Case 2009-OR-01 Problems with DRG 530-541
|Target version:||Expert Group 2013|
|Case type:||Major||Owner / responsible:||Nordic Casemix Centre|
|MDC:||OR||Old forum status:||FURTHER ACTITEM - Further active item|
|Target Grouper:||COMMON, DEN, EST, FIN, ICE, LAT, NOR, SWE|
Last updated: 2012-05-07
Status: FURTHER ACTITEM - Further active item
By: Several DRG-users/CPK, Sweden
Expert Network 2009-02-26 - Postpone recommended
Board 2009-03-20 - Recommendation accepted
Expert Network 2010-03-12 - Postpone recommended
Board 2010-04-09 - Recommendation accepted
Expert Network 2011-03-17 - Postpone recommended
Board 2011-04-04 - Recommendation accepted
Expert Network 2012-03-08 - Change recommended
Board 2012-03-19 - Recommendation accepted
Expert Network 2013-03-12 Change recommended
Board 2013-03-27 Change accepted
CPK ID 283
In NordDRG 2008 seven new DRGs (530, 531, 534, 535, 537, 538 and 541) were introduced with the purpose to handle cases treated under general anaesthesia (or anaesthesia with similar resource demands) but without any surgical procedure, see [[http://documents.norddrg.net/issues/201]]. Previously these cases were grouped to DRG 477. The new DRGs were designed only for inpatient care with duration of two days ore more.
The new DRGs have been questioned by a lot of DRG users. The medical relevance is poor (see examples below) and on principal grounds it is considered doubtful that general anaesthesia shall rule the grouping of care with a rather long duration. General anaesthesia is probably not the dominating cost driver in these cases. It is also considered absurd that general anaesthesia affects the grouping of inpatients but not the grouping of outpatients. It should be the reverse since the cost for general anaesthesia is a far larger part of the total cost of an outpatient visit.
In the table below there are examples where the NordDRG users find the medical relevance of the new groups poor. The conservative DRGs, that are achieved when the anaesthesia code is omitted, describe better the medical problem.
CPK has tried to analyse the outcome of the new grouping logic resulting in DRG 530-541. These DRGs are not present in the latest (year 2007) national cost database but the west region cost database containing data from year 2006 and 2007 has been grouped with NordDRG version 2009 and the result for the groups 530-541 is as follows.
In the further analyse, only DRG 530 is represented since it is the biggest group, most groups have too few cases to analyse. The result should be applicable to the other groups. Only cases with LOS > 1 day are included. In table 2 we show how the new logic affects different DRGs where cases has been transferred to DRG 530.
The mean cost for patients in DRG 530 is 24 000 SEK and they are transferred from different DRGs with quite a varying mean cost (13 000 – 48 000 SEK). The CV for these DRGs are both increased and decreased but the changes are very small. The conclusion from this analysis is that from an economic perspective both the new and old logic is OK. From a medical point of view it is more problematic. The old logic for inpatients is much more medical meaningful.
We want the Nordic Expert group to discuss the meaningfulness of the DRGs 530-541. We think that they can be omitted and replaced by similar groups for outpatients.
Expert Network 2009-02-26
The meeting recommended to postpone the case. Sweden should analyse it further. Obstetric cases for DRG 530 will disappear, see [[http://documents.norddrg.net/issues/202]].
We have previously noted that DRG 530-541 are rather poor for medical description. Now we have studied the economic effects of having the cases in DRG 530-541 grouped to the corresponding conservative DRGs according to the principal diagnoses.
In the study, all cases in DRG 530-541 in the Swedish cost database for inpatients 2008 were grouped with NordDRG 2010 and then regrouped after that the anesthesia codes with the diagnosis property 00X10 hade been replaced by a dummy code. shows the original and the new DRGs for all 5 331 cases and shows what happens in the affected receiving DRGs.
There were more than 200 DRGs involved, originally containing almost 490 000 cases. After regrouping, the number of cases in the receiving DRGs was increased with the 5 331 cases from DRG 530-541. That is only an increase with 1 %. The cv value for the receiving DRGs was sometimes decreased and sometimes increased but the changes were generally very small. In summary the study shows very small changes in the receiving DRGs and these changes are acceptable in order to get a better medical description.
Therefore we want a change in the Swedish version of NordDRG 2011 so that DRG 530-541 are deleted and all the cases that now are grouped to these DRGs instead will be grouped to conservative DRGs according to the principal diagnoses. Technical solutions to achieve such a change are suggested below.
1. The OR property is removed from all the anesthesia codes. Then the cases will be grouped to conservative DRGs according to the principal diagnoses, as desired.
2. A. To keep the possibility for surgical partitioning of smaller procedures performed with general anesthesia (or similar) some new rules can be added to the drglogic table. Existing rules with OR = S or P and with a specified procedure property containing the letter S or E are supplemented with new rules that are copies of the existing rules but with an empty OR field and the value 00X10 in a dgprop field. Such new rules are necessary only when there exists procedure codes with the procedure property in question but without OR property 1.
2. B. Alternative 2A will increase the table drglogic with many new rules and the updating process when new procedure codes are introduced will be more complicated. An easier way to keep the possibility for surgical partitioning of smaller procedures performed with general anesthesia is to give the anesthesia codes an inactive OR property that is activated only in the presence of a procedure code containing the letter S or E. Such a system should be possible since a similar system already exists for CC properties.
Expert Network 2010-03-12
The meeting recommended to still postponing the case.
Recommended alt 2b but should await for the result at of the work with the new CC-logic.
Mats Fernström, NPK 2011-03-04
We have tested alternative 2B with the NPK grouper (designed by Janos Nagy). All the codes for greater anaesthesia (defined by the diagnosis property 00X10) were given an inactive OR property that was converted to OR=1 if any code with a procedure property with the letter S was present or converted to OR=2 if any code with a procedure property with the letter E was present (without a simultaneous presence of a procedure property with the letter S).
The procedure property 17S02 caused a problem. That property is given to practically all codes for endoscopic procedures and it activates the OR property for greater anaesthesia. Thus, all in-patients with endoscopy and greater anaesthesia but without a principal diagnosis for history of malignancy (dgcat 17M05) were grouped to DRG 477, although they were coded correctly. We found out that the best way to solve that problem is to delete DRG 412 (History of malignancy w endoscopy), the only DRG that uses the procedure property 17S02. DRG 412 is not very relevant in a medical perspective and there are few cases in Sweden. Both in a medical and a cost perspective it is OK to let the cases go to DRG 411 (History of malignancy w/o endoscopy). The merged group could be named just “History of malignancy”. For cost data, see table below.
After deletion of DRG 412 the preliminary results of the test were very promising. Most of the cases in DRG 530-541 were transferred to conservative groups (as desired) and the rest of the cases were transferred to DRG 477 because they had an unusual or incorrect combination of principal diagnosis and procedures. As a matter of fact, the test has shown that a lot of miscoded cases are hidden in DRG 530-541, which is yet another reason to get rid of these DRGs. In miscoded cases DRG 477 is much better than DRG 530-541, because then the DRG user becomes aware of the miscoding.
We still have to make a full scale test with the new logic on the whole Swedish patient register, including outpatients, to see if there are any negative side effects. A report will be presented at the meeting in Reykjavik.
Expert Network 2011-03-17*
The meeting recommended that the proposal to remove drg 530, 531, 534, 535, 537, 538 and 541 should be tested in the CC-grouper
The case should be postponed.
Mats Fernström, NPK 2011-03-18
The result of the full scale test on the Swedish PAR 2009 was presented at the meeting in Reykjavik. The following technical changes had been done in the grouper:
- All codes for greater anaesthesia (identified by the diagnosis property 00X10) hade been given OR property 3 instead of 1.
- The grouper had been hard coded so that any existing OR property with the value 3 was converted to 1 if any code with a procedure property containing the letter S was present in the case. Some procedure properties were excluded from that rule, however, namely:
- 00S99 Unspecified or impossible procedure
- 99S13 Bilateral procedure
- 17S02 Endoscopy (this property will disappear 2012 according to a decision of the meeting).
- The grouper had also been hard coded so that any existing OR property with the value 3 was converted to 2 if any code with a procedure property containing the letter E, but no procedure property containing the letter S, was present in the case.
- The 22 rules for “surgical” DRGs that only demands OR=1 (no demand for surgical procprop) had been complemented with rules that makes it possible to get these DRGs by registration of greater anaesthesia, although the OR property 1 had been withdrawn. How these new rules were constructed is shown in the sheet “New rules” in the file .
- The result of the grouping of PAR showed that the inactive OR logic has a neglectible effect on the outpatient grouping. Only 3 cases out of more than 10 millions had a DRG change. The effect on inpatient grouping is shown in the sheets “Result-gross” and “Result-details” in the file .
In summary there was no negative side-effects detected with the exception that one case incorrectly was grouped to DRG 461 instead of DRG 470 but that can be avoided. The rule with 014D709 has to be complemented with a new similar rule but without S and with 00X10.
For easier maintenance of the system we recommend that the exclusions from the hard coded rule that changes OR from 3 to 1 (see above) are specified in a new definition table.
Expert Network 2011-10-27
The meeting recommended adding this problem in to the CC-grouper. Martti Virtanen should think about the problem and try to find a solution.
Expert Network 2012-03-08
At the meeting information was given concerning Martti’s own suggestion with another technical solution than Mats suggested. Martti’s solution for the DRG 530 to 540 means that the cases are grouped to surgical rest groups in each MDC. DRG 531 (A39N in the CC-grouper) must remain according to Martti. Martti’s suggestion also means that pulmonary problem and neurology problem will be handled differently.
Martti must put up his documentation in Forum. Sweden will thereafter evaluate Martti’s suggestion with Mats and Sweden might there after come with additional changes for the 2014 CC-grouper.
The case will still remain opened until Sweden has made the controls mentioned above.
Mats Fernström, NPK, Sweden 2013-02-19 (National ID = C475)*
Sweden is not at all satisfied with the model proposed last year by Nordic Casemix Centre (NCC model). Here are some examples where the grouping is totally wrong.
- All cases with cardioversion, correctly registered with the code for general anesthesia, are grouped to the surgical DRGs E35 “Övriga kärloperationer …” (former DRG 120 “Other circulatory system o. r. procedures”) with an average cost that is more than three times greater than the mean cost for the cardioversion cases.
- Cases with bronchoscopy in general anesthesia (not unusual in children) are grouped to DRG D20N “Respiratorbehandling för sjukdomar i andningsorganen” (former DRG 475A “Respiratory system diagnosis with respirator therapy”) with an average cost that is also more than three times greater than the mean cost for the bronchoscopy cases.
- Cases with x-ray or MR in general anesthesia are grouped to different surgical DRGs (where the text includes the word “operation”) with usually much higher costs than the cases with x-ray/MR.
Besides, in all the mentioned cases the DRG text is misleading. Cardioversion, x-ray and MR are not operations and bronchoscopy is certainly not respirator treatment. Furthermore, the Finnish model disguise incorrectly coded cases that should go to DRG Z60N (previously DRG 477).
In Sweden we want a model with the following grouping principles:
- Cases with greater anesthesia combined with minor surgery (procedures without OR property but with surgical or endoscopic procedure properties) shall be grouped as if the intervention would have OR property. (This is the original NordDRG principle, present in both models, and it must be retained.)
- Cases with greater anesthesia without any significant (surgical or similar) interventions (no intervention with surgical or endoscopic properties) shall be grouped to specified surgical DRG's, only if resource use (cost data) indicates that this is correct. The descriptive text of all these specified DRG's must be adjusted for the change of the content.
- All other cases with greater anesthesia (anesthesia alone or in combination with interventions without both OR property and surgical/endoscopic procedure properties) shall assigned to conservative medical DRG's according to the diagnosis/diagnoses (in contrast to the NCC model.)
- No cases should be assigned to DRG Z60N/477 because of use of anesthesia codes (which was the case in original NordDRG).For this following changes are necessary:
For point 1.
OR=1 property is removed from all anesthesia codes. They will retain the property 00X10 'General anesthesia'
Each surgical rule (not just rule for surgical DRG) is followed by a rule without demand for OR (OR=1) but demand for anesthesia (diagnosis property 00X10).
In the area for multiple trauma all diagnosis property positions are occupied in some rules and another solution is needed. If anesthesia codes are also given surgical property (00S10 'General anesthesia') and that property is used in the position secondary procedure, this can be solved.
If all interventions with a given procedure property have property OR=1, then the added rule is not needed and shall be omitted.
In situation where effect anesthesia on DRG assignment was not considered appropriate, the rule were also omitted. - this causes a low number of cases assigned to DRG Z50/Z60 (477/468)
For point 2
Like in the NCC model there is a surgical partitioning to DRG A39N (former DRG 531) for cases in MDC 01 (with the exception for infection and tumor cases?) since that is better in a cost perspective. Also like in the NCC model there is a surgical partitioning for cases in MDC 14, 16, 18 and 21.
Preliminary results of the grouping with the Swedish modified model are presented in the tables below. Compared to the grouping result with the present NCC model, 3 281 cases (0,36 %, cost outliers excluded) were grouped to a different DRG. Out of these cases, 96 % were “better” grouped, both in a medical and a cost perspective. The numbers of correct coded cases in DRG Z50/Z60 were minimal.
We believe that the Swedish modified model makes it possible to more precisely control the effect of anesthesia on the grouping process and we find the grouping results to be excellent. Therefore we already have decided to introduce the model in the Swedish version of NordDRG 2014.
Detailed grouping results can be presented at the Expert Network meeting 2013-03-11.
Martti Virtanen 2013-02-28
I have taken the freedom to edit the proposal from Sweden without changing the content of it.
The comparison to the NCC model that I designed on last minute for NordDRG 2012 SWE is only valid for Sweden. I am afraid that this is a major change also for the other countries. I am sorry I have not have time to analyze this earlier.
The notes from Sweden are obviously valid, the very simplified model I proposed does have some drawbacks.
This discussion started as the case [[http://documents.norddrg.net/issues/201]] and it is reason to remind that the original problem where 148 cases per year in Sweden assigned to DRG 468 or 477 (in 2002). Now we would have 145 cases out of which 76 are correctly coded. We have not reached much.
Expert group 2013-03-12
The expert group had a long discussion on this matter. Nordic Casemix Centre proposed that the additional procedure property could be skipped by replacing the dgprop4 (21X02 'Open wound of lower legg and ankle) at the 18 multitrauma rules with 00X10. This would mean that cases with fracture of lower legg would be accepted as part of multitrauma without an open wound of the region if anesthesia would be present. Sweden analyzed the change and found that this is acceptable since in 2011 data from Sweden it would affect only 1 case.
However, because the removal of OR=1 property from the anesthesia codes will affect all versions of NordDRG, this decision must be common. The other countries have to analyze the model before they can decide about it's use.
Martti Virtanen 2013-05-07
Norway proposes (NOR case HD-0051) that the low number of anesthesia codes existing in Norway should have OR=2.
The proposal does not indicate why the change would be done. In the old logic it would affect the outpatient grouping of cases with anesthesia but without OR=1 interventions. They would still be assigned to the surgical or outpatient intervention groups.
This problem is solved in the accepted proposal (from Sweden) by adding anesthesia based rules to all DRG's where anesthesia may have effect on grouping which is valid also for Norway.
The Norwegian anesthesia codes are linked to the existing NCSP+ anesthesia codes and will get the property 00X10. The new rules of the accepted Nordic version must be added also to the NordDRG 2014 NOR logic.
It seems unnecessary to solve the very complicated problem again in the Norwegian grouper.
Cases with greater anesthesia combined with minor surgery (procedures without OR property but with surgical or endoscopic procedure properties) shall be grouped as if the intervention would have OR property.
Cases with greater anesthesia in MDC's 1, 14, 16, 18 and 21 will be grouped to intervention DRG's even if no other intervention is performed. In MDC 1 the intervention DRG is the existing anesthesia DRG 531/A39N that is retained but cases with malignancy are not grouped to that DRG.
Other cases with greater anesthesia will be assigned to conservative DRG's if no intervention with OR=1 is performed. In this case if the procedures performed do not fit with the MDC (i.e. do not have procedure property for that MDC) the case may be assigned to DRG 477 or 468 (Z50/Z60)
OR=1 property is removed from all anesthesia codes i.e from all codes that now have and will retain the property 00X10 'General anesthesia'
Each surgical rule (not just rule for surgical DRG) is followed by a new rule that is a copy of the existing rule but without demand for OR (OR=S is removed) but demand for anesthesia (dgprop is given value 00X10 in the first empty dgprop available). The rules for DRG 468(Z50E,Z50C and Z50A) and the rules for DRG 477 (Z60N) do not need any added rule.
In the area for multiple trauma for 2 rules in short therapy (100D50049, 100D50050) and for 16 rules in inpatient care (100D50149, 100D50150, 100D50349, 100D50350, 400D51049, 400D51050, 400D51149, 400D51150, 400D51249, 400D51250, 400D51349, 400D51350, 400D51449, 400D51450) the dgprop4 value 21X02 is substituted by 00X10.
The rules for anesthesia DRG's (DRG=53xX, Dgprop=00X10) are removed (listed below with current ord values of the rules) and the DRG's become inactive.
- MDC 15, rules 015D8260, 015D8261, 015D8290
- MDC 2, rule 402D07420 (402D074207 and 402D074209 in Swedish version)
- MDC 3, rule 403D1640 (403D164009 in Swedish version)
- MDC 4, rule 404D040
- MDC 5, rule 405D210000 (405D210007 and 405D210009 in Swedish version)
- MDC 6 rule 406D1130 (406D113001, 406D113005 and 406D113009 in Swedish version)
- MDC 7 rule 407D15003 (407D130106 and 407D13010 in Swedish version)
- MDC 8 rule 408D310 (and 408D319 in Swedish version)
- MDC 9 rule 409D1323 (409D132309 in Swedish version)
- MDC 30 rule 409D51202 (Swedish version)
- MDC 10 rule 410D0930 (410D093005 and 410D093009 in Swedish version)
- MDC 11 rule 411D16980 (411D169801, 411D169805 and 411D169809 in Swedish version)
- MDC 12 rule 412D1130 (412D113009 in Swedish version)
- MDC 13 rule 413D1310 (413D131005 and 413D131009 in Swedish version)
- MDC 21 rule 422D02412 (421D081500, 421D081505 and 421D081509 in Swedish version)
For MDC 1 the current rule 401D0730 for DRG 531/A39N 'General anaesthesia for neurological problem' will remain unchanged. The rule will be moved after the current rule for DRG 011X (A43E) and the rule(s) for spinal disorders and injuries (DRG 009/A40E,A40C and A40A) currently (401D090000, 401D090005 and 401D090009 in Swedish version) is placed after the new position of the DRG 531/A39N rule.
For MDC 16 rule 418D012 (416D031109 in Swedish version) DRG will be changed 394X 'Other o. r. procedures of the blood and blood forming organ' (already changed to 394X/R05N in Swedish version)
For MDC 21 rule 421D08124 (421D081500, 421D081505 and 421D081509 in Swedish version) DRG will be changed to 443X 'Other o. r. procedures for injuries' (Already changed to 443N, 443C adn 443M /U19E, U19C and U19A in Swedish version)
MDC's 14 and 18 do not have anesthesia related rules currently. Thus the basic addition of 00X10 rules will take care of the necessary changes.