While we're on the subject about where to store data, I'm wondering about the correct value to store in the CDM.
Most of the PHQ items are scored with values from 0-3 reflecting "not at all" to "nearly every day", though the 0-3 score values are assigned sequence numbers from 1 to 4. (drawn from https://r.details.loinc.org/AnswerList/LL358-3.html) However, the 10th question of the PHQ 9 https://r.details.loinc.org/LOINC/69722-7.html?sections=Comprehensive is "How difficult have these made it for you to do your work, take care of things at home, or get along with other people." The standard results for that 10th question range from "not difficult at all" to "extremely difficult." (https://r.details.loinc.org/AnswerList/LL359-1.html). There are no score equivalents for those results, but each answer choice does have a "sequence number" from 1-4. Therefore, what value should we store in the CDM for the PHQ items? Is it the "score value" from 0-3, or the "sequence number" from 1-4? I think I like the score since that is a number that is directly added to give the final score, but there is no score and only a sequence number for the 10th question. Mark From: Campbell, James R [mailto:campb...@unmc.edu] Sent: Friday, March 15, 2019 12:52 PM To: Weiner, Mark; GPC STANDARDS Cc: Keith Marsolo Subject: RE: Question from Mark Weiner of PaTH regarding LOINC coding in OBS_CLIN and PRO_CM EXTERNAL MESSAGE. DO NOT open attachments or click links from unknown senders or unknown emails. ________________________________ Mark The list you have is apparently older. Here is a frequency analysis from Nebraska of LOINC coding in OBS_CLIN PRO_CM and OBS_GEN as we have currently deployed it. I will be adding 75-100 new data items soon. Jim From: Gpc-dev <gpc-dev-boun...@listserv.kumc.edu<mailto:gpc-dev-boun...@listserv.kumc.edu>> On Behalf Of Campbell, James R Sent: Friday, March 15, 2019 10:12 AM To: Weiner, Mark <mark.wei...@tuhs.temple.edu<mailto:mark.wei...@tuhs.temple.edu>>; GPC STANDARDS <gpc-dev@listserv.kumc.edu<mailto:gpc-dev@listserv.kumc.edu>> Cc: Keith Marsolo <keith.mars...@cchmc.org<mailto:keith.mars...@cchmc.org>> Subject: Question from Mark Weiner of PaTH regarding LOINC coding in OBS_CLIN and PRO_CM Non-UNMC email Hi Mark! I hope all is well with you. I will have an expanded map of OBS_CLIN and PRO_CM data for our extracts soon including OB, mental health and neuropathology. Do you want a copy? Initially there was no guidance from PCORI and I dumped eveything in OBS_CLIN. More recently I built the tables based upon the guidance in the PCORnet document as recounted below. I used the LOINC class type to assign codes to the individual tables as all our clinical and lab data is LOINC coded. Checking LOINC you will see that PHQ-2 and 9 are classtype 4 and are now in our PRO_CM table. Unfortunately, the OBS_GEN table remains undefined and unuseful from my point of view. Jim CDMV41 p19: "The LOINC Class Type variable has 4 values (as of May, 2018): 1 = Lab Class, 2 = Clinical Class, 3 = Claims Attachment, 4 = Survey. This information can be found on search.loinc.org or in the LOINC Core files. In general, observations with a class type of "1" will be stored in LAB_RESULT_CM. Observations with a class type of "2" will be loaded into OBS_CLIN, and observations with a class type of "4" will generally be stored in PRO_CM, but not definitively." ________________________________ From: Weiner, Mark <mark.wei...@tuhs.temple.edu<mailto:mark.wei...@tuhs.temple.edu>> Sent: Friday, March 15, 2019 7:50 AM To: Campbell, James R; 'GPC STANDARDS'; Research Informatics Group, UNMC Subject: RE: PCORI OBS_CLIN Non-UNMC email Hello again! While some in our PaTH network have already incorporated many of the flowsheet items you nicely crosswalked to LOINC codes, we are gearing up for a more network-wide implementation in the coming cycle. One of our members asked why PHQ was being placed in the Obs_clin table when it could be considered a PRO measure and placed in the PRO_CM table. Looking at the content of that table, the information certainly fits the structure. My sense was that in the past, it was directed at specific PRO measures like PROMIS, but now that you can specify LOINC codes for PRO measures in the PRO_CM table, the PHQ data could just as easily fit there as in OBS_CLIN. What are your thoughts on where PHQ data belongs? Thanks! Mark From: Weiner, Mark Sent: Thursday, July 12, 2018 3:06 PM To: 'Campbell, James R'; GPC STANDARDS; Research Informatics Group, UNMC Subject: RE: PCORI OBS_CLIN Thanks for your detailed reply and slide decks. My sense is that the hard part is what you seemed to do manually - map the FLO_MEAS_ID from the IP_FLO_GP_DATA table into LOINC codes. The good news is that many of your commonly recorded flowsheet items (like braden score and GCS data) are also frequently recorded within our institution, and within the other institutions within our PaTH CDRN. That means, to the extent there are matches, we can leverage your LOINC mappings to allow us to use our flowsheet rows in the CLIN_OBS table without us having to reinvent the wheel. This is especially useful for the "low" numbered FLOW_MEAS_IDs which are standardized by Epic, Of course, local variation creates more work. For example, for some reason, the FLO_MEAS_ID for our R_PAIN_SCORE is in the "custom" ID range with a value of 3040104280, so a blind application of your mapping into our system would make it seem we didn't have any pain scores, when in fact, the absence is related to a mismatch in code - we aren't using the 6066 code for some reason. We also have a number of variations on the theme of a "generic" pain score with an OB pain score, a PEDS PAIN SCORE, among others. Not sure if these inflections have different LOINC codes, but I'm sure you also have a sense for this variation. I wonder if there is a way we can crowdsource this project to map the FLO_MEAS_IDs that should be common across institutions into the associated LOINC codes. It will add to the ability of the Epic institutions in PCORI to fill their OBS_CLIN table and would ensure a consistent mapping to the extent there may be two similar LOINC codes that COULD match to a single FLO_MEAS_ID. I'm not going to push this issue to hard within my network for the current cycle of data characterization, but after this first CDM 4.1 submission, we may be able to work to collectively expand the mapping. Mark From: Campbell, James R [mailto:campb...@unmc.edu] Sent: Thursday, July 12, 2018 1:50 PM To: Weiner, Mark; GPC STANDARDS; Research Informatics Group, UNMC Subject: Re: PCORI OBS_CLIN WARNING: **** EXTERNAL Message. DO NOT open attachments or click links from unknown senders or unknown emails. **** ________________________________ Hi Mark! I include below the stats for frequency of occurrence of LOINC coded data that we store in our first roll-out of OBS_CLIN. I am also enclosing two presentation from Epic meetings in 2016-7 where we demonstrated our implementation of clinical and laboratory facts in i2b2 (and then to CDM) using ONC standards SNOMED CT and LOINC. You are quite correct in your analysis of Epic Clarity data structures and clinical/lab data that begins with an order event ends up in ORDER_RESULTS. The clinical observations that are listed however are all flowsheet data in our deployment of Epic, and the extract that I developed to load those facts employs a map from Flowsheet row to LOINC code that supports coding of the data. I have not been successful with our build team in deploying flowsheet maps into Epic data structures and so the map file resides with our extract library at this point. We have published our extracts on the userweb - you can find them I think in the slide set. If you are interested, Nebraska will be glad to share our code with you for this or any other feature we have deployed. Jim OBSCLIN_CODE RAW_OBSCLIN_NAME COUNT(*)--SELECT* 8867-4 Heart rate 41195391 9279-1 Respiratory rate 30020150 8310-5 Body temperature 22551370 8478-0 Mean blood pressure 19685904 8462-4 Diastolic blood pressure 17794069 8480-6 Systolic blood pressure 17794069 39156-5 Body mass index (BMI) [Ra 7574390 38208-5 Pain severity - Reported 7039613 50064-5 Ideal body weight 6583916 8302-2 Body height 6524735 3141-9 Body weight Measured 5922939 9267-6 Glasgow coma score eye op 4270690 9268-4 Glasgow coma score motor 4268539 9270-0 Glasgow coma score verbal 4266029 9269-2 Glasgow coma score total 4262288 3140-1 Body surface area Derived 3301569 38222-6 Sensory perception Braden 2827295 38229-1 Moisture exposure Braden 2827250 38223-4 Physical activity Braden 2827192 38224-2 Physical mobility Braden 2827028 38225-9 Nutrition intake pattern 2826941 38226-7 Friction and shear Braden 2826089 38227-5 Braden scale total score 2824397 9187-6 Urine output 2536706 3151-8 Inhaled oxygen flow rate 2194966 59576-9 Body mass index (BMI) [Pe 334323 11784-6 Cervical canal external o 102323 11867-9 Effacement Cervix 90605 11881-0 Uterus Fundal height Tape 84044 11876-0 Fetal presentation palpat 43333 79991-6 Left ventricular Ejection 36724 8287-5 Head Occipital-frontal ci 32128 55283-6 Fetal Heart rate 22784 ________________________________ From: Weiner, Mark <mark.wei...@tuhs.temple.edu<mailto:mark.wei...@tuhs.temple.edu>> Sent: Wednesday, July 11, 2018 9:59:39 AM To: Campbell, James R Subject: PCORI OBS_CLIN Non-UNMC email Someone circulated data on the Nebraska experience with populating he PCORI CDM OBS_CLIN table using data from the EHR that corresponded to LOINC codes having a LOINC_CLASS = 2. We have Epic an our institution, and we were doing the same thing with data that happened to be in our ORDER_RESULTS table where most tests have a component_ID that was linked to a LOINC code directly in our Clarity system. In that way, we were able to pick up non-blood/urine tests like PFTS that were stored in the same ORDER_RESULTS table as more routine blood tests. I saw in your data that you had other items like Braden scores, GCS, and pain scores. We had that data too, but it was in our flowsheet data, not the order_results table, and we didn't have an easy crosswalk between those flowsheet items and the corresponding LOINC code. Presuming you use Epic, too, what was the Clarity source for most of your OBS_CLIN data? Did you have an automated crosswalk between items like Braden Score and the associated LOINC code? I can see looking it up manually, or doing a string match with the LOINC dictionary, but I envision that will not have a perfect alignment. I appreciate your insight. Thanks! Mark ------------------- Mark Weiner, MD, FACP, FACMI Assistant Dean for Informatics Professor of Clinical Sciences and Medicine Lewis Katz School of Medicine at Temple University 3440 N. Broad Street, Kresge Hall, Rm 219 Philadelphia, PA 19140 t 215-707-8079 f 215-707-3160 ________________________________ This electronic message is intended to be for the use of the named recipient, and may contain information that is confidential or privileged. This communication may contain protected health information (PHI) that is legally protected from inappropriate disclosure by the Privacy Standards of the Health Insurance Portability and Accountability Act (HIPAA) and relevant Pennsylvania Laws. You can direct questions concerning PHI or HIPAA to the Corporate Compliance and Privacy Officer at (215) 707-5605. If you are not the intended recipient, please note that any dissemination, distribution or copying of this communication is strictly prohibited. If you have received this message in error, you should notify the sender immediately by telephone or by return e-mail and delete and destroy all copies of this message. The information in this e-mail may be privileged and confidential, intended only for the use of the addressee(s) above. Any unauthorized use or disclosure of this information is prohibited. If you have received this e-mail by mistake, please delete it and immediately contact the sender. The information in this e-mail may be privileged and confidential, intended only for the use of the addressee(s) above. Any unauthorized use or disclosure of this information is prohibited. If you have received this e-mail by mistake, please delete it and immediately contact the sender. The information in this e-mail may be privileged and confidential, intended only for the use of the addressee(s) above. Any unauthorized use or disclosure of this information is prohibited. If you have received this e-mail by mistake, please delete it and immediately contact the sender.
_______________________________________________ Gpc-dev mailing list Gpc-dev@listserv.kumc.edu http://listserv.kumc.edu/mailman/listinfo/gpc-dev