Diabetes mellitus currently affects about 285 million adults worldwide, and this figure is expected to rise to over 400 million adults by 2030 [¹]. Based on self-reported data, the prevalence of diagnosed diabetes among Australian adults is 4.4% [²]. It is possible that the true prevalence is as much as twice that, and likely to increase further given an aging population, more sedentary lifestyles, rising rates of obesity, and a reduction in the rates of diabetes-related mortality [^3-5].

The day to day management of diabetes is demanding and can take a heavy psychological and social toll, which may in turn result in poor control of blood glucose levels and an increased risk of complications [⁶,⁷]. From the patient's perspective, minimising the burden imposed by diabetes requires an approach that ensures services are integrated, accessible and affordable. They should also be patient-centred, with a strong emphasis on supporting patients' confidence and ability to effectively manage their illness [⁸,⁹]. To this end, patient reported outcomes such as quality of life and assessments of quality of care are becoming more widely used indicators of health care systems, and are now commonly considered to be critical to the evaluation of the responsiveness of health systems in meeting the needs of their users [^10-12].

There is a growing literature on the interaction between various patient-reported outcomes, demographic factors, the self-management of patients with chronic illnesses, and medical outcomes [¹³,¹⁴]. For instance, diabetes patients with higher levels of active self-management enjoy better health outcomes [¹⁵,¹⁶]; more engaged, informed, confident, and skilled patients are more likely to perform activities that will promote their own health, and are more likely to have their health care needs met [¹⁷]. Fostering patients to take on a meaningful role in their own care is therefore central to improving quality of care and health outcomes.

There is a need to better understand the realities of living with diabetes in order to tailor adequate and appropriate medical and psychosocial interventions [¹⁸,¹⁹]. Many studies have focused on quality of life [²⁰,²¹], patient activation [¹⁵,¹⁶], resource utilisation [²²,²³], or the clinical aspects of diabetes [²⁴,²⁵], but there has not been a concerted effort to simultaneously address all of these in order to gain a more holistic understanding. The Living With Diabetes Study (LWDS) described here extends the focus of previous research into diabetes beyond medical endpoints to encapsulate a broader range of outcomes that contribute to good health and improved quality of life. In particular the LWDS considers: how diabetes affects participants' quality of life, including their mental health and well-being; how satisfied people with diabetes are with the range of health services they use; how people with diabetes manage their condition; and the natural progression of diabetes over time. The findings from the LWDS will provide a more comprehensive picture of the everyday experiences of people living with diabetes and inform health policy planning and service delivery.

This paper describes the study rationale and procedures for the LWDS and reports the baseline characteristics of a cohort of 3951 people with diabetes. The approach of the LWDS differs from other national and international studies of diabetes in its focus on examining the natural trajectory of diabetes and its treatment from a psychosocial perspective. It is hoped that this will lead to a greater understanding of how to improve the life and quality of care of people with diabetes. The large cohort will allow for the undertaking of multivariate statistical analyses, and the longitudinal nature of the data enables the investigation of temporal effects.

Wherever possible standardised scales and questions were used in the LWDS questionnaire to measure the domains of interest. We relied on previous health services research and psychosocial research to select measures sensitive to change and with adequate face, content and construct validity. The importance of distinguishing between health-related quality of life and disease-specific quality of life has been highlighted by previous researchers [¹¹,⁵⁰]. On this basis, both are measured in the LWDS. Health-related quality of life measures a patient's symptoms and functioning. The EQ-5D is one of the most widely used preference-based measures of health-related quality of life [⁵¹]. Disease-specific quality of life, which captures the broader multidimensional, subjective and dynamic features of quality of life, relates more specifically to a person's perception of how a specific disease has impacted on their life [⁵²]. Disease-specific measurement instruments include those aspects of life considered to be the most important by patients and clinicians resulting in a more detailed assessment of the issues and concerns relevant to the specific disease, its treatments and complications [¹³,³⁷,⁵³]. Several recent reviews of diabetes-specific quality of life instruments conclude that there is good evidence that the ADDQoL, used in the LWDS, is a reliable measure of disease-specific quality of life with good face and content validity [^52-55]. A recent assessment of 37 measures of patient-assessed quality of care designed for use with people with chronic illness found the PACIC, used in the LWDS, to be the most appropriate as determined by its psychometric properties and perceived applicability and relevance [⁴¹]. The inclusion of these self-reported patient assessments of quality of life and quality of care will allow for the reliable assessment of the impact that the progression of diabetes has on these important health and well-being outcomes.

The LWDS study has the inherent limitations of self-report surveys. The reliability and validity of self-report health measures varies across behaviours and outcomes. Self-reported health service utilisation data is subject to recall bias and underreporting, especially for older adults and frequent users of primary care [⁵⁶,⁵⁷]. Cross-referencing the self-report service utilisation data with MBS data will improve the reliability and validity of this information. Self-report data on health information, such as co-morbidities, has been found to be of variable quality, but is generally satisfactory for well-known conditions [⁵⁸,⁵⁹]. Previous research has found a positive, albeit weak, correlation between self-reported HbA1c values and medical record data [⁶⁰]. Self-report of treatment types including oral agents and insulin use are generally valid [⁶¹].

While recruitment at baseline was slightly lower than anticipated, it is similar to that of other studies of this nature [⁶²] and is consistent with research showing that participation rates in large cohort studies appear to be decreasing. It is estimated that rates have declined from about 80% to 30% or 40% over the past several decades [⁶³]. Effective participant retention is vital to the success of the LWDS and we have instituted an anti-attrition strategy to ensure as many cohort members as possible remain in the study. Past research suggests that the attrition in follow-up for postal surveys can be decreased by the provision of a monetary incentive such as a lottery ticket or a prize [⁶⁴], with the magnitude less important than an incentive per se [⁶⁵]. Although losses to follow-up are inevitable, the study's comprehensive retention strategy to date has been successful in limiting participant drop-out. However, it is possible that those individuals whose health deteriorates more markedly over time will discontinue study participation, biasing findings towards healthier participants.

Use of a national disease register with high coverage of the target population in the recruitment process has been effective for enrolling a large representative sample of people with diabetes, covering a broad range of socio-demographic and clinical characteristics. Specifically, the proportion of males was consistent with estimates provided by Australia's National Health Survey that 56% of Australians with diabetes are male [²]. The large proportion of LWDS participants with low levels of educational attainment is not unexpected given the older age distribution of the sample. Similarly, it is unsurprising that a large proportion of the sample (44.3%) was retired and one in 15 reported being unable to work.

In 2005-06 the average gross annual household income in Australia was around $68,000 (AUD) [⁶⁶]. While it is difficult to make direct comparisons based on the income brackets used in the LWDS, the finding that 73.7% of participants live in households earning less than $60,000 (AUD) per annum indicates that overall the sample are more economically disadvantaged than the general population of Australians. This differential is consistent with what would be expected given the previously reported relationship between low socio-economic status and prevalence of diabetes [⁶⁷,⁶⁸] as well as the high levels of retirement among the LWDS participants.

A similar pattern holds for health concession cards. In Australia low income is one of the criteria for eligibility for a government health concession card. Possession of the card can significantly reduce the out-of-pocket expenses for consumers through subsidised medical care, hospital treatment and some medications. Australia's 2004-05 National Health Survey reported 35% of persons 15 years and over were covered by a government health card, a figure considerably lower than the 62.0% of LWDS participants reporting that they hold a health care card. The higher LWDS proportion will most likely be accounted for by the older age and poorer health status of people with diabetes.

Another area in which the study is not representative of the general population relates to the proportion of Indigenous Australians. Research shows that diabetes is more common among Indigenous Australians [⁶⁹] than among their non-Indigenous counterparts. However, with census data showing that 2.5% of the Australian population report an Indigenous background [⁷⁰] and only 1.8% of the study population reporting that they are of Indigenous origin the LWDS cannot be generalised to this population.

Less than one in 20 LWDS participants relied on diet alone as the treatment pathway for their diabetes. Previous researchers have suggested that people managing their diabetes through diet and exercise alone may not have a high level of need for the NDSS's services and would therefore be less likely to register with the scheme [²⁶]. Our findings support this assumption. Less than 5% of the LWDS sample reported having a diagnosis of type 1 diabetes. The true prevalence of type 1 diabetes in Australia is estimated to be approximately 10% [²]. The underrepresentation of patients with type 1 diabetes is attributed to the lower likelihood of NDSS registrants with type 1 diabetes consenting to participate in research as there is not systematic updating of research consent status at the age of 18 among those registered as a child [²⁸]. We acknowledge this as a weakness of the LWDS with implications for the generalisability of the study to people with type 1 diabetes. Future analyses of the data will stratify by diabetes type.

On the other hand, the diabetes complications reported by the LWDS participants at baseline were very much in keeping with the available Australian statistics for the Australian diabetic population as a whole [⁷¹]. For example, 2.1% of respondents reported having foot ulcers, with the available comparable figure for the entire Australian diabetic population also 2.1%. Additionally, 6.2% of the sample, compared to 6.3% of the Australian diabetic population, reported kidney disease; 5.1% of the sample compared to 5.0% of the Australian diabetic population reported having had a stroke; 8.8% of the sample versus 8.6% of the Australian diabetic population reported neuropathy; and 41.4% of men on the study reported erectile dysfunction as a complication, while 30.2% of the general male diabetic population had the same complaint. The discrepancy for erectile dysfunction may stem from the differing data collection methods used to obtain the estimates.

Lifestyle changes constitute an important aspect of the management of diabetes, in particular type 2 diabetes. For example, obesity in those with type 2 diabetes complicates management of the disease by increasing insulin resistance and with that, blood glucose concentrations. This however, is reversible, such that weight loss of just 5% of body weight may improve insulin sensitivity and glycaemic control [⁷²,⁷³]. Research undertaken with US adults found that those with diabetes were more likely to be physically inactive (61%) than those without diabetes (42%) [⁷⁴]. Recent comparable physical activity rates for the Australian general population are not available. We can however compare the LWDS with the Australian Diabetes, Obesity and Lifestyle Study (AusDiab) conducted in1999 -2000 which found that 59.3% of Australians with diabetes were insufficiently active [⁷⁵], which is higher than the 50.7% found for the LWDS sample. Recent suggestions of a trend towards increased levels of physical activity among Australians may explain the difference [⁷⁶]. It is well established that on average people with diabetes are more likely to be overweight than people without diabetes. Both the LWDS and the AusDiab study [⁷¹] found that approximately four out of five diabetic patients are overweight or obese. The baseline estimates for health risk behaviours for participants of the LWDS suggest that there is considerable scope for behavioural lifestyle modification among Australians with diabetes.

The wide range of information collected from the LWDS participants allows for an in-depth exploration of the multidimensional nature of diabetes. The availability of longitudinal data allows the LWDS to contribute towards a deeper understanding of the dynamics of living with diabetes; and to build complex psychosocial models of the determinants of disease progression, quality of life and models of patients' assessments of the quality of their care. This will allow for the identification of key targets for intervention strategies, and will contribute to public health policy, especially as it relates to resource allocation and planning.

MD, JD, RW, BM and SP all had financial support from Queensland Health.

MD made a substantial contribution to the conception and design of the study as well as interpretation of the data and was involved in drafting the manuscript. JD contributed to the interpretation of the data and was involved in drafting the manuscript. RW supervised the statistical analyses and was involved in drafting the manuscript. BM contributed to interpretation of the data and was involved in drafting the manuscript. SP performed the statistical analysis and was involved in drafting the manuscript. CB contributed to the conception of the study and was involved in revising the intellectual content of the manuscript. All authors read and approved the final manuscript.

The pre-publication history for this paper can be accessed here:

http://www.biomedcentral.com/1471-2458/12/8/prepub

The Living with Diabetes Study is funded by Queensland Health.