We shall now explain how we assigned patients year by year to the three comparison groups for the study. We confine ourselves here to the assignment of diabetes patients. We first identified all diabetes patients and their pricing arrangements in terms of the individual insurance claims made for them in a particular year. The final assignment to groups was made on the basis of the pricing arrangement associated that year with the GP with whom they were registered.
9 Type 2 diabetes mellitus for persons aged ≥18.
10 Cardiovascular risk management, called VRM from 2011 onwards
Patients categorised initially into three groups on the basis of insurance claims Three types of diabetes patients were distinguished:
1. Patients being treated in a disease management programme reimbursed by bundled payments. They were labelled ‘bundled payment patients’.
2. Patients being treated in a disease management programme for which no bundled payment claims were submitted. They were labelled ‘management fee patients’.11
3. Patients being treated for diabetes but not in a disease management programme. They were labelled
‘care-as-usual patients’.
Figure A3.1 Time frame of the study and procedure codes potentially used to claim diabetes management programmes under NZa policy guidelines.
Procedure codes 15###
NZa Innovation policy guideline
Procedure codes 40###
NZa Bundled Payments policy guideline
Procedure codes 11###
NZa Primary Care Products policy guideline
Procedure codes 14###
2006 2007 2008 2009 2010
NZa M&I policy guideline Effective date of policy guideline
Start of study period
01/01/2007 End of study period
31/12/2010
The three types of patients were identified in the Vektis databases in different ways.
1. Bundled payment patients
Patients being treated in bundled payment disease management programmes were identified using procedure code numbers. For the 2007-2009 period, we began by retrieving the code numbers for disease management programmes of any kind deriving from the Innovation policy guideline (15###
codes). Using the descriptions provided by Vektis, two RIVM researchers then distinguished the patients by whether they were in diabetes or vascular risk management programmes and by whether these were reimbursed by bundled payments or otherwise.
For 2010, we identified patients in programmes contracted according to the Bundled Payments policy guideline (40### codes). Many claims that year still bore 15### procedure codes, as many care groups had older contracts extending into 2010. As these contracts expired and new ones took effect, care groups switched to the 40### codes (deriving from the Bundled Payments policy guideline) to claim disease management reimbursement. Table A3.1 shows the code numbers denoting diabetes or VRM programmes on a bundled payment basis that were used for this study.
Table A3.1 Procedure codes used to identify patients in diabetes and vascular risk management (VRM) programmes reimbursed by bundled payments.
Diabetes 40001 15000 15001 15004 15005 15007 15008 15009 15010 15011 15012 15013 15014 15015 15016 15017 15018 15019 15020 15022 15027 15028 15029 15030 15031 15032 15033 15034 15035 (beginning 2010) 15036 15037 15038 15039 15040 15043 15044 15045 15060 15061 15062 15064 15065 15076 15087 15088 15089 15091 15092 (Menzis policyholders in 2010) 15093 15098 15099 15100 15105 15106 15111 15118
Vascular risk 40011
2. Management fee patients
Procedure codes were also used to identify patients in disease management programmes supported by management fees under the Innovation policy guidelines. All codes flagged by two researchers as suggesting a management fee contract were submitted to that care group’s preferential insurer for verification.
Additional data from Achmea to identify management fee patients
The Achmea insurance group reimbursed GPs who had management fee contracts during the period under study via a supplement to their capitation allowances. As such patients were not identifiable by procedure codes, Achmea provided additional data files (see also section A3.1.3) that enabled Vektis to readily and reliably identify the management fee patients in its databases.
Table A3.2 Procedure codes to identify patients in diabetes and vascular risk management (VRM) programmes not reimbursed by bundled payments.
Diabetes 15021 15025 15026 15044 15075 15035 (including 2010) 15092 (except Menzis policyholders in 2010)
Vascular risk not applicable
3. Care-as-usual patients
Diabetes patients receiving care as usual were identified using two criteria. First, they were to have no insurance claims for bundled or management fees in the calendar year in question; second, they were to have claims for specified types of medicines. The determining marker for diabetes patients was the range of A10B codes in the ATC classification (Table A3.3). No care-as-usual VRM patients could be identified, since no drugs were prescribed that were uniquely linkable to VRM.
Defining comparison groups on the basis of GP classification
For all the patients thus far selected for the initial sample for a particular year, we next identified the GP with whom they were registered, using the codes for the capitation allowances claimed for those patients by GPs in that year, as recorded in Vektis’s ELIS H database. The GPs were then categorised as ‘bundled payment GPs’, ‘management fee GPs’ or ‘care-as-usual GPs’ on the basis of the types of patients (see above) that they claimed for. Bundled payment GPs were ones that had bundled payment patients in their patient population at any time within the calendar year in question. Management fee GPs were ones with management fee patients, and care-as-usual GPs were ones that had no bundled payment or manage-ment fee patients in the year in question. The GP categorisation was performed separately for each year of the study, as GPs could change categories from year to year.
Table A3.3 ATC codes used to identify patients receiving care as usual (CVZ, 2009).
Diabetes ATC code Drug names or descriptions
A10BA02 Metformin
A10BB01 Glibenclamide
A10BB03 Tolbutamide
A10BB09 Gliclazide
A10BB12 Glimepiride
A10BD02 Metformin and glibenclamide
A10BD03 Metformin and rosiglitazone
A10BD04 Glimepiride and rosiglitazone
A10BD05 Metformin and pioglitazone
A10BD07 Metformin and sitagliptin
A10BD08 Metformin and vildagliptin
A10BF01 Acarbose
A10BG02 Rosiglitazone
A10BG03 Pioglitazone
A10BH01 Sitagliptin
A10BH02 Vildagliptin
A10BX02 Repaglinide
A10BX04 Exenatide
Vascular risk management not applicable
under a bundled payment arrangement. In other words, our bundled payment comparison group could contain both bundled payment patients and care-as-usual patients. Similarly, the management fee comparison group could also contain some CAU patients. The care-as-usual comparison group consisted solely of patients receiving care as usual. Figure A3.2 depicts the procedure by which we assigned patients to comparison groups.