View Article PDF

The Virtual Diabetes Register (VDR) is an annually updated national register of all patients with diabetes mellitus. The VDR was designed to monitor the prevalence of diabetes in New Zealand and support quality improvement initiatives.1,2 The register is compiled from publicly available health data based on patients’ use of health services, including hospitalisation records, laboratory test results and pharmaceutical dispensing information. As of 2018, the VDR estimates that there are 253,480 patients with diabetes in New Zealand.1 However, this is lower than the number forecast prior to 2017, due to a change in the algorithm used to define the VDR population.3

Although the VDR is a valuable tool in that it allows for the description of diabetes prevalence by geography, age, ethnicity and gender,3,4 it does also have limitations. Currently, the VDR does not contain diagnosis information, and thus it is unable to discriminate between the different types of diabetes, namely type 1 diabetes (T1D) and the more common type 2 diabetes (T2D). This poses a problem if VDR data is to be used for research purposes, as often times a clinically defined population (eg, patients with T1D) is required for analysis. In some cases, this can be circumvented via the use of clinically confirmed patient registers, though these are generally restricted to a few individual district health board (DHB) databases that contain limited current up to date data rather than a national dataset. Hence algorithms using additional clinical data to differentiate between T1D and T2D are being developed to define these subset populations from the VDR.  However, differentiating between T1D and T2D can be difficult, as T2D is becoming more common in youth and young adults and the onset of T1D in adulthood is becoming increasingly recognised.5 Further, approximately 10% of patients with T1D will have negative islet cell antibody titres, and a similar proportion of patients with T2D will have mildly elevated autoantibody titres.5 Consequently, misclassification of the type of diabetes is common in the community. Approximately one third of patients with T2D are misclassified as T1D,6 and approximately 40% of adults with T1D are misdiagnosed with T2D.7

One algorithm designed to identify patients with T1D was published by McKergow et al in 2017,8 and it has been used to define the T1D patient population in further work with VDR data.9 However, the diagnostic accuracy of the McKergow algorithm has not been validated against a confirmed T1D clinical register. Hence, the aim of this study was to determine the accuracy of the McKergow algorithm by using it to predict which patients in the 2017 VDR dataset had T1D as compared to a known population of patients with confirmed T1D at the Waikato Regional Diabetes Services at Waikato District Health Board (WDHB) for the same time period. Ethics approval was granted for this study by the Health and the Disability Ethics Committee (ref: 17/NTB/222).

The Waikato VDR dataset used in this study included all patients registered in the national dataset and domiciled in the Waikato DHB region during 2017. As per the McKergow algorithm,8 this included patients who died during 2017 and those not enrolled in a primary health organisation (PHO) (n=23,211). As such, this population is larger than the 21,767 reported on the VDR website,1 as the latter excludes patients who are not alive or not enrolled in a PHO as of 31 December 2017.

The WDHB diabetes clinical register included 1,303 patients who have all had T1D diagnosed by an endocrinologist.9 Patients in the initial WDHB dataset who were not part of the VDR extract were excluded (n=85), such that 1,218 patients were included in this study.

The McKergow algorithm was applied to the VDR data as described previously8 but with the inclusion of additional data for 2014–2017 to account for the time period of study. This included diabetes medication dispensing data (1 January 2006–31 December 2017), hospital diagnosis and discharge data (1 January 1988–31 December 2017) and death records (2017) to determine the number of patients with T1D (Figure 1). Although not defined in the McKergow study, dispensed insulin in our study included insulin glargine, insulin isophane with insulin neutral, insulin aspart with insulin aspart protamine, insulin neutral, insulin zinc suspension, insulin neutral, insulin lispro, insulin glulisine, insulin aspart, insulin isophane and insulin lispro with insulin lispro protamine. Similarly, oral hypoglycaemics and alpha glucosidase inhibitors included metformin hydrochloride, vildagliptin with metformin hydrochloride, glibenclamide, gliclazide, glipizide, vildagliptin, pioglitazone and acarbose. Linkage between the datasets was undertaken using master national health index (NHI) numbers.

Summary

Abstract

Aim

Method

Results

Conclusion

Author Information

Lynne Chepulis: Senior Research Fellow, Waikato Medical Research Centre, University of Waikato, Hamilton. Christopher Mayo: Medical Student, Faculty of Medical and Health Sciences, University of Auckland, Auckland. Ryan Paul: Endocrinologist, Waikato Regional Diabetes Service, Waikato District Health Board, Hamilton.

Acknowledgements

Correspondence

Lynne Chepulis, Medical Research Centre, University of Waikato, Private Bag 3015, Hamilton

Correspondence Email

lynnec@waikato.acz.nz

Competing Interests

Nil.

Ministry of Health. Virtual Diabetes Register. Available from https://www.health.govt.nz/our-work/diseases-and-conditions/diabetes/about-diabetes/virtual-diabetes-register-vdr [accessed April 2020]. 2019.

2. Jo E., Drury P. Development of a virtual diabetes register using information technology in New Zealand. Healthcare Informatics Research, 2015. 21(1): p. 49-55.

3. Chan WC, Papaconstantinou D, Lee M et al., Can administrative health utilisation data provide an accurate diabetes prevalence estimate for a geographical region? Diabetes Research and Clinical Practice, 2018. 139: p. 59-71.

4. Coppell KJ, Mann J, Willians SM et al., Prevalence of diagnosed and undiagnosed diabetes and prediabetes in New Zealand: findings from the 2008/09 Adult Nutrition Survey. New Zealand Medical Journal, 2013. 126(1370): p. 23-42.

5. Oram RA, Patel K, Hill A et al., A type 1 diabetes genetic risk score can aid discrimination between type 1 and type 2 diabetes in young adults. Diabetes Care, 2016. 39(3): p. 337-344.

6. De Lusignan S, Sadek N, Mulnier H et al., Miscoding, misclassification and misdiagnosis of diabetes in primary care. Diabetic Medicine, 2012. 29(2): p. 181-189.

7. Thomas NJ, Lynam A, Hill AV et al., Type 1 diabetes defined by severe insulin deficiency occurs after 30 years of age and is commonly treated as type 2 diabetes. Diabetologia, 2019. 62(7): p. 1167-1172.

8. McKergow E, Parkin L, Barson DJ et al., Demographic and regional disparities in insulin pump utilization in a setting of universal funding: a New Zealand nationwide study. Acta Diabetologica, 2017. 54(1): p. 63-71.

9. Tamatea J, Chepulis L, Goldsmith J, et al., Glycaemic control across the lifespan in a cohort of New Zealand patients with type 1 diabetes mellitus. Internal Medicine Journal (online) https://doi.org/10.1111/imj.14816, 2020.

10. Wheeler BJ, Braund R, Galland B et al., District health board of residence, ethnicity and socioeconomic status all impact publicly funded insulin pump uptake in New Zealand patients with type 1 diabetes. New Zealand Medical Journal 2019. 132(1491): p. 78-89.

Contact diana@nzma.org.nz
for the PDF of this article

View Article PDF

The Virtual Diabetes Register (VDR) is an annually updated national register of all patients with diabetes mellitus. The VDR was designed to monitor the prevalence of diabetes in New Zealand and support quality improvement initiatives.1,2 The register is compiled from publicly available health data based on patients’ use of health services, including hospitalisation records, laboratory test results and pharmaceutical dispensing information. As of 2018, the VDR estimates that there are 253,480 patients with diabetes in New Zealand.1 However, this is lower than the number forecast prior to 2017, due to a change in the algorithm used to define the VDR population.3

Although the VDR is a valuable tool in that it allows for the description of diabetes prevalence by geography, age, ethnicity and gender,3,4 it does also have limitations. Currently, the VDR does not contain diagnosis information, and thus it is unable to discriminate between the different types of diabetes, namely type 1 diabetes (T1D) and the more common type 2 diabetes (T2D). This poses a problem if VDR data is to be used for research purposes, as often times a clinically defined population (eg, patients with T1D) is required for analysis. In some cases, this can be circumvented via the use of clinically confirmed patient registers, though these are generally restricted to a few individual district health board (DHB) databases that contain limited current up to date data rather than a national dataset. Hence algorithms using additional clinical data to differentiate between T1D and T2D are being developed to define these subset populations from the VDR.  However, differentiating between T1D and T2D can be difficult, as T2D is becoming more common in youth and young adults and the onset of T1D in adulthood is becoming increasingly recognised.5 Further, approximately 10% of patients with T1D will have negative islet cell antibody titres, and a similar proportion of patients with T2D will have mildly elevated autoantibody titres.5 Consequently, misclassification of the type of diabetes is common in the community. Approximately one third of patients with T2D are misclassified as T1D,6 and approximately 40% of adults with T1D are misdiagnosed with T2D.7

One algorithm designed to identify patients with T1D was published by McKergow et al in 2017,8 and it has been used to define the T1D patient population in further work with VDR data.9 However, the diagnostic accuracy of the McKergow algorithm has not been validated against a confirmed T1D clinical register. Hence, the aim of this study was to determine the accuracy of the McKergow algorithm by using it to predict which patients in the 2017 VDR dataset had T1D as compared to a known population of patients with confirmed T1D at the Waikato Regional Diabetes Services at Waikato District Health Board (WDHB) for the same time period. Ethics approval was granted for this study by the Health and the Disability Ethics Committee (ref: 17/NTB/222).

The Waikato VDR dataset used in this study included all patients registered in the national dataset and domiciled in the Waikato DHB region during 2017. As per the McKergow algorithm,8 this included patients who died during 2017 and those not enrolled in a primary health organisation (PHO) (n=23,211). As such, this population is larger than the 21,767 reported on the VDR website,1 as the latter excludes patients who are not alive or not enrolled in a PHO as of 31 December 2017.

The WDHB diabetes clinical register included 1,303 patients who have all had T1D diagnosed by an endocrinologist.9 Patients in the initial WDHB dataset who were not part of the VDR extract were excluded (n=85), such that 1,218 patients were included in this study.

The McKergow algorithm was applied to the VDR data as described previously8 but with the inclusion of additional data for 2014–2017 to account for the time period of study. This included diabetes medication dispensing data (1 January 2006–31 December 2017), hospital diagnosis and discharge data (1 January 1988–31 December 2017) and death records (2017) to determine the number of patients with T1D (Figure 1). Although not defined in the McKergow study, dispensed insulin in our study included insulin glargine, insulin isophane with insulin neutral, insulin aspart with insulin aspart protamine, insulin neutral, insulin zinc suspension, insulin neutral, insulin lispro, insulin glulisine, insulin aspart, insulin isophane and insulin lispro with insulin lispro protamine. Similarly, oral hypoglycaemics and alpha glucosidase inhibitors included metformin hydrochloride, vildagliptin with metformin hydrochloride, glibenclamide, gliclazide, glipizide, vildagliptin, pioglitazone and acarbose. Linkage between the datasets was undertaken using master national health index (NHI) numbers.

Summary

Abstract

Aim

Method

Results

Conclusion

Author Information

Lynne Chepulis: Senior Research Fellow, Waikato Medical Research Centre, University of Waikato, Hamilton. Christopher Mayo: Medical Student, Faculty of Medical and Health Sciences, University of Auckland, Auckland. Ryan Paul: Endocrinologist, Waikato Regional Diabetes Service, Waikato District Health Board, Hamilton.

Acknowledgements

Correspondence

Lynne Chepulis, Medical Research Centre, University of Waikato, Private Bag 3015, Hamilton

Correspondence Email

lynnec@waikato.acz.nz

Competing Interests

Nil.

Ministry of Health. Virtual Diabetes Register. Available from https://www.health.govt.nz/our-work/diseases-and-conditions/diabetes/about-diabetes/virtual-diabetes-register-vdr [accessed April 2020]. 2019.

2. Jo E., Drury P. Development of a virtual diabetes register using information technology in New Zealand. Healthcare Informatics Research, 2015. 21(1): p. 49-55.

3. Chan WC, Papaconstantinou D, Lee M et al., Can administrative health utilisation data provide an accurate diabetes prevalence estimate for a geographical region? Diabetes Research and Clinical Practice, 2018. 139: p. 59-71.

4. Coppell KJ, Mann J, Willians SM et al., Prevalence of diagnosed and undiagnosed diabetes and prediabetes in New Zealand: findings from the 2008/09 Adult Nutrition Survey. New Zealand Medical Journal, 2013. 126(1370): p. 23-42.

5. Oram RA, Patel K, Hill A et al., A type 1 diabetes genetic risk score can aid discrimination between type 1 and type 2 diabetes in young adults. Diabetes Care, 2016. 39(3): p. 337-344.

6. De Lusignan S, Sadek N, Mulnier H et al., Miscoding, misclassification and misdiagnosis of diabetes in primary care. Diabetic Medicine, 2012. 29(2): p. 181-189.

7. Thomas NJ, Lynam A, Hill AV et al., Type 1 diabetes defined by severe insulin deficiency occurs after 30 years of age and is commonly treated as type 2 diabetes. Diabetologia, 2019. 62(7): p. 1167-1172.

8. McKergow E, Parkin L, Barson DJ et al., Demographic and regional disparities in insulin pump utilization in a setting of universal funding: a New Zealand nationwide study. Acta Diabetologica, 2017. 54(1): p. 63-71.

9. Tamatea J, Chepulis L, Goldsmith J, et al., Glycaemic control across the lifespan in a cohort of New Zealand patients with type 1 diabetes mellitus. Internal Medicine Journal (online) https://doi.org/10.1111/imj.14816, 2020.

10. Wheeler BJ, Braund R, Galland B et al., District health board of residence, ethnicity and socioeconomic status all impact publicly funded insulin pump uptake in New Zealand patients with type 1 diabetes. New Zealand Medical Journal 2019. 132(1491): p. 78-89.

Contact diana@nzma.org.nz
for the PDF of this article

View Article PDF

The Virtual Diabetes Register (VDR) is an annually updated national register of all patients with diabetes mellitus. The VDR was designed to monitor the prevalence of diabetes in New Zealand and support quality improvement initiatives.1,2 The register is compiled from publicly available health data based on patients’ use of health services, including hospitalisation records, laboratory test results and pharmaceutical dispensing information. As of 2018, the VDR estimates that there are 253,480 patients with diabetes in New Zealand.1 However, this is lower than the number forecast prior to 2017, due to a change in the algorithm used to define the VDR population.3

Although the VDR is a valuable tool in that it allows for the description of diabetes prevalence by geography, age, ethnicity and gender,3,4 it does also have limitations. Currently, the VDR does not contain diagnosis information, and thus it is unable to discriminate between the different types of diabetes, namely type 1 diabetes (T1D) and the more common type 2 diabetes (T2D). This poses a problem if VDR data is to be used for research purposes, as often times a clinically defined population (eg, patients with T1D) is required for analysis. In some cases, this can be circumvented via the use of clinically confirmed patient registers, though these are generally restricted to a few individual district health board (DHB) databases that contain limited current up to date data rather than a national dataset. Hence algorithms using additional clinical data to differentiate between T1D and T2D are being developed to define these subset populations from the VDR.  However, differentiating between T1D and T2D can be difficult, as T2D is becoming more common in youth and young adults and the onset of T1D in adulthood is becoming increasingly recognised.5 Further, approximately 10% of patients with T1D will have negative islet cell antibody titres, and a similar proportion of patients with T2D will have mildly elevated autoantibody titres.5 Consequently, misclassification of the type of diabetes is common in the community. Approximately one third of patients with T2D are misclassified as T1D,6 and approximately 40% of adults with T1D are misdiagnosed with T2D.7

One algorithm designed to identify patients with T1D was published by McKergow et al in 2017,8 and it has been used to define the T1D patient population in further work with VDR data.9 However, the diagnostic accuracy of the McKergow algorithm has not been validated against a confirmed T1D clinical register. Hence, the aim of this study was to determine the accuracy of the McKergow algorithm by using it to predict which patients in the 2017 VDR dataset had T1D as compared to a known population of patients with confirmed T1D at the Waikato Regional Diabetes Services at Waikato District Health Board (WDHB) for the same time period. Ethics approval was granted for this study by the Health and the Disability Ethics Committee (ref: 17/NTB/222).

The Waikato VDR dataset used in this study included all patients registered in the national dataset and domiciled in the Waikato DHB region during 2017. As per the McKergow algorithm,8 this included patients who died during 2017 and those not enrolled in a primary health organisation (PHO) (n=23,211). As such, this population is larger than the 21,767 reported on the VDR website,1 as the latter excludes patients who are not alive or not enrolled in a PHO as of 31 December 2017.

The WDHB diabetes clinical register included 1,303 patients who have all had T1D diagnosed by an endocrinologist.9 Patients in the initial WDHB dataset who were not part of the VDR extract were excluded (n=85), such that 1,218 patients were included in this study.

The McKergow algorithm was applied to the VDR data as described previously8 but with the inclusion of additional data for 2014–2017 to account for the time period of study. This included diabetes medication dispensing data (1 January 2006–31 December 2017), hospital diagnosis and discharge data (1 January 1988–31 December 2017) and death records (2017) to determine the number of patients with T1D (Figure 1). Although not defined in the McKergow study, dispensed insulin in our study included insulin glargine, insulin isophane with insulin neutral, insulin aspart with insulin aspart protamine, insulin neutral, insulin zinc suspension, insulin neutral, insulin lispro, insulin glulisine, insulin aspart, insulin isophane and insulin lispro with insulin lispro protamine. Similarly, oral hypoglycaemics and alpha glucosidase inhibitors included metformin hydrochloride, vildagliptin with metformin hydrochloride, glibenclamide, gliclazide, glipizide, vildagliptin, pioglitazone and acarbose. Linkage between the datasets was undertaken using master national health index (NHI) numbers.

Summary

Abstract

Aim

Method

Results

Conclusion

Author Information

Lynne Chepulis: Senior Research Fellow, Waikato Medical Research Centre, University of Waikato, Hamilton. Christopher Mayo: Medical Student, Faculty of Medical and Health Sciences, University of Auckland, Auckland. Ryan Paul: Endocrinologist, Waikato Regional Diabetes Service, Waikato District Health Board, Hamilton.

Acknowledgements

Correspondence

Lynne Chepulis, Medical Research Centre, University of Waikato, Private Bag 3015, Hamilton

Correspondence Email

lynnec@waikato.acz.nz

Competing Interests

Nil.

Ministry of Health. Virtual Diabetes Register. Available from https://www.health.govt.nz/our-work/diseases-and-conditions/diabetes/about-diabetes/virtual-diabetes-register-vdr [accessed April 2020]. 2019.

2. Jo E., Drury P. Development of a virtual diabetes register using information technology in New Zealand. Healthcare Informatics Research, 2015. 21(1): p. 49-55.

3. Chan WC, Papaconstantinou D, Lee M et al., Can administrative health utilisation data provide an accurate diabetes prevalence estimate for a geographical region? Diabetes Research and Clinical Practice, 2018. 139: p. 59-71.

4. Coppell KJ, Mann J, Willians SM et al., Prevalence of diagnosed and undiagnosed diabetes and prediabetes in New Zealand: findings from the 2008/09 Adult Nutrition Survey. New Zealand Medical Journal, 2013. 126(1370): p. 23-42.

5. Oram RA, Patel K, Hill A et al., A type 1 diabetes genetic risk score can aid discrimination between type 1 and type 2 diabetes in young adults. Diabetes Care, 2016. 39(3): p. 337-344.

6. De Lusignan S, Sadek N, Mulnier H et al., Miscoding, misclassification and misdiagnosis of diabetes in primary care. Diabetic Medicine, 2012. 29(2): p. 181-189.

7. Thomas NJ, Lynam A, Hill AV et al., Type 1 diabetes defined by severe insulin deficiency occurs after 30 years of age and is commonly treated as type 2 diabetes. Diabetologia, 2019. 62(7): p. 1167-1172.

8. McKergow E, Parkin L, Barson DJ et al., Demographic and regional disparities in insulin pump utilization in a setting of universal funding: a New Zealand nationwide study. Acta Diabetologica, 2017. 54(1): p. 63-71.

9. Tamatea J, Chepulis L, Goldsmith J, et al., Glycaemic control across the lifespan in a cohort of New Zealand patients with type 1 diabetes mellitus. Internal Medicine Journal (online) https://doi.org/10.1111/imj.14816, 2020.

10. Wheeler BJ, Braund R, Galland B et al., District health board of residence, ethnicity and socioeconomic status all impact publicly funded insulin pump uptake in New Zealand patients with type 1 diabetes. New Zealand Medical Journal 2019. 132(1491): p. 78-89.

Contact diana@nzma.org.nz
for the PDF of this article

Subscriber Content

The full contents of this pages only available to subscribers.

LOGINSUBSCRIBE