Author Article

Luc te Marvelde
Research Fellow, Cancer Council Victoria

Linked statewide data provides insight into cancer treatment and outcomes in Victoria 15/11/2018


Publication date Thursday 15 November 2018
2 minute read

Luc te Marvelde


Linked statewide data provides insight into cancer treatment and outcomes in Victoria 

Understanding variations in patterns of cancer care and associated outcomes is instrumental to identifying areas for improvement in outcomes for Victorians diagnosed with cancer. Linked population-based statewide data can play an important role in identifying variations in care and outcomes.
 
There are various routinely collected statewide data sources, which include relevant data on cancer patients. The population-based Victorian Cancer Registry (VCR) provides comprehensive, accurate information on cancer incidence. Pathology laboratories, hospitals and certain? screening programs are required to notify the VCR of new cancer diagnoses. The VCR includes data on patient and tumour characteristics, including cancer site, morphology, summary stage (restricted to some cancers), age at diagnosis, sex and residential address. 

Data on treatment, complications and survival is available as in-patient (VAED) or out-patient (VINAH) activity for public and private Victorian health services, presentations to public hospital emergency departments (VEMD), public & private radiation therapy (VRMDS), and deaths (VDI). These datasets and over 20 others are routinely linked by the Centre for Victorian Data Linkage at the Department of Health and Human Services. Data linkage allows the retrospective reconstruction of treatment pathways, complications and survival for each Victorian diagnosed with cancer. Outcomes for patients can be compared according to age, sex, socio-economicposition, geographic residential location or the health service that provided the treatment. Admissions due to complications in health services other than the health service that provided the treatment are captured. Thus, linked data includes information on diagnosis, treatment and outcome on Victorians diagnosed with cancer and can be used to identify variations in treatment and outcomes.
 
The use of administrative data for research purposes is often criticised due to its lack of clinical detail and potential for exaggerating complexity of the episode for the purpose of receiving additional funding (‘gaming’). ‘Gaming’ is mitigated by a routine VAED audit process. The lack of clinical details can have many causes. For example, nationally agreed hospital coding rules mean that only selected comorbidities are coded in the hospital data. As a result, comorbidities are underestimated. Further, extracting structured data from pathology reports is time consuming and hence tumour pathology data in the VCR is currently restricted to selected data items for selected tumour streams. Other limitations include the absence of interstate data, the type of chemotherapy received (captured in the Pharmaceutical Benefits Scheme; PBS) and radiology (captured in the Medicare Benefits Schedule; MBS). In some instances, these data limitations restrict the interpretation of the variation and local confirmation from medical records data is needed.
 
Many of these limitations can be addressed by changing coding practices or current processes. For example, advances in ‘natural language processing’ allows (semi) automatic extraction of structured data from free-text digital pathology reports. This technology will be available to the VCR in the near future. The lack of comorbidity data is now partly addressed with the introduction of new codes in ICD-10-AM Version 9. These new codes relate to 29 chronic conditions which, from July 2015 onwards, have to be coded if patients are known to have any of these conditions. Finally, the DHHS is working towards including PBS and MBS data in the linked dataset in the future.
 
In summary, a wealth of statewide linked data is available to be used to identify variations in patterns of care and outcomes for Victorians diagnosed with cancer. Given the data limitations some care has to be taken in interpreting observed variations, but at a minimum administrative data can guide prioritisation of local investigations, with the ultimate goal of improving outcomes for cancer patients in Victoria.

Back to top

 


 

 

 

print text size increase