Query check CCI CXR vs LOS: Difference between revisions

From CCMDB Wiki
Jump to navigation Jump to search
No edit summary
 
(30 intermediate revisions by 2 users not shown)
Line 1: Line 1:
{{Data Integrity Check
{{Data Integrity Check
|DIC_summary=Confirm that a [[:Category:Labs Imaging]] count is not unreasonably high
|DIC_summary=Confirm that a [[:Category:Labs Imaging]] count is not unreasonably high
|DIC_related_concepts=CXR (plain film); LOS; Arrive DtTm; Dispo DtTm
|DIC_related_concepts=CXR (plain film); LOS; Service tmp entry; Dispo DtTm
|DIC_firmness=soft check
|DIC_firmness=soft check
|DIC_timing=complete
|DIC_timing=complete
Line 9: Line 9:
|DIC_backlogged=No
|DIC_backlogged=No
}}
}}
We have had some medical [[:Category:Labs Imaging]] count entries that are so high that they are likely errors.  
* Not currently implemented; starting point for the query exists but doesn't yet include Julie's values. Also, isn't being called from anywhere.
 
=== Discussion ===
{{Discuss |
* Just came across this... the started query includes additional dxs now, as per [[#CCI collect count each]]. Does the proposed accepted count make sense for all the dxs? Your validation table was for CXR. [[User:Ttenbergen|Ttenbergen]] 11:50, 2023 May 3 (CDT)
** Did some checking on all 13 LAB Imaging from CCI picklist and found counts more than the LOS only in ECHO, AXR and CXR. Emailed Lisa and Pagasa March 14, 2024 to check if correct or not. Waiting for their feedbacks. In the email, I also propose another query (e.g. the counts per calendar day per patient must not be more than one) as alternative to the threshold limits LOS +- 3 STD shown below.  --[[User:JMojica|JMojica]] 10:55, 2024 March 18 (CDT)
*** That query is tighter and would therefore have more false positives. What does Lisa think about that? [[User:Ttenbergen|Ttenbergen]] 09:51, 2024 March 20 (CDT)
**** I made some limits per LOS group for CXR, AXR, ECHO and Blood gases and emailed the results to Tina and Lisa 3/25/2024 for discussion. ---[[User:JMojica|JMojica]] 13:28, 2024 March 25 (CDT)
***** Further emails with Julie and Lisa about details. Discussed with Lisa who is reviewing some of the errors and will hold off until she is done. [[User:Ttenbergen|Ttenbergen]] 12:46, 2024 April 8 (CDT)
}}
 
We have had some medical [[:Category:Labs Imaging]] count entries that are so high that they are likely errors.


=== What are likely maximum entries? ===
=== What are likely maximum entries? ===
The count per patient stay is related to the [[LOS]]. The threshold limits are either Mean + 3 Standard Deviation or Maximum value of Chest X-ray count per patient day for various LOS category listed in the table below.  They are derived from July 1988 to June 2017 Lab data. The LOS is based on calendar dates where partial stay is considered as  full day.  
The count per patient stay is related to the [[LOS]]. The threshold limits are either Mean + 3 Standard Deviation or Maximum value of Chest X-ray count per patient day for various LOS category listed in the table below.  They are derived from July 1988 to June 2017 Lab data. The LOS is based on calendar dates where partial stay is considered as  full day.  


*LOS based on calendar dates = [[Dispo DtTm]] - [[Arrive DtTm]] + 1
*LOS based on calendar dates = [[Function LOS()|Function LOS()]] + 1
{{Discuss | who=Julie | question= would we not use [[Accept DtTm]] here? Because we could have CXRs on days before arrival...
* we can use [[Accept DtTm]]  but with the Tmp Service, use the the first service start dttm instead. --[[User:JMojica|JMojica]] 09:18, 2021 September 2 (CDT)}}
** if patient is still in the unit,  [[Dispo DtTm]] is replaced by today's date
** if patient is still in the unit,  [[Dispo DtTm]] is replaced by today's date
*** check is marked as only running on complete records, if that is so then we don't need to replace with today's date
*Chest X-ray (CXR) per patient day = Number of chest X-rays / LOS Based on calendar dates
*Chest X-ray (CXR) per patient day = Number of chest X-rays / LOS Based on calendar dates


Line 45: Line 55:
{{CCMDB.accdb cross checks triggering for correct values}}
{{CCMDB.accdb cross checks triggering for correct values}}


{{DJ |
== Checks ==
* is there a function we can use for the threshold? That would be easier to implement than a list like that. If SAS as a way to generate functions (like excel does for its trendlines) then could you run that? [[User:Ttenbergen|Ttenbergen]] 09:02, 2021 September 2 (CDT)
This uses [[Function LOS()]] to calculate the [[LOS]].  
* looks like from the list above, the mean plus 3 std of 2 per day is common except when LOS <1. we can use the function twice of LOS. the CXR should not exceed  twice of LOS.  when LOS <1, should not exceed 5. When LOS is 51 or more, should not exceed LOS. --[[User:JMojica|JMojica]] 09:18, 2021 September 2 (CDT) }}
 
=== CCI collect count each ===
* [[AXR (abdominal plain X-ray)]] 3.OT.10.VA
* [[Blood gas, arterial]] 2.ZZ.13.RE
* [[Blood gas, venous]] 2.ZZ.13.RA
* [[CXR (plain film)]] 3.GY.10.VA
 
"CCI collect count each" count should not be higher than:
* if LOS < 1, 5
* if LOS 1-50, LOS * 2
* if LOS > 50, LOS
 
=== CCI collect count days ===
"CCI collect count days" should not be higher than LOS. See [[query check_CCI_vs_LOS_count_days_component]] and [[query check_CCI_vs_LOS_count_days_picklist]].
 
=== CCI collect count units ===
Discussed with Julie that this one is likely uncheckable. [[User:Ttenbergen|Ttenbergen]] 15:47, 2021 September 2 (CDT)
 
=== CCI collect each ===
Discussed with Julie and no specific checks required for this one.
 
== Log ==
* 2022-02-10 - confirmed that this would not be affected by the change of including a time in the [[Px Date]]
* 2022-01-19 - Julie confirmed that we don't need to check this per [[Boarding Loc]] - per profile is enough (especially since it's a soft check)


== Related articles ==
== Related articles ==
Line 54: Line 87:
[[Category:Labs]]
[[Category:Labs]]
[[Category:Length of stay]]
[[Category:Length of stay]]
[[Category:CCI Px check]]
[[Category:Data Integrity Checks]]

Latest revision as of 12:46, 2024 April 8

Data Integrity Checks
Summary: Confirm that a Category:Labs Imaging count is not unreasonably high
Related: CXR (plain film), LOS, Service tmp entry, Dispo DtTm
Firmness: soft check
Timing: complete
App: CCMDB.accdb
Coding: query check_CCI_CXR_vs_LOS
Uses L Problem table: not relevant for this app
Status: needs review
Implementation Date: not entered
Backlogged: No
  • Cargo


  • SMW


  • Categories: 
  • form:
  • Not currently implemented; starting point for the query exists but doesn't yet include Julie's values. Also, isn't being called from anywhere.

Discussion

  • Just came across this... the started query includes additional dxs now, as per #CCI collect count each. Does the proposed accepted count make sense for all the dxs? Your validation table was for CXR. Ttenbergen 11:50, 2023 May 3 (CDT)
    • Did some checking on all 13 LAB Imaging from CCI picklist and found counts more than the LOS only in ECHO, AXR and CXR. Emailed Lisa and Pagasa March 14, 2024 to check if correct or not. Waiting for their feedbacks. In the email, I also propose another query (e.g. the counts per calendar day per patient must not be more than one) as alternative to the threshold limits LOS +- 3 STD shown below. --JMojica 10:55, 2024 March 18 (CDT)
      • That query is tighter and would therefore have more false positives. What does Lisa think about that? Ttenbergen 09:51, 2024 March 20 (CDT)
        • I made some limits per LOS group for CXR, AXR, ECHO and Blood gases and emailed the results to Tina and Lisa 3/25/2024 for discussion. ---JMojica 13:28, 2024 March 25 (CDT)
          • Further emails with Julie and Lisa about details. Discussed with Lisa who is reviewing some of the errors and will hold off until she is done. Ttenbergen 12:46, 2024 April 8 (CDT)
  • SMW


  • Cargo


  • Categories

We have had some medical Category:Labs Imaging count entries that are so high that they are likely errors.

What are likely maximum entries?

The count per patient stay is related to the LOS. The threshold limits are either Mean + 3 Standard Deviation or Maximum value of Chest X-ray count per patient day for various LOS category listed in the table below. They are derived from July 1988 to June 2017 Lab data. The LOS is based on calendar dates where partial stay is considered as full day.

  • LOS based on calendar dates = Function LOS() + 1
    • if patient is still in the unit, Dispo DtTm is replaced by today's date
      • check is marked as only running on complete records, if that is so then we don't need to replace with today's date
  • Chest X-ray (CXR) per patient day = Number of chest X-rays / LOS Based on calendar dates
LOS_days N Mean Std Dev Minimum Median 75th Pctl 90th Pctl 95th Pctl 99th Pctl Maximum mean + 3 Std Dev Limits of CXR per patient day
1 4021 1.16 0.42 1 1 1 2 2 3 5 2.42 5
2 23910 0.83 0.31 0.5 1 1 1 1.5 2 4 1.76 4
3-10 52600 0.64 0.34 0.1 0.67 0.88 1 1.25 1.6 3 1.66 3
11-20 7784 0.7 0.31 0.05 0.71 0.92 1.09 1.19 1.42 2.47 1.63 2
21-30 1870 0.68 0.3 0.03 0.69 0.9 1.05 1.14 1.36 2.57 1.58 2
31-40 660 0.62 0.3 0.03 0.64 0.84 1 1.09 1.32 1.67 1.52 2
41-50 311 0.54 0.29 0.02 0.53 0.73 0.93 1.02 1.21 1.58 1.41 2
51-100 347 0.41 0.26 0.02 0.38 0.6 0.76 0.85 1.12 1.2 1.19 1
> 100 101 0.19 0.21 0.01 0.11 0.2 0.48 0.73 0.88 1.09 0.82 1

CCMDB.accdb cross checks triggering for correct values

If the actual correct data leads to a cross-check error CCMDB.accdb will not allow you to send. In that case, enter a value that is acceptable and then email the Data Processor using the "generate email" button.

Checks

This uses Function LOS() to calculate the LOS.

CCI collect count each

"CCI collect count each" count should not be higher than:

  • if LOS < 1, 5
  • if LOS 1-50, LOS * 2
  • if LOS > 50, LOS

CCI collect count days

"CCI collect count days" should not be higher than LOS. See query check_CCI_vs_LOS_count_days_component and query check_CCI_vs_LOS_count_days_picklist.

CCI collect count units

Discussed with Julie that this one is likely uncheckable. Ttenbergen 15:47, 2021 September 2 (CDT)

CCI collect each

Discussed with Julie and no specific checks required for this one.

Log

  • 2022-02-10 - confirmed that this would not be affected by the change of including a time in the Px Date
  • 2022-01-19 - Julie confirmed that we don't need to check this per Boarding Loc - per profile is enough (especially since it's a soft check)

Related articles

Related articles:
Retrieved from "https://ccmdb.kuality.ca/index.php?title=Query_check_CCI_CXR_vs_LOS&oldid=161207"