the hypothesis : “Mean value of “Stay_Time_Hours” for each category of mode_of_separation is equal.

Step7: Assume the hypothesis : “Mean value of “Stay_Time_Hours” for each category of ed_visit_type is equal.

Step8: Assume the hypothesis : “Mean value of “Stay_Time_Hours” for each category of triage_category is equal.

Step9: Find the significant difference between the means of groups.

Step 10: Conclude final analysis model.

Results:

Result#1 – Percentage of patients falling under different categories like : mode_of_separation, ed_visit_type, triage_category.

• Analysis 1- After doing analysis of entire data it was found that 97.76% people who visit emergency department are of ed_visit_type =1 . So we need to target these people to reduce the time of stay.

• Analysis 2- Also 56% of people who visit emergency department are of mode of separation = 4 and 28.9% of people who visit emergency department are of mode of separation = 1.

So we need to target these people to reduce the time of stay.

• Analysis 3- Also 42.7% of people who visit emergency department are of triage category = 4 and 33% of people who visit emergency department are of Triage category= 1.

So we need to target these people to reduce the time of stay.

Please find details of Data Analysis of variables which are main part of study as shown below on next page:-

_visit_type
ed_visit_type	Frequency	Percent	Cumulative Frequency	Cumulative Percent
1	607031	97.76	607031	97.76
2	5743	0.92	612774	98.69
3	4699	0.76	617473	99.44
4	196	0.03	617669	99.48
5	354	0.06	618023	99.53
6	907	0.15	618930	99.68
8	532	0.09	619462	99.76
9	44	0.01	619506	99.77
10	1271	0.20	620777	99.98
11	74	0.01	620851	99.99
12	3	0.00	620854	99.99
13	72	0.01	620926	100.00
Frequency Missing = 77

mode_of_separation
mode_of_separation	Frequency	Percent	Cumulative Frequency	Cumulative Percent
1	179761	28.95	179761	28.95
2	14381	2.32	194142	31.27
3	652	0.10	194794	31.37
4	347706	56.00	542500	87.37
5	2435	0.39	544935	87.76
6	25721	4.14	570656	91.90
7	19317	3.11	589973	95.01
8	1285	0.21	591258	95.22
9	8006	1.29	599264	96.51
10	11737	1.89	611001	98.40
11	4354	0.70	615355	99.10
12	4559	0.73	619914	99.83
13	1042	0.17	620956	100.00
Frequency Missing = 47

triage_category
triage_category	Frequency	Percent	Cumulative Frequency	Cumulative Percent
1	6441	1.04	6441	1.04
2	81844	13.19	88285	14.23
3	205740	33.16	294025	47.39
4	265045	42.72	559070	90.11
5	61387	9.89	620457	100.00
Frequency Missing = 546

 Result#2: Minimum, Maximum and outlier values in existing data set

As per details are shown in table below:

N	Mean	Minimum	Maximum
620999	4.8277838	-8759.93	3679.50

Outlier values are: 3679.50000 and -8759.93333

Minimum Value: 0.2

Maximum Value: 21.6

Mean Value: 4.8

Result#3: Minimum and Maximum values in cleaned data set

After cleaning data by removing outliers below are details of the Minimum, Maximum and Mean values:

N	Mean	Minimum	Maximum
621003	4.7653256	0.2000000	21.6000000

Result#4: Null Hypothesis: Mean value of Dependent Variable: Stay_Time_Hours is Equal for all 5 groups of triage_category.

After applying Anova test to compare mean values of all 5 groups of triage category, we get below result   :       

Group Level Information
Group	Levels	Values
triage_category	5	1 2 3 4 5

Method	DF	Sum of Squares	Mean Square	F Value	Pr > F
Model	4	803399.338	200849.835	14479.4	<.0001
Error	620452	8606531.830	13.871
Corrected Total	620456	9409931.169

Stay_Time_Hours Mean
4.768647

Conclusion:

Here default value of alpha=0.05 and p-value that we have calculated from Anova test is .0001. Now since p-value is less than alpha so we can reject Null hypothesis. So we conclude that mean values from Group1 to Group5 are not equal i.e. mean value of at least two groups out of 5groups are not equal and is different from each other.so patients that comes under the group whose mean value is higher will be our focus group in order to reduce the time duration of stay for that group.

Result#5: Null Hypothesis: Mean value of Dependent Variable: Stay_Time_Hours is Equal for all 13 groups of mode_of _separation

After applying Anova test to compare mean values of all 13 groups of  mode_of _separation, we get below    result: 

Group Level Information
Group	Levels	Values
mode_of_separation	13	1 2 3 4 5 6 7 8 9 10 11 12 13

Method	DF	Sum of Squares	Mean Square	F Value	Pr > F
Model	12	2243872.610	186989.384	16182.3	<.0001
Error	620943	7175092.088	11.555
Corrected Total	620955	9418964.698

Stay_Time_Hours Mean
4.765329

Conclusion:

Here default value of alpha=0.05 and p-value that we have calculated from Anova test is .0001. Now since p-value is less than alpha so we can reject Null hypothesis. So we conclude that mean values from Group1 to Group13 are not equal i.e. mean value of at least two groups out of 13groups are not equal and is different from each other.so patients that comes under the group whose mean value is higher will be our focus group in order to reduce the time duration of stay for that group.

Result#6: Null Hypothesis: Mean value of Dependent Variable: Stay_Time_Hours is Equal for all 12 groups of ed_visit_type

Group Level Information
Group	Levels	Values
ed_visit_type	12	1 2 3 4 5 6 8 9 10 11 12 13

Method	DF	Sum of Squares	Mean Square	F Value	Pr > F
Model	11	78598.278	7145.298	475.02	<.0001
Error	620914	9339927.200	15.042
Corrected Total	620925	9418525.478

Stay_Time_Hours Mean
4.765329

Conclusion:

Here default value of alpha=0.05 and p-value that we have calculated from Anova test is .0001. Now since p-value is less than alpha so we can reject Null hypothesis. So we conclude that meanvalues from Group1 to Group12 are not equal i.e. mean value of at least two groups out of 12 groups arenot equal and is different from each other.so patients that comes under the group whose mean value is  higher will be our focus group in order to reduce the time duration of stay for that group.

Anon., n.d. NSW Health Admitted Patient Data Collection. [Online] Available at:   HYPERLINK

“https://internal.health.nsw.gov.au/im/ims/ap/index.html ”   https://internal.health.nsw.gov.au/im/ims/ap/index.html .

Anon., n.d. NSW Health Data Dictionary. [Online] Available at:   HYPERLINK

“https://internal.health.nsw.gov.au/im/ims/standards/nsw-health-data-dictionary.html  ”   https://internal.health.nsw.gov.au/im/ims/standards/nsw-health-data-dictionary.html .

Anon., n.d. NSW Health Emergency Department Data Collection. [Online] Available at:   HYPERLINK

“https://internal.health.nsw.gov.au/data/collections/edc/  ”   https://internal.health.nsw.gov.au/data/collections/edc/ .

(Anon., 2017)www.sas.com  (Anon., n.d.)