Case 2003-OR-03 OR property of anaesthesia codes
|Priority:||Error correction||Spent time:||-|
|Target version:||Expert Group 2008|
|Case type:||Owner / responsible:|
|MDC:||OR||Old forum status:||CLOSITEM - Closed item|
|Target Grouper:||COMMON, DEN, EST, FIN, ICE, LAT, NOR, SWE|
Last updated: 2008-04-07
Status: CLOSITEM - Closed item
Initiated - 2003-09-17
By: Mats Fernström (CPK), Sweden
Expert Network 2004-03-30 - Postpone recommended
Steering Group 2004-04-15 - Recommendation accepted
Expert Network 2005-04-08 - Postpone recommended
Steering Group 2005-04-19 - Recommendation accepted
Expert Network 2006-03-22 - Postpone recommended
Steering Group 2006-04-19 - Recommendation accepted
Expert Network 2007-03-28 - Change recommended
Steering Group 2007-04-11 - Recommendation accepted
Expert Network 2008-03-06 - Correction accepted
Steering Group 2008-03-31 - Recommendation accepted.
Some codes for anaesthesia (in Sweden ZXH20 'Ledningsanestesi', ZXH40 'Spinalanestesi', ZXH50 'Epiduralanestesi' & ZXH60 'Generell anestesi') have the property OR 1, which was given to the codes because we wanted a surgical partitioning of minor procedures if they were performed under greater anaesthesia. With the new logic in NordDRG (version 2004) we have to discuss this idea once more. Do we really want surgical partitioning of minor procedures when the patients are staying overnight one or more nights? In these cases, I think, the cost of the procedure is only a small part of the total cost, even with anaesthesia. In the short-term cases, however, the cost of the anaesthesia is a significant part of the total cost. Therefore I want the Expert group to consider a change of the OR property of the anaesthesia codes from OR 1 to OR 2. Short-term cases with minor procedures and anaesthesia will then have surgical partitioning to a DRG in the 800-series and "long-term" cases will have medical partitioning to a regular conservative DRG.
There is at least one positive side effect of such a change. Tiny procedures, i.e. procedures without any procedure property (like bandages and casts) performed under anaesthesia (rather frequent in children) are today always allocated to DRG 477. With the suggested change they will be allocated to a DRG in the 800-series (short-term cases) or to a conservative DRG. Negative side effects?
Expert Network 2004-03-30
The meeting recommended to postpone the case for further investigation. CPK, Sweden will investigate the case further with economic analysis for next years update.
We have analysed the effect of removing the OR 1 property from the Swedish anaesthesia codes. In the cost database from year 2003 we exchanged the anaesthesia codes with OR 1 to other codes without any properties and then re-grouped the material. The method is described in detail in the appendix "Case_2003-OR-03_Tabell".
The re-grouping resulted in a DRG transfer in 469 cases (0,11 %). In most of the cases (72 %) the new DRG was better, concerning costs, than the old group. 148 of the cases had previously been grouped to DRG 468 or 477 but were now grouped to more meaningful DRGs. 77 % of them were also better grouped concerning costs. See "Case_2003-OR-03_Tabell" and use the auto filter in the headlines fore more details.
In summary the removal of OR 1 from the anaesthesia codes affects only a very small portion of the DRGs in an inpatient material. In affected cases, however, it leads to a better DRG concerning costs in most cases and the number of cases with correct primary coding in DRG 468/477 is substantially reduced.
Martti Virtanen 2005-03-01
The analysis by CPK is very interesting and extremely useful. I have modified it somewhat (see "Case_2003-OR-03_Tabell"). Case mix systems are about creating groups with as small economical variance within the group as possible. Any analysis should take this in account. In this case (as in general) it means that analysis should be concentrated on relative changes more than on absolute changes.
In the whole material the cost of the 469 cases is 21'000 SEK (66%) higher than the mean cost of the groups the cases would be moved to. The cost is 11'000 SEK (3.8%) lower than the mean of the current groups. This was the original reason for the solution NordDRG has now – the cases with anaesthesia are more expensive than cases without anaesthesia. Furthermore the disturbance in the conservative groups is larger than in the surgical groups. This is true also for the absolute difference of individual cost and DRG mean cost (85% in current groups and 102% in the new groups). The proposed change would definitely not increase the homogeneity of the DRGs although as indicated by the analysis of CPK the distance of most cases from the new DRG mean is in absolute terms smaller than from the current DRG. This is true for all MDCs with a reasonable amount of cases although in some MDCs the difference is minimal.
DRG 468 is problematic but the number of cases is so low that it is not a real problem for the system. However, it would be interesting to analyze which were the procedures that resulted in assignment to DRG 468 instead of 477. They do not have OR=1 property but are regarded as major procedures! That is probably wrong and should be corrected.
DRG 477 is clearly a problem. These differences of individual cost of these cases is in the mean 102% of the cost of other cases in DRG 477 and 125% of the cost of the conservative DRGs at issue. In MDC 01 the problem is even worse (153% vs 196%). There are 44 cases in MDC 01 and 134 cases in total.
One possible way to solve the problem could be to create a new group for cases with anaesthesia without OR=1 procedure and without procedure related to principal diagnosis MDC. Additionally, procedures without OR=1 should not have properties that are regarded as major procedures. MDC 01 could even have a group of its own of this type because the number of cases is 50 per year (468 and 477 combined). However, with the exception of a small number of extremely expensive cases there is not much difference from the other MDCs in financial terms.
CPK, Mats Fernström 2005-03-09
The main reason for the change is to reduce the number of correct coded cases in DRG 477. The most frequent complaint on NordDRG comes from pediatric clinics. On younger children they often use full anesthesia for moderate and minor procedures and they are not happy to end up in DRG 477 or 477O. They don't consider the combinations of diagnosis and procedure as "rare".
A simple way to solve the problem is to withdraw the OR 1 property from the anesthesia codes and I think that we can agree on that such a change has an almost neglectible effect on the NordDRG system in general. I am aware of the original thought that anesthesia must lead to a surgical partitioning but that old principle can be questioned since we now define inpatients in a new way. We are talking about patients staying two or more days in the hospital and then anesthesia is probably not that dominating cost driver.
Martti Virtanen 2005-03-15
Mats pointed out that relative changes are not straight forward because they are asymmetric. I fully agree, but homogeneity is a relative term and all indicators of homogeneity like V% are dependent on relative rather than absolute terms.
However, I had two main points:
DRG 468 problem.
We should check the properties of procedures not having the property OR=1 and having a procedure property that results in PROC=99S90 (which is secondary defined by a number of other procedure properties, see Case 2002-OR-05 []). This should not happen and it is most probably an error. The cases will belong to 477-group in this analysis after the correction and are included in the DRG 477 problem analysis below.
The attached table (drg 468) shows all 68 codes of this type in NCSP+. When they are corrected the national versions will be corrected simultaneously. The number of codes in the national versions may be different from NCSP+.
The options are either to change the property to another in the same MDC area that does not have Extens-property (value 0 instead of 1) or to remove the procedure property without substitution. I propose the first model but I suggest that experts in surgery check that the proposed properties are acceptable and that second option cannot be applied.
DRG 477 problem.
We should avoid DRG 477 in this type cases. Instead of interfering with the relatively well working grouping of the other cases, I would recommend 2 new DRG's.
Division to neurological and other problems seems to be important. Compared with the conservative DRG's the cases are in the mean about 10% more expensive. The number of cases is acceptable. The groups are medically meaningful. The only problem is relatively high variance. Removing 5 cases with expenses more than 600'000 SEK (6% and 2% of the cases in each group) results in much better variance.
This analysis is only about inpatients. The outpatients need to be analyzed separately, because the differences in the DRG assignment process.
Expert Network 2005-04-08
The meeting accepted that the changes related to DRG 468 problem. This means that all cases currently assigned to DRG 468 will be assigned to DRG 477.
The meeting recommended the case to be otherwise postponed for further investigation. The effects of the change of OR=1 to OR=2 for all anaesthesia codes should be compared with the possibility to create new DRG (DRGs) for independent anaesthesia treatment.
Mats Fernström 2005-12-10
About the DRG 468 problem.
The procedure AAXC04 (Temporary closure of brain artery) is an intracranial procedure. It should have OR 1 and then it is not a problem any more. Otherwise I fully support Marttis suggestion concerning all the other procedure codes.
About the DRG 477 problem.
I am a bit skeptical to Marttis suggestion. It seems to be a rather complicated way to solve the problem and it gives us another two DRGs. Besides, I doubt the medical relevance of DRG XX23 which will contain cases from all MDCs except MDC 01.
Therefore I want to discuss a third alternative to solve the DRG 477 problem. Anesthesia codes with OR 1 could have procedure properties leading to the lowest surgical DRG for every (or almost every) MDC. Cases without any other significant procedure will then be grouped to the surgical rest groups. Some of the DRG texts have to be modified but otherwise this change is rather simple to perform and there will be no new groups and all cases will remain in the correct MDC. I don’t think we have to worry about the homogeneity. The number of cases is small and rest groups will for natural reasons always be less homogenous.
Martti Virtanen 2006-03-14
This is an extremely complex problem. I have now analyzed it based on the Finnish data, which basically gives similar results as the Swedish one.
I started by grouping Finnish data with Finnish 2006 FIN-version grouper. Then I made a new version of the grouper where anesthesia codes have OR=2 property instead of OR=1. The results can be seen in the first two sheets of the Excel book . The DRG texts are in Finnish, I am sorry I did not have time to change them. The essential finding is that the weighted average of variance increases by 0.8%-points when the change is introduced. This difference mostly disappears when the data is trimmed (2-step trimming), the remaining increase is 0.1%-points.
The next step was to create a separate dataset out of those cases that changed their DRG because of the change in grouper. Then the difference of the mean costs of these cases could be compared with the mean costs of the total groups (see two last sheets of the excel book). In the average the difference of the real costs and the ‘price’ from the full data was 28% of true costs in old model and 40% in the new model (OR=2 for anesthesia). This is the reason for the slightly higher variance in the new model in the first step. Mostly the costs is lower than the average of the total group in the old model and higher in the new model. The hospital would “win” round 4.8 mil EUR from these cases if untrimmed values would be used and loose 0.6 mil EUR in trimmed values are used with the old model. Currently a partial trimming is applied, which means that the true is between these figures. With the new model the loose round 5 mil EUR both without and with trimming. Of this a little over half is from inpatient care and the rest from short therapy. Obviously the costs is transferred to prices of other patients and other DRG’s so that at the end the budget is balanced.
Thus the proposed model of changing OR-property from 1 to 2 for codes on general anesthesia and similar procedures would result in slight increase of variance and underestimation of the cost of the cases at issue. This is basically the same results as in the Swedish data last year.
The anesthesia is an important procedure as such but it also indicates other problems that increase treatment cost. To recognize anesthesia procedures as interventions that affect DRG assignment is in principle correct from medical point of view. Thus anesthesia should have a procedure property that could be used for assignment of patients in all MDC’s. Table ‘+Analysis of DRG 477 cases in Helsinki anesthesia data+’ looks at the problem from this angle. There are 4 MDC’s without any cases which means that a new DRG in each MDC is out of question. However, in 10 MDC’s the Helsinki data has an acceptable amount of cases (at least 35 in 25% of Finland) and the cost of treatment varies among these groups even after trimming enough to make separate groups reasonable. The rest of the cases form a group with acceptable homogeneity.
The basic problem of any grouping model for these cases is that this rather small group includes a substantial amount high cost intensive care cases. The six most expensive cases cost more than 100’000 EUR each when the mean is less than 5’000 EUR. Although this problem can be controlled by trimming the real solution would be to develop a grouping algorithm for intensive care. However, the cases here for example do not have any common denominator that could be used as an indicator (apart from the price).
As a solution for the anesthesia problem my proposal is that general anesthesia and similar procedures are recognized with procedure property that results in assignment to special new DRG’s in MDC’s 1, 3, 5, 6, 7, 8, 10, 11 and 16. Cases in other MDC’s are assigned to a new DRG in MDC 23. As for the short therapy grouping, at present not changes seem to be necessary. The matter will be discussed during the NordPol-project.
Expert Network, 2006-03-22
The meeting stated that the problem with anaesthesia codes has increased. Analyses on Finnish data suggested that the proposal may be relevant from a clinical point of view (description) but the resulting change in (relative) financial consequences would be significant compared to the present situation, and not necessarily in a desired direction. Analyses must consider the effects for inpatients and short therapy separately. The resulting changes in assignment to DRG 477 must be considered. General underreporting of anaesthesia codes contributes to the problem. These codes may be proxies for resource use, but there may also be other indicators that need consideration. In conclusion, the meeting felt that a solution of the anaesthesia problem is not possible without further investigation and careful thinking about introducing new grouping principles, a situation that applies to other areas too (e.g. intensive care). The meeting recommended postponing the case for further investigation.
Martti Virtanen, 2006-03-28
Since the matter is causing continuous trouble in hospitals, a solution has to be reached for NordDRG 2008. If no further information will be available I recommend that the model with special groups for anaesthesia will be used.
The properties of 74 procedures in the table DRG 468 problems will be changed as indicated in the table.
When performed without anaesthesia or other code with OR=1 property the procedures result in assignment to DRG 477 instead of 468. When the codes have MDC specific effect on DRG assignment the DRG will of lower weight than currently.
NordDRG 2006. The case will be further active. The Nordic Centre should do further analysis.
Expert Network 2007-03-28
The meeting suggested to create 7 new DRG’s.
MDC 1,5, 8 and 11 are big groups and should be under resp MDC’s.
MDC 7 and 4 are rather expensive groups and should also be under resp MDC’s.
MDC 3, 6, 10 and 16 are not so big and could lead to av postMDC drg in MDC 23.
The anaesthesia codes should keep OR 1 Otherwise the technical solution will be formulated by the Nordic Centre.
A new procedure property 00S10 ‘General anesthesia’ is given to the procedures listed below
In each MDC a new rule is created immediately after the last surgical DRG. The new rule is a copy of the last surgical rule but the procedure property is changed to 00S10. The new rules must always have OR=’S’ and may not have any diagnosis properties nor any age or sex limitations. These properties need to be changed at some places. If there are several groups of ‘S’-rules for the MDC at issue, the new rule is placed immediately after the last one.
- In MDC 01 the new rule is a copy of the current rule 401D0722, placed immediately after it (401D0730) and 01S05 is changed to 00S10.
- In MDC 04 the new rule is placed after 404D031 and may be a copy of it, but OR hast have value ‘S’ which it does not have in rule 404D031.
- In MDC 06 the new rule is placed after 406D1121 and may be a copu of it, but the diagnosis property 06X08 has to be removed from the new rule.
- In MDC 10 the new rule is placed after 410D0921 and may be a copy of it, but the diagnosis property 10X01 and the sex value ‘M’ have to be removed from the new rule
Special cases needing more changes:
- In MDC 12 the new rule has to be placed after the rule 412D1121 and it is created as a copy of this rule. In the copy the value of sex ‘M’ has to be retained.
- In MDC 13 the new rule has to be placed after the rule 413D1301. In the copy the value of sex ‘F’ has to be retained.
- In MDC 14 the new rule has to be placed after the rule 414D123. Her also the value of sex ‘F’ has to be retained.
In MDC 18, MDC 19 and MDC 23 no new rule is needed.
For neonatology (MDC 15) three new rules are placed after current rule 015D8251. The first is a copy of current rule 015D8250 but procprop1 is changed to 00S10. The second is a copy of current rule 015D8251 with the same change. The third is a copy of current rule 015D8281. Procop1 is given value 00S10 and OR is changed to ‘S’ as stated above.
7 new DRG:s are created:
The value of local DRG is the same as DRG in all national versions. The national authorities are responsible for delivering the national names for these DRG’s, in alpha version the English names may be used.
The new rules in the listed MDC’s will result in these DRG’s i.e. the value of DRG on those lines has to be changed according to the list above. All other new rules result in DRG 530 and the value of DRG is changed to 530.
Cases with anesthesia without significant procedures are assigned to one of the seven new DRG's for general anesthesia if no procedure with procedure property is performed. In specified MDC the DRG is according to the MDC, others will be assinged to DR 530 which is tabled as MDC 23 DRG although the cases have not MDC 23 principal dx.
Martti Virtanen, 2008-02-19
During the alpha testing in Finland it was found out that the rule based on procedure property will change the assignment in situations whit secproc rules. To avoid this the solution was done using a diagnosis property (00X10) given to the same codes and with the same meaning as the originally planned procedure property code (00S10).
Expert Network 2008-03-06
The meeting accepted the correction. Error corrected. The case will be closed.