The simplified vascular staging instrument produced a single diabetes severity score which has shown to predict hospitalisation and mortality and can be applied using only data collected routinely in health practice settings including hospital admissions. It has good construct validity shown by a positive association between disease severity and hospital cost and disease severity and mortality (Figures
, ). The distribution of disease severity reflects a common observation in clinical services that the majority of people when diagnosed with diabetes are already in stage 2, having already accumulated a degree of vascular damage (Figure
It is unknown if this instrument would have good construct validity when applied to data of other population groups. The sample size is small, the cohort is young and Indigenous Australians have a lipid profile which differs from that of non-Indigenous Australians [9
]. However given the structure of the instrument, there is no reason to presume it will not prove useful in other populations.
A strength of the instrument is its use of routinely collected data from robust sources. Complex testing is not necessary to estimate vascular stage, which can be costly and physically invasive, repetitive and time consuming for participants. Annual and condition specific health checks are becoming core business of primary health care services and hospital separation data is routinely collected in many countries, which can be linked to other population data. The feasibility of this type of research will be enhanced with the adoption of electronic data linkage, which is becoming more common, following its success in Sweden [18
], Western Australia [19
] and Canada [20
]. With future increased availability of linked primary health care and hospital datasets, this instrument could be used more widely and at low cost.
A benefit of a staged disease severity approach is its ability to assign disease severity according to clinical vascular changes, rather than relying on date of diagnosis. Compelling evidence suggests that vascular changes begin to occur years prior to diabetes diagnosis [21
]. Harris et al., 1992 [21
] studied two groups of patients with diabetes, in the United States and Australia, and found that on-set of diabetic retinopathy was likely to have occurred 4–7
years before clinical diagnosis of diabetes. Furthermore other data in their study suggested that on-set of diabetes may in fact occur 9–12
years before clinical diagnosis. The staged disease severity approach based on clinical markers could capture vascular changes that may occur prior to diagnosis and potentially adjusts for the effect of lifestyle changes and medical and pharmacological advances in diabetes management that may prevent or slow diabetes-related vascular disease progression. In this respect a vascular staging instrument may be more sensitive than diabetes duration as a proxy for health status.
Comparing the performance of the NSW instrument and the simplified version was not feasible, given the more extensive data requirements of the more complex NSW instrument. Two existing severity indexes that use administrative data were found, one did not include the use of screen-detected micro/macrovascular risks[22
], and both when applied derived severity scores within complication categories [22
]; this was considerably more complex than we required. Professionals in collaboration [22
] developed a diabetes complication severity index (DCSI) consisting of a 13 point scale with 7 complication categories. Rosenzweig and colleagues [23
] index produces 6 separate scores each with 4 levels of severity. Both indexes discussed [22
] were developed to assist health organizations identify care intensity to provide targeted interventions based on diabetes disease stage, which requires a more finely tuned staging instrument than our purpose of adjusting for heath status as a possible confounder.
Clinical trials have shown that well managed type 2 diabetes can slow the rate of progression of diabetes-related vascular complications reducing the incidence of vascular events [24
]. However this promise is not always realised in primary care practice settings. One aim of primary health care services is the early detection of diabetes (and other chronic diseases) and early implementation of management to keep people in vascular severity stages one (no evidence of microvascular or macrovascular risk) or two (screen detected microvascular comorbidities or macrovascular risk factors) and prevent progression to more severe stages (irreversible vascular impairment). A clear association between the degree of vascular severity, hospital cost [26
] and mortality [17
] is established in the literature. Therefore the returns of slowing or preventing diabetes related vascular disease progression should reduce diabetes-related hospital costs and mortality, and enhance the quality of life of patients and their families.
The practical application of this instrument could be within population health research groups and health care delivery settings. At a population level the instrument may be used to inform health services planning for instance by identifying the distribution of vascular severity within a population group at a given point in time. Knowing the percent of people in each disease severity stage may help inform targeted planning of service requirements. In addition, applying the instrument repeatedly over time could allow the rate of diabetes-related vascular disease progression to be calculated and compared over time and across groups. With the inclusion of hospital separations the instrument could also be used to monitor how well a primary care program is performing in managing diabetes.
We propose to use the diabetes vascular severity staging instrument to derive a single severity score which will be used in statistical analysis as a proxy for health status of Indigenous Australian adults with type 2 diabetes. Validating the instrument in future research and testing its generalisability with similar and diverse population groups would be desirable.