12 month washout period

[ashtrik203]

Hello Everyone,

I wanted to have like a one-year wash-out period in my analysis , i.e. if the same type of event occurs within one year then it is not included as an outcome event.

A sample of my data looks something like this

Patientkey	Diagnosis date
1123	21/11/2018
1123	01/01/2019
1123	22/05/2021

So for this patient, I want my count to be 2, as I will not be counting the 21/11/2018 as it occurs in less than 1 years of 01/01/2019.

I did this, but I am aware that it is not appropriate.

count(${<([DiagnosisDate]-Above [DiagnosisDate])>'365'>} [PatientKey])

Any help would be greatly appreciated.

Thanks 

[ASEDavidSu]

You can try these formulas, hope it helps.

[ashtrik203]

Thank you for the reply. Unfortunately, my count is coming to be '0'

Not sure what is not working

[sidhiq91]

@ashtrik203  Could you please try the below logic:

In the back end the script would be as below:

NoConcatenate
Temp:
Load Patientkey,
Date(Date#([Diagnosis date],'DD/MM/YYYY'),'MM/DD/YYYY') as [Diagnosis date]

Inline [
Patientkey,Diagnosis date

1123,21/11/2018

1123,01/01/2019

1123,22/05/2021

];

NoConcatenate
Temp1:
Load *,
if(Previous(Patientkey)=Patientkey and
(Interval([Diagnosis date]-Previous([Diagnosis date]),'DD'))>'365','1','0') as Flag
Resident Temp
Order by Patientkey,[Diagnosis date];

Concatenate(Temp1)
Load Patientkey,
Min([Diagnosis date]) as [Diagnosis date],
'1' as Flag
Resident Temp
group by Patientkey;

Drop table Temp;

Exit Script;

In the front end the expression would be as:

Count({<Flag={'1'}>}Patientkey)

Let me know if this resolved your issue, if yes please like and accept it as a solution.

[ashtrik203]

Thank you, @sidhiq91 .

Unfortunately, i cannot get it to work. The flags are not appropriately placed when I run it on my data. not sure why

Myabe to explain it further, I have multiple people in my record. I have to count all people who had say a headache. Among these patients, if someone had headache mutliple times, then I will only count them if they are more than I year apart. If there is only one date for headache for a person, then it's straight forward, I just count them once

Patientkey	Condition	Diagnosis date
1123	Headache	21/11/2018
1156	Stomach-ache	22/6/1994
1123	Headache	01/01/209
1136	Headache	51/11/2002
1156	Fever	5/08/2005
1123	Headache	22/05/2021

Output

Total Headache- 3 (2 for 1123 and 1 for 1136)

Total fever- 1 (for 1156)

Total Stomach-ache- 1(for 1156)

Hope it's clearer now

Any help would be appreciated 

[sidhiq91 · Accepted solution]

@ashtrik203 I have followed the same logic as above and got the desired output. Attaching the screen shot as well.

Please see the code used in the backend.

NoConcatenate
Temp:
Load Patientkey,
Condition,
Date(Date#([Diagnosis date],'DD/MM/YYYY'),'MM/DD/YYYY') as [Diagnosis date]

Inline [
Patientkey,Condition,Diagnosis date

1123,Headache,21/11/2018

1156,Stomach-ache,22/6/1994

1123,Headache,01/01/2019

1136,Headache,15/11/2002

1156,Fever,5/08/2005

1123,Headache,22/05/2021

];

NoConcatenate
Temp1:
Load *,
if(Previous(Patientkey)=Patientkey and Previous(Condition)=Condition and
(Interval([Diagnosis date]-Previous([Diagnosis date]),'DD'))>'365','1','0') as Flag
Resident Temp
order by Patientkey,Condition,[Diagnosis date];

Concatenate(Temp1)
Load Patientkey,
Condition,
Min([Diagnosis date]) as [Diagnosis date],
'1' as Flag
Resident Temp
group by Patientkey,Condition;

Drop table Temp;

exit script;

In the front end to count number of Patients I used:  Count({<Flag={1}>}Patientkey)

This should resolve your issue, If yes please like and accept it as a solution.