Skip to main content
NCBI home page
Springer logoLink to Springer
. 2026 Feb 12;35(3):68. doi: 10.1007/s11136-026-04173-4

Health-related quality of life among adults with chronic diseases: a linked, retrospective cohort study using the EQ-5D-5L in Queensland, Australia

Takuto Endo 1,✉, Susan Linden Clemens 1
PMCID: PMC12901249  PMID: 41677969

Purpose

People with chronic diseases are known to have lower EQ-5D-5L utility scores, but data are not readily available in an Australian context. Using linked administrative hospital and population survey data, we aimed to calculate utility scores for adults with different disease profiles.

Methods

We conducted a retrospective cohort study using a cross-sectional population-level health survey (2022–2023) linked to administrative hospital data for adults 18 years and older in Queensland, Australia to assess: (1) chronic disease and comorbidity prevalence, (2) Health-related quality of life (HRQoL) differences among adults with pre-existing chronic conditions, and (3) to model differences in disutility days by chronic diseases.

Results

The mean EQ-5D-5L utility score for the cohort was 0.917, but was lower among those with chronic diseases, for example chronic obstructive pulmonary disease (COPD), coronary heart disease (CHD) and diabetes had corresponding mean values of 0.780, 0.850 and 0.832, respectively. After adjustment, on average the disutility days that could be averted by preventing chronic diseases equalled approximately a month annually for some conditions, ranging from 14.8 days for CHD to 36.5 days for COPD. Prevalence estimates using linked administrative hospital data were comparable to results from the National Health Survey, which used self-report, although comorbidity was found to be substantially higher in the current study.

Conclusion

People living with chronic diseases have substantially higher number of disutility days annually. Preventing or delaying onset of chronic conditions would likely improve HRQoL and positively impact individuals, society and the economy.

Supplementary Information

The online version contains supplementary material available at 10.1007/s11136-026-04173-4.

Keywords: Health-related quality of life, EQ-5D-5L, Chronic disease, Population health survey, Utility score, Data linkage

Plain language summary

Why is this study needed?

Health-related quality of life (HRQoL) is lower among those with chronic diseases than those without. While there are methods to evaluate HRQoL numerically, estimates for people living with chronic diseases are not available in an Australian context.

What is the key problem/issue/question this manuscript addresses?

The key questions this manuscript addresses are to: (1) assess chronic disease and comorbidity prevalence using population-level survey data linked to administrative health records, (2) quantify HRQoL differences among adults with pre-existing chronic conditions, and (3) model differences in disutility days by listed chronic diseases.

What is the main point of your study?

Using a population health survey linked to hospital admission and emergency presentation data, this study estimated the prevalence of selected chronic diseases in Queensland, Australia, and evaluated differences in HRQoL scores and the number of days not in optimal health (disutility day) among people with different disease profiles.

What are your main results and what do they mean?

Survey data linked to administrative hospital data provided chronic disease prevalence estimates comparable to a national health survey, although multimorbidity estimates were higher. Multimorbidity is commonly associated with higher health care costs and utilisation, and is a growing challenge to healthcare sector sustainability. Robust multimorbidity prevalence better informs healthcare system planning. Population-level HRQoL scores were also lower among Queensland adults with chronic diseases. A high number of disutility days could be avoided if chronic disease onset could be averted or delayed.

Supplementary Information

The online version contains supplementary material available at 10.1007/s11136-026-04173-4.

Introduction

In 2022–23, $34.7 billion was spent on diseases and injuries in Queensland, Australia [1]. Three disease groups contributed most to health system expenditure—cancer ($4.0 billion), cardiovascular diseases ($3.3 billion) and musculoskeletal disorders ($3.1 billion), with most diseases within groups considered chronic. Chronic disease attributable health burden was high at 16.4% (cancer), 11.8% (cardiovascular diseases) and 12.7% (musculoskeletal conditions) of total burden [2].

Those with chronic conditions experience increased pain, discomfort, and reduced mobility, and often have ongoing complex health needs, especially in those with additional comorbidities. This can impact labour participation and social interactions for patients and their carer, and quality of life may be substantially reduced. Increased medical cost and reduced labour participation increase pressure on patients and their families, as well as the economy overall [3]. Globally, multimorbidity is commonly associated with higher health care costs and utilisation, and is a growing challenge to healthcare sector sustainability [4–7].

Health-related quality of life (HRQoL), measured using a variety of validated instruments, quantifies how health impacts quality of life. Preference-based HRQoL instruments use scoring algorithms to generate health utility scores where 1 represents full health. Utility scores are used to produce quality-adjusted life years (QALY) and life expectancy (QALE), and are used widely across health economics research as an outcome measure. The EQ-5D [8] is recommended by the National Institute for Health and Care Excellence for economic evaluations [9] and is one of the most commonly used HRQoL instruments globally. The multidimensional EQ-5D measures mobility, self-care, usual activities, pain/discomfort and anxiety/depression.

A less common but important use of utility scores is as a summary measure of population health, due to associations with health outcomes such as disease severity, health care utilisation and cost, and mortality. They have been used to predict outcomes such as coronary heart disease (CHD), ventricular arrhythmias, heart failure and atrial fibrillation [10–12], opioid misuse [13], HIV [14], pulmonary arterial hypertension [15], and arthritis [16].

Due to these diverse uses, a recent systematic review catalogued EQ-5D utility scores for chronic diseases [17]. Despite heterogeneity across studies, people living with chronic conditions had substantially reduced HRQoL. Gaps in knowledge were noted for several diseases and countries, including Australia. These exist because administrative health care records and population surveys infrequently include the EQ-5D, despite it being a recommended patient reported outcome measure in Australia [18] and internationally [8, 9]. The Queensland preventive health survey (QPHS), however, has reported population norms using Australian value sets using both the 3- and 5-level EQ-5D [19, 20]. Population surveys, however, often lack information on health conditions, or collect such information by participant self-report which may introduce recall bias.

The aims of this study are to (1) assess chronic disease and comorbidity prevalence using population-level survey data linked to administrative health records, (2) quantify HRQoL differences among adults with pre-existing chronic conditions, and (3) model differences in disutility days by listed chronic diseases.

Methods

Data

The QPHS is a general population, cross-sectional, computer-assisted telephone interview survey of modifiable risk factors. It has been Queensland’s primary modifiable risk factor surveillance system for over two decades. The adult survey scope is community-living adults 18 years or older residing in Queensland with approximately 12,500 adults participating annually.

The EQ-5D-5L were collected on both the 2022 and 2023 surveys. The EQ-5D-5L consists of questions asking people to rank their current mobility, ability to look after themselves and to do usual activities, physical pain and discomfort, and mental health on a 5-level Likert scale [21]. Respondents also completed the accompanying visual analog scale (VAS—which was described verbally because it was a telephone survey) question asking “Now I would like you to think of a scale between 0 and 100, where 0 is the worst health you can imagine, and 100 is the best health you can imagine. What number between 0 and 100 best describes your health today?” Survey methods are also documented elsewhere [20, 22].

EQ-5D-5L utility scores were calculated using the Australian value set, which can range from − 0.301 to 1 [23]. Available modifiable risk factors were: smoking (both years), height and weight (both years), alcohol consumption (2022), nutrition (2023), physical activity (2023) and sun protection (2023). Routine sociodemographic variables were also collected. Area-based measures for the index of relative socioeconomic advantage and disadvantage (IRSAD) [24] and remoteness [25] were based on participants’ residential address. The IRSAD uses a collection of relevant measures from the 5-yearly Census to create an index score that is then expressed as quartiles or deciles ranging from most advantaged to most disadvantaged. Similarly, remoteness categories are defined nationally based on connectivity to major services.

Consent for data linkage was asked during the 2022 and 2023 telephone survey interviews (see Supplementary materials) with participants able to withdraw consent if desired through a confidential link on the survey website. Data Linkage Queensland (DLQ), a node of the Population Health Research Network [26], probabilistically linked QPHS records using an index file containing unit-record level data from various health information systems [27].

Hospital admission records were sourced from the Queensland Hospital Admitted Patient Data Collection (QHAPDC; from 1 July 2007 onwards) and emergency department presentation records from the Emergency Data Collection (EDC; from 1 July 2008 onwards). The scope of QHAPDC (public and private hospitals) and EDC (public only) are facilities in Queensland. QHAPDC and EDC record diagnoses information for each event using the International Statistical Classification of Diseases and Related Health Problems, Tenth Revision, Australian Modification (ICD-10-AM).

Whether chronic conditions were present at survey interview was assigned by assuming that if a chronic disease (see Supplementary information Table S1 for ICD-10-AM codes) was recorded at an earlier time point, the person had that condition thereafter. For example, if a participant was linked to a record with a chronic obstructive pulmonary disease (COPD) diagnosis in 2010, the participant was considered to have COPD thereafter, and thus, at the time of telephone interview. Both the primary and other diagnoses were used, and a person could be diagnosed with more than one chronic disease at a single presentation or over time. Diseases considered were those likely to be identified in tertiary care settings, specifically COPD, CHD, diabetes (excluding gestational), stroke, cancer and renal disease (the listed chronic diseases). Charlson Comorbidity Index [28] (CCI) conditions were also assigned, but without applying the age component. Listed and CCI diseases were assigned from participants’ earliest linked record to the month preceding the survey interview (exact survey date was unavailable).

Statistical analysis

Unless otherwise stated, analyses were population weighted. Records missing a given covariate value were excluded when it was used in specific analyses, but were retained otherwise. See Endo et al. [20] for missing value proportion for each indicator. When participants were excluded from analysis (for example, linkage non-consent), the original study design information was retained rather than only using the subset’s population weights, using subset() method of R’s survey package [29, 30].

Population characteristics and prevalence estimates

Population-weighted proportions estimated the prevalence of adults living with listed chronic diseases overall and by sociodemographic variables. Chi-squared (Inline graphic) tests were conducted to test demographic differences among adults living with listed chronic diseases (for example, whether there were more males than females with COPD). Weighted, univariable quasi-binomial logistic regression assessed whether prevalence estimates differed across different demographic profiles (for example, to test if COPD prevalence was higher among males than females).

Health utility score and disutility days

Population-weighted mean EQ-5D-5L utility scores were calculated for each listed chronic disease by population subgroups. A bootstrap method with 15,000 replicates was used to estimate 95% confidence intervals (CIs) for the mean utility score. To assess the convergence of confidence intervals, samples were taken with various replication counts. CIs were computed using the percentiles from the bootstrap sampled mean estimates for the utility score to account for non-normal distributions. A Wilcoxon rank-sum test was conducted between the mean value for each listed disease cohort and those without a chronic disease. The None of the listed chronic disease group are those who did not have any listed chronic diseases nor any CCI conditions. The number of disutility days (days in less than full health) were calculated by multiplying the difference between full health (utility score of 1.0) and the mean utility score by − 365, as in Mujica-Mota et al. [31].

Propensity score analysis

One component of the current study is to understand the association between pre-existing chronic diseases and subsequent HRQoL. As an observational study, bias may arise from factors such as participant sociodemographic characteristics. Propensity score (PS) analysis was used to mitigate this risk.

Briefly, PSs model the probability of receiving certain treatments, defined in the current study as the probability of being diagnosed with a chronic disease, based on modelling using covariates. PS analysis assumes that participants with similar distributions of covariates would have similar PSs [32]. Directed acyclic graphs (DAGs) are a visual tool representing hypothetical causal relationships between variables (see Supplementary Fig. S1) [33]. Cancer was omitted from these analyses as the category was considered too broad to draw a meaningful DAG, with substantial variability in cancer aetiology, treatment, survival and risk of recurrence or secondary tumours [34]. Because pre-existing chronic conditions were hypothesised to be associated with reduced HRQoL, a formative model DAG was utilized [33]. DAGs were generated using daggitty and ggdag packages [35, 36].

The QPHS is a modifiable risk factor survey with covariates limited to routine sociodemographic variables. Condition-specific covariates were unavailable which simplified DAG models. Age at survey was included as it was considered important to adjust for its strong effects on both the risk of disease onset and the HRQoL outcome [37]. While an earlier study reported males have higher utility scores than females [20], sex was not considered to directly impact utility scores, but may impact disease onset.

DAG development considered chronic diseases potentially associated with developing other chronic diseases. For example, while diabetes is a risk factor for CHD, stroke and kidney disease onset, the timing of disease onset could not be reliably ascertained. DAG paths between comorbid conditions, including diabetes, were therefore not connected to the chronic diseases of interest. The CCI was derived as a continuous variable, however, due to limited numbers of participants with high CCI scores, censored CCI versions were considered. The value at which it should be censored was determined through the PS modelling steps and the associated diagnostics. Because the CCI includes some of the listed chronic diseases, those were removed list-wise by participant to avoid perfect prediction. For example, analysis of adults living with diabetes removed diabetes and diabetes complications from the CCI derivation.

The inclusion of risk factor information, such as obesity, smoking and socioeconomic status in PS analysis and DAGs was carefully considered. Because a long look-back period was used to identify chronic diseases, some characteristics may have changed between diagnosis and the survey, which may bias findings. For this reason, only smoking was treated as a pre-exposure variable, as smoking initiation typically occurs during teen/early adulthood and is unlikely after disease onset. Smoking was categorised as current/ex-smoker and never smoker due to the lack of a smoking cessation date among ex-smokers. Smoking is known to be associated with COPD, CHD, diabetes, stroke and renal diseases as well as a number of CCI conditions, and therefore pathways were included in all DAGs. Smoking has been shown to be associated with lower utility scores [38], and thus, direct paths to the outcome as well as chronic conditions were specified.

PS models were developed for each listed chronic disease, with age (continuous), smoking status and CCI as covariates. Although there is no path from CCI to the listed chronic disease, it was included for efficiency in modelling outcomes of interest [37]. In order to estimate the effect of having the chronic disease on health utility scores, average treatment effects on treated (ATT, with ‘treated’ indicating pre-existing chronic diseases) were estimated by calculating the PS for having the disease at survey administration. PSs were calculated using logistic regression with the WeightIt package [39].

ATT was calculated by fitting a population-weighted linear regression to the ATT and population-weighted data. When covariates were included in linear models, obesity was derived from self-reported height and weight, collected during the survey, using standard BMI cutoffs and age was included as a continuous variable. E-values were calculated to understand the effect any unmeasured confounders would need on both the treatment (chronic disease) and the outcome (utility score) to nullify the estimates [40, 41]. All analyses were done using R version 4.5.0 [42]. ChatGPT contributed to the literature search, statistical programming, to determine the most appropriate R package to produce plots, and early phase of model diagnostics but not in the selection of the final model. It was not used to write or draft text of the manuscript, nor interpret or translate findings.

Results

Linkage results

Of the 25,170 survey participants, 21,297 (84.6%) consented and were linked to administrative health records. The demographic differences in the adults who did and did not consent to linkage are shown in Supplementary Table S2. Linkage identified a small number of duplicated participants between the 2022 and 2023 surveys (0.4%), with the 2022 survey responses retained. No duplicates were found within a survey year, resulting in 21,202 unique records (10,377 for 2022, 10,825 for 2023, see Supplementary Fig. S2).

Overall, 18,920 participants were linked to at least one admitted episode or emergency presentations on or before the survey month. This translates to 27.6% (2022) and 27.5% (2023) having at least one admitted patient episode of care, and 25.6% (2022) and 22.7% (2023) for emergency presentations, in the 12-months before the surveys. Mean and median duration from the first linked admitted episode or emergency presentations to the survey was 10.0 years and 11.2 years, respectively (Supplementary Fig. S3).

Proportion of adults living with chronic diseases

As a general population survey, the proportion of participants diagnosed with a listed chronic disease could be considered the population prevalence. Table 1 summarises the prevalence of adults with a past diagnosis of the listed chronic diseases and Supplementary Table S3 presents sociodemographic differences. For example, on average, 2.9% of adults 18 years and older lived with COPD and 5.9% with diabetes, while 73.9% had none of the listed diseases. Chronic diseases prevalence increased with age, and tended to be higher for retired or unemployed adults, and those living in disadvantaged areas.

Table 1.

Prevalence (%; 95% CI) of listed chronic diseases using linked QPHS and hospital data, 2022–2023

Cohort Group COPD CHD Diabetes Stroke Cancer Renal disease CCI ≥ 1 CCI ≥ 2 None of the listed chronic disease
Overall All 2.9 (2.7–3.2) 9.2 (8.7–9.6) 5.9 (5.6–6.3) 1.2 (1.1–1.4) 5.7 (5.3–6.0) 1.6 (1.4–1.8) 23.2 (22.5–23.8) 12.0 (11.6–12.5) 73.9 (73.2–74.5)
Sex Males (Reference) 3.0 (2.7–3.3) 10.4 (9.8–11.0) 6.6 (6.1–7.1) 1.4 (1.2–1.7) 5.9 (5.4–6.4) 1.7 (1.5–2.0) 22.3 (21.4–23.2) 12.4 (11.8–13.1) 74.4 (73.5–75.3)
Females 2.9 (2.6–3.2) p = 0.75 7.9 (7.4–8.5) p < 0.01 5.2 (4.8–5.7) p < 0.01 1.1 (0.9–1.3) p = 0.04 5.4 (5.0–6.0) p = 0.22 1.5 (1.2–1.7) p = 0.16 24.1 (23.1–25.0) p < 0.01 11.6 (10.9–12.3) p = 0.08 73.3 (72.3–74.3) p = 0.12
Age 18–49 years (Reference) 0.3 (0.2–0.5) 1.9 (1.6–2.2) 1.7 (1.4–2.0) 0.2 (0.2–0.4) 1.4 (1.1–1.7) 0.3 (0.2–0.4) 11.5 (10.7–12.3) 3.0 (2.7–3.5) 87.5 (86.7–88.3)
50–64 years 2.8 (2.4–3.3) p < 0.01 11.8 (10.8–12.8) p < 0.01 7.0 (6.2–7.8) p < 0.01 1.4 (1.1–1.9) p < 0.01 6.9 (6.2–7.8) p < 0.01 1.1 (0.8–1.5) p < 0.01 26.3 (24.9–27.6) p < 0.01 14.3 (13.3–15.5) p < 0.01 68.9 (67.5–70.4) p < 0.01
65–79 years 8.7 (7.9–9.6) p < 0.01 22.7 (21.4–24.1) p < 0.01 15.1 (14.0–16.2) p < 0.01 3.1 (2.6–3.7) p < 0.01 14.3 (13.2–15.4) p < 0.01 4.2 (3.6–4.9) p < 0.01 47.1 (45.5–48.7) p < 0.01 29.9 (28.5–31.4) p < 0.01 47.1 (45.5–48.7) p < 0.01
80 + years 13.8 (11.6–16.3) p < 0.01 35.0 (31.8–38.3) p < 0.01 18.1 (15.7–20.9) p < 0.01 5.6 (4.3–7.3) p < 0.01 19.3 (16.7–22.1) p < 0.01 11.8 (9.7–14.2) p < 0.01 62.0 (58.6–65.3) p < 0.01 43.9 (40.5–47.3) p < 0.01 31.6 (28.5–34.9) p < 0.01
Employment status Employed/Student (Reference) 0.7 (0.6–0.9) 4.7 (4.3–5.1) 2.8 (2.5–3.1) 0.4 (0.3–0.6) 3.0 (2.7–3.3) 0.4 (0.3–0.5) 14.5 (13.8–15.2) 5.7 (5.2–6.1) 83.3 (82.5–84.0)
Retired 9.5 (8.7–10.5) p < 0.01 24.0 (22.7–25.3) p < 0.01 15.5 (14.4–16.7) p < 0.01 3.6 (3.1–4.2) p < 0.01 14.8 (13.7–15.9) p < 0.01 5.7 (5.0–6.5) p < 0.01 49.3 (47.8–50.9) p < 0.01 31.8 (30.3–33.2) p < 0.01 44.8 (43.2–46.4) p < 0.01
Home duties/career 3.6 (2.6–5.1) p < 0.01 9.6 (7.9–11.8) p < 0.01 6.3 (4.9–8.2) p < 0.01 1.5 (0.8–2.6) p < 0.01^ 6.3 (4.8–8.2) p < 0.01 1.3 (0.7–2.3) p < 0.01^ 30.0 (26.7–33.5) p < 0.01 14.8 (12.5–17.5) p < 0.01 67.6 (64.0–71.0) p < 0.01
Unemployed/unable to work 6.7 (5.5–8.3) p < 0.01 13.7 (11.8–15.9) p < 0.01 11.5 (9.7–13.6) p < 0.01 3.0 (2.1–4.3) p < 0.01 7.3 (5.8–9.1) p < 0.01 3.3 (2.3–4.6) p < 0.01 35.7 (32.7–38.8) p < 0.01 21.3 (18.9–24.0) p < 0.01 61.6 (58.4–64.6) p < 0.01
Socioeconomic status Most disadvantaged 5.3 (4.7–6.0) p < 0.01 13.4 (12.4–14.5) p < 0.01 8.9 (8.1–9.9) p < 0.01 1.7 (1.3–2.1) p < 0.01 6.0 (5.4–6.8) p = 0.53 2.3 (1.8–2.8) p < 0.01 29.0 (27.5–30.5) p < 0.01 15.4 (14.3–16.6) p < 0.01 67.1 (65.6–68.7) p < 0.01
Q2 3.9 (3.4–4.5) p < 0.01 10.8 (9.9–11.8) p < 0.01 7.3 (6.6–8.1) p < 0.01 1.3 (1.0–1.7) p = 0.05 6.2 (5.5–7.0) p = 0.36 1.7 (1.3–2.1) p = 0.04 26.0 (24.7–27.4) p < 0.01 14.2 (13.2–15.3) p < 0.01 70.7 (69.2–72.1) p < 0.01
Q3 2.3 (1.9–2.8) p = 0.03 9.0 (8.1–10.0) p < 0.01 5.8 (5.1–6.7) p < 0.01 1.1 (0.8–1.5) p = 0.29 5.4 (4.7–6.2) p = 0.57 1.4 (1.1–1.8) p = 0.22 22.6 (21.2–24.1) p < 0.01 11.6 (10.6–12.7) p < 0.01 74.5 (73.0–76.0) p < 0.01
Q4 2.0 (1.6–2.5) p = 0.18 7.4 (6.5–8.3) p = 0.03 4.7 (4.1–5.5) p < 0.01 1.3 (1.0–1.8) p = 0.07 5.1 (4.4–5.9) p = 0.29 1.7 (1.3–2.2) p = 0.03 19.9 (18.6–21.4) p = 0.75 10.1 (9.1–11.1) p = 0.49 77.6 (76.1–79.0) p = 0.74
Most advantaged (reference) 1.6 (1.2–2.1) 6.0 (5.2–6.9) 3.4 (2.8–4.1) 0.8 (0.6–1.2) 5.7 (4.9–6.6) 1.1 (0.8–1.5) 19.6 (18.1–21.2) 9.6 (8.5–10.7) 78.0 (76.3–79.5)
Remoteness Major cities (reference) 2.4 (2.1–2.7) 8.1 (7.5–8.6) 5.3 (4.9–5.8) 1.2 (1.0–1.4) 5.3 (4.8–5.7) 1.6 (1.3–1.8) 21.5 (20.6–22.4) 11.0 (10.4–11.7) 75.9 (74.9–76.8)
Inner regional 4.2 (3.7–4.7) p < 0.01 11.6 (10.8–12.5) p < 0.01 7.1 (6.4–7.8) p < 0.01 1.3 (1.0–1.6) p = 0.62 6.8 (6.2–7.5) p < 0.01 1.7 (1.4–2.1) p = 0.41 27.2 (25.9–28.5) p < 0.01 14.4 (13.5–15.4) p < 0.01 69.1 (67.7–70.4) p < 0.01
Outer regional 3.4 (2.9–4.0) p < 0.01 10.2 (9.3–11.2) p < 0.01 6.5 (5.7–7.3) p < 0.01 1.4 (1.1–1.8) p = 0.41 5.6 (5.0–6.4) p = 0.38 1.6 (1.2–2.0) p = 0.98 24.5 (23.0–25.9) p < 0.01 12.5 (11.5–13.6) p = 0.02 72.2 (70.7–73.7) p < 0.01
Remote 4.4 (3.4–5.7) p < 0.01 11.0 (9.3–12.9) p < 0.01 7.5 (6.1–9.3) p < 0.01 1.4 (0.7–2.7) p = 0.72^ 6.8 (5.2–8.7) p = 0.07 1.5 (0.9–2.4) p = 0.84 25.9 (23.4–28.7) p < 0.01 14.2 (12.2–16.6) p < 0.01 70.8 (68.0–73.5) p < 0.01
Very remote 3.5 (2.5–5.0) p = 0.04 9.9 (7.9–12.2) p = 0.08 8.4 (6.5–10.7) p < 0.01 1.5 (0.9–2.5) p = 0.47^ 4.9 (3.6–6.6) p = 0.66 1.5 (0.8–2.6) p = 0.88^ 23.9 (20.8–27.3) p = 0.15 13.3 (10.9–16.0) p = 0.07 73.3 (69.8–76.5) p = 0.13
Open in a new tab

Excludes records with invalid or unknown response

COPD, Chronic obstructive pulmonary disease; CHD, Coronary heart disease; CCI, Charlson comorbidity index

Age was not used for CCI calculation

^RSE > 25%

Quasi-binomial logistic regression was used for statistical tests

Comorbidity

The proportion of adults living with more than one listed chronic diseases was also estimated. Of adults 18 years and over with COPD, 24.5% also had diabetes. Similarly, 12.1% of adults living with diabetes also had COPD. Among those 65 years and older, this increased to 26.1% and 16.2%, respectively. See Supplementary Table S4 for additional diseases.

Average EQ-5D-5L utility and VAS score

There were 21,151 valid EQ-5D-5L utility scores out of 21,202 participant records and 35.2% reported having a utility score of 1.0, with 0.3% having utility value below 0. CIs from bootstrap samplings were relatively stable after 1000 iterations (Supplementary Fig. S4). A greater number of disutility days were observed for all diseases including the CCI and, for the majority of diseases, younger cohorts with chronic disease had more disutility days than the oldest cohort with none of the listed diseases.

In the current study, the mean utility score was 0.917, translating to an average of 30.2 disutility days annually among the general adult population, with reduced scores in those with chronic conditions. For example, mean utility scores in those with COPD, CHD and diabetes were 0.780, 0.850 and 0.832, respectively (Table 2). Disutility days reduced to 23.5 days for those without any of the listed chronic diseases, but increased to 80.4 days, 61.4 days, 54.8 for adults with COPD, diabetes and CHD, respectively. Similar patterns were found for mean VAS (Supplementary Table S5).

Table 2.

Mean health utility (%; 95% CI) and disutility days by chronic diseases status and sociodemographic characteristics, 18 years and over, Queensland, Australia

Cohort Group Measure COPD CHD Diabetes Stroke Cancer Renal disease CCI ≥ 1 CCI ≥ 2 None of the listed chronic disease (Reference) Overall
Overall All Mean utility score 0.780 (0.759–0.800) p < 0.01 0.850 (0.840–0.860) p < 0.01 0.832 (0.818–0.845) p < 0.01 0.817 (0.784–0.848) p < 0.01 0.883 (0.873–0.893) p < 0.01 0.792 (0.762–0.820) p < 0.01 0.860 (0.854–0.867) p < 0.01 0.842 (0.833–0.850) p < 0.01 0.936 (0.933–0.938) 0.917 (0.915–0.920)
Disutility days 80.4 (73.1, 88.0) 54.8 (51.2, 58.5) 61.4 (56.7, 66.3) 66.7 (55.3, 78.8) 42.7 (39.1, 46.4) 75.9 (65.8, 86.9) 51.0 (48.7, 53.3) 57.8 (54.6, 61.0) 23.5 (22.7, 24.3) 30.2 (29.4, 31.0)
Sex Males Mean utility score 0.801 (0.774–0.827) p < 0.01 0.863 (0.851–0.874) p < 0.01 0.848 (0.831–0.865) p < 0.01 0.845 (0.810–0.878) p < 0.01 0.888 (0.875–0.900) p < 0.01 0.802 (0.760–0.839) p < 0.01 0.873 (0.865–0.881) p < 0.01 0.856 (0.844–0.867) p < 0.01 0.941 (0.938–0.944) 0.925 (0.922–0.928)
Disutility days 72.6 (63.2, 82.3) 50.2 (46.0, 54.5) 55.4 (49.4, 61.7) 56.4 (44.4, 69.4) 41.0 (36.4, 45.7) 72.5 (58.7, 87.7) 46.4 (43.5, 49.4) 52.7 (48.6, 57.0) 21.5 (20.5, 22.6) 27.3 (26.3, 28.4)
Females Mean utility score 0.758 (0.726–0.789) p < 0.01 0.833 (0.816–0.850) p < 0.01 0.811 (0.790–0.832) p < 0.01 0.781 (0.722–0.835) p < 0.01 0.878 (0.862–0.893) p < 0.01 0.781 (0.738–0.821) p < 0.01 0.849 (0.839–0.858) p < 0.01 0.827 (0.813–0.840) p < 0.01 0.930 (0.927–0.934) 0.910 (0.906–0.913)
Disutility days 88.5 (77.2, 99.8) 60.8 (54.8, 67.3) 68.9 (61.5, 76.7) 80.1 (60.4, 101.3) 44.5 (38.9, 50.4) 79.9 (65.2, 95.7) 55.2 (51.8, 58.6) 63.3 (58.4, 68.2) 25.5 (24.2, 26.7) 33.0 (31.8, 34.3)
Age 18–49 years Mean utility score 0.805 (0.716–0.884) p < 0.01 0.854 (0.819–0.885) p < 0.01 0.849 (0.808–0.885) p < 0.01 0.938 (0.894–0.970) p = 0.69 0.906 (0.878–0.931) p < 0.01 0.868 (0.689–0.966) p = 0.35 0.880 (0.867–0.892) p < 0.01 0.863 (0.836–0.887) p < 0.01 0.941 (0.938–0.943) 0.933 (0.930–0.936)
Disutility days 71.4 (42.3, 103.8) 53.2 (41.9, 66.0) 55.1 (41.8, 70.2) 22.5 (10.8, 38.8) 34.2 (25.2, 44.5) 48.1 (12.3, 113.5) 43.8 (39.3, 48.6) 50.0 (41.2, 59.9) 21.7 (20.7, 22.7) 24.4 (23.4, 25.5)
50–64 years Mean utility score 0.725 (0.675–0.771) p < 0.01 0.853 (0.833–0.872) p < 0.01 0.824 (0.796–0.849) p < 0.01 0.734 (0.658–0.803) p < 0.01 0.887 (0.867–0.905) p < 0.01 0.791 (0.728–0.847) p < 0.01 0.846 (0.832–0.859) p < 0.01 0.830 (0.811–0.847) p < 0.01 0.928 (0.923–0.933) 0.905 (0.900–0.910)
Disutility days 100.4 (83.5, 118.6) 53.7 (46.8, 60.9) 64.4 (55.3, 74.3) 97.0 (71.9, 124.8) 41.2 (34.6, 48.6) 76.3 (55.7, 99.3) 56.2 (51.5, 61.2) 62.2 (55.7, 69.1) 26.2 (24.5, 28.0) 34.5 (32.7, 36.4)
65–79 years Mean utility score 0.795 (0.766–0.821) p < 0.01 0.854 (0.840–0.868) p < 0.01 0.831 (0.813–0.849) p < 0.01 0.816 (0.766–0.862) p < 0.01 0.883 (0.869–0.896) p < 0.01 0.767 (0.722–0.808) p < 0.01 0.861 (0.852–0.870) p < 0.01 0.843 (0.831–0.855) p < 0.01 0.926 (0.920–0.932) 0.895 (0.889–0.900)
Disutility days 75.0 (65.4, 85.3) 53.2 (48.2, 58.4) 61.5 (55.2, 68.3) 67.0 (50.5, 85.3) 42.6 (37.9, 47.7) 85.1 (70.2, 101.6) 50.8 (47.6, 54.1) 57.2 (52.8, 61.7) 26.9 (24.6, 29.3) 38.4 (36.5, 40.4)
80 + years Mean utility score 0.795 (0.753–0.837) p < 0.01 0.829 (0.803–0.854) p < 0.01 0.830 (0.794–0.864) p < 0.01 0.871 (0.825–0.910) p = 0.06 0.852 (0.817–0.885) p = 0.06 0.808 (0.763–0.850) p < 0.01 0.845 (0.827–0.863) p < 0.01 0.839 (0.817–0.860) p < 0.01 0.906 (0.887–0.922) 0.868 (0.855–0.881)
Disutility days 74.7 (59.5, 90.3) 62.3 (53.3, 72.0) 61.9 (49.6, 75.3) 47.0 (32.9, 63.8) 53.8 (42.1, 66.8) 70.1 (54.9, 86.5) 56.5 (50.0, 63.3) 58.7 (51.2, 66.7) 34.5 (28.4, 41.3) 48.0 (43.5, 52.9)
Employment status Employed/student Mean utility score 0.883 (0.850–0.913) p < 0.01 0.910 (0.899–0.920) p < 0.01 0.897 (0.878–0.914) p < 0.01 0.891 (0.855–0.921) p < 0.01 0.930 (0.920–0.939) p < 0.01 0.838 (0.734–0.914) p < 0.01 0.911 (0.904–0.918) p < 0.01 0.904 (0.892–0.915) p < 0.01 0.947 (0.944–0.949) 0.941 (0.939–0.943)
Disutility days 42.6 (31.9, 54.9) 32.9 (29.2, 36.8) 37.5 (31.5, 44.4) 40.0 (28.7, 52.8) 25.6 (22.2, 29.3) 59.2 (31.4, 97.2) 32.4 (30.0, 35.0) 35.0 (31.1, 39.5) 19.5 (18.7, 20.3) 21.6 (20.8, 22.4)
Retired Mean utility score 0.800 (0.776–0.824) p < 0.01 0.846 (0.833–0.860) p < 0.01 0.836 (0.818–0.852) p < 0.01 0.845 (0.809–0.876) p < 0.01 0.879 (0.865–0.892) p < 0.01 0.793 (0.760–0.824) p < 0.01 0.857 (0.848–0.865) p < 0.01 0.843 (0.831–0.854) p < 0.01 0.922 (0.916–0.928) 0.890 (0.884–0.895)
Disutility days 72.9 (64.3, 81.9) 56.1 (51.1, 61.1) 60.0 (54.1, 66.3) 56.6 (45.2, 69.7) 44.1 (39.3, 49.3) 75.5 (64.3, 87.7) 52.2 (49.1, 55.4) 57.5 (53.3, 61.7) 28.4 (26.2, 30.8) 40.3 (38.4, 42.2)
Home duties/carer Mean utility score 0.772 (0.695–0.844) p < 0.01 0.841 (0.800–0.880) p < 0.01 0.802 (0.740–0.860) p < 0.01 0.765 (0.561–0.943) p = 0.57 0.874 (0.826–0.916) p = 0.09 0.769 (0.659–0.870) p < 0.01 0.834 (0.805–0.861) p < 0.01 0.824 (0.783–0.862) p < 0.01 0.917 (0.904–0.929) 0.892 (0.879–0.904)
Disutility days 83.2 (56.9, 111.3) 57.9 (43.9, 73.1) 72.4 (51.2, 94.9) 85.9 (20.7, 160.3) 45.9 (30.7, 63.6) 84.3 (47.6, 124.6) 60.7 (50.7, 71.4) 64.3 (50.5, 79.1) 30.3 (26.0, 34.9) 39.5 (35.2, 44.2)
Unemployed/unable to work Mean utility score 0.551 (0.481–0.617) p < 0.01 0.632 (0.583–0.680) p < 0.01 0.643 (0.589–0.695) p < 0.01 0.605 (0.486–0.720) p < 0.01 0.691 (0.628–0.749) p < 0.01 0.735 (0.637–0.816) p = 0.13 0.650 (0.618–0.681) p < 0.01 0.652 (0.613–0.690) p < 0.01 0.810 (0.790–0.829) 0.749 (0.731–0.766)
Disutility days 163.9 (139.7, 189.3) 134.2 (116.6, 152.2) 130.3 (111.3, 150.2) 144.3 (102.0, 187.7) 112.9 (91.7, 135.7) 96.9 (67.3, 132.6) 127.8 (116.6, 139.3) 127.1 (113.2, 141.4) 69.4 (62.4, 76.6) 91.6 (85.4, 98.1)
Socioeconomic status Most disadvantaged Mean utility score 0.774 (0.741–0.805) p < 0.01 0.825 (0.805–0.844) p < 0.01 0.802 (0.777–0.826) p < 0.01 0.834 (0.787–0.875) p < 0.01 0.869 (0.847–0.889) p < 0.01 0.773 (0.718–0.824) p < 0.01 0.833 (0.820–0.846) p < 0.01 0.813 (0.795–0.830) p < 0.01 0.917 (0.911–0.923) 0.891 (0.885–0.896)
Disutility days 82.5 (71.0, 94.6) 63.9 (57.0, 71.1) 72.3 (63.5, 81.6) 60.4 (45.5, 77.6) 47.7 (40.4, 55.7) 82.9 (64.3, 102.8) 60.9 (56.3, 65.6) 68.2 (62.1, 74.6) 30.3 (28.2, 32.5) 39.9 (37.8, 42.0)
Q2 Mean utility score 0.769 (0.727–0.808) p < 0.01 0.859 (0.841–0.877) p < 0.01 0.844 (0.819–0.867) p < 0.01 0.837 (0.782–0.884) p < 0.01 0.878 (0.855–0.898) p < 0.01 0.806 (0.747–0.856) p < 0.01 0.865 (0.853–0.876) p < 0.01 0.845 (0.828–0.861) p < 0.01 0.929 (0.924–0.934) 0.912 (0.907–0.916)
Disutility days 84.3 (70.1, 99.7) 51.3 (44.9, 58.2) 56.9 (48.4, 65.9) 59.4 (42.4, 79.6) 44.5 (37.1, 53.0) 70.9 (52.4, 92.3) 49.2 (45.2, 53.7) 56.6 (50.8, 62.7) 25.9 (24.0, 27.7) 32.2 (30.5, 34.0)
Q3 Mean utility score 0.805 (0.756–0.849) p < 0.01 0.853 (0.829–0.876) p < 0.01 0.834 (0.800–0.865) p < 0.01 0.792 (0.705–0.870) p < 0.01 0.866 (0.840–0.890) p < 0.01 0.746 (0.649–0.831) p < 0.01 0.862 (0.848–0.876) p < 0.01 0.843 (0.822–0.863) p < 0.01 0.938 (0.933–0.942) 0.920 (0.915–0.925)
Disutility days 71.3 (55.2, 89.1) 53.5 (45.2, 62.4) 60.4 (49.2, 72.9) 75.8 (47.4, 107.7) 48.9 (40.1, 58.5) 92.8 (61.5, 128.2) 50.3 (45.2, 55.5) 57.2 (49.9, 65.0) 22.8 (21.1, 24.6) 29.3 (27.5, 31.2)
Q4 Mean utility score 0.744 (0.672–0.810) p < 0.01 0.837 (0.807–0.866) p < 0.01 0.824 (0.788–0.857) p < 0.01 0.767 (0.670–0.854) p < 0.01 0.897 (0.874–0.918) p < 0.01 0.799 (0.741–0.851) p < 0.01 0.849 (0.832–0.866) p < 0.01 0.824 (0.798–0.849) p < 0.01 0.940 (0.936–0.945) 0.922 (0.917–0.927)
Disutility days 93.6 (69.3, 119.5) 59.3 (49.0, 70.3) 64.3 (52.0, 77.5) 85.0 (53.1, 120.6) 37.6 (30.1, 46.2) 73.5 (54.4, 94.5) 54.9 (48.8, 61.2) 64.3 (55.0, 73.9) 21.8 (20.2, 23.5) 28.5 (26.7, 30.5)
Most advantaged Mean utility score 0.830 (0.777–0.879) p < 0.01 0.890 (0.868–0.911) p < 0.01 0.878 (0.846–0.908) p < 0.01 0.872 (0.818–0.922) p < 0.01 0.904 (0.882–0.924) p < 0.01 0.855 (0.793–0.908) p < 0.01 0.897 (0.883–0.909) p < 0.01 0.892 (0.874–0.908) p < 0.01 0.948 (0.943–0.952) 0.937 (0.933–0.942)
Disutility days 62.0 (44.1, 81.5) 40.1 (32.6, 48.3) 44.4 (33.5, 56.3) 46.6 (28.6, 66.3) 35.1 (27.8, 43.1) 53.0 (33.4, 75.7) 37.7 (33.1, 42.6) 39.6 (33.6, 45.9) 19.1 (17.6, 20.8) 22.8 (21.3, 24.5)
Remoteness Major cities Mean utility score 0.776 (0.742–0.808) p < 0.01 0.852 (0.837–0.867) p < 0.01 0.832 (0.811–0.851) p < 0.01 0.802 (0.755–0.847) p < 0.01 0.881 (0.865–0.895) p < 0.01 0.777 (0.732–0.817) p < 0.01 0.863 (0.854–0.872) p < 0.01 0.843 (0.829–0.855) p < 0.01 0.938 (0.935–0.941) 0.921 (0.918–0.925)
Disutility days 81.8 (70.1, 94.1) 54.0 (48.6, 59.7) 61.5 (54.3, 68.9) 72.1 (55.8, 89.3) 43.5 (38.2, 49.2) 81.6 (66.7, 97.7) 50.1 (46.8, 53.4) 57.4 (52.8, 62.2) 22.6 (21.5, 23.7) 28.7 (27.6, 29.8)
Inner regional Mean utility score 0.770 (0.737–0.803) p < 0.01 0.840 (0.823–0.857) p < 0.01 0.811 (0.786–0.836) p < 0.01 0.814 (0.750–0.867) p < 0.01 0.881 (0.865–0.897) p < 0.01 0.811 (0.767–0.852) p < 0.01 0.846 (0.835–0.857) p < 0.01 0.831 (0.816–0.845) p < 0.01 0.928 (0.923–0.932) 0.904 (0.899–0.909)
Disutility days 83.9 (72.0, 95.9) 58.3 (52.3, 64.5) 68.9 (60.0, 78.2) 67.8 (48.7, 91.3) 43.3 (37.6, 49.4) 68.9 (54.0, 85.2) 56.1 (52.1, 60.3) 61.8 (56.6, 67.3) 26.4 (24.7, 28.1) 35.0 (33.3, 36.7)
Outer regional Mean utility score 0.807 (0.765–0.845) p < 0.01 0.857 (0.834–0.877) p < 0.01 0.862 (0.838–0.883) p < 0.01 0.871 (0.822–0.911) p < 0.01 0.887 (0.866–0.906) p < 0.01 0.842 (0.792–0.884) p < 0.01 0.868 (0.855–0.880) p < 0.01 0.849 (0.831–0.866) p < 0.01 0.933 (0.928–0.937) 0.915 (0.910–0.920)
Disutility days 70.6 (56.6, 85.9) 52.1 (44.8, 60.6) 50.4 (42.7, 59.0) 47.2 (32.5, 64.9) 41.3 (34.2, 49.0) 57.8 (42.3, 75.8) 48.2 (43.8, 52.8) 55.1 (48.7, 61.8) 24.5 (23.0, 26.1) 31.0 (29.3, 32.7)
Remote Mean utility score 0.788 (0.712–0.853) p < 0.01 0.851 (0.809–0.888) p < 0.01 0.841 (0.796–0.882) p < 0.01 0.865 (0.726–0.941) p = 0.55 0.914 (0.885–0.940) p = 0.05 0.745 (0.594–0.865) p = 0.01 0.881 (0.859–0.900) p < 0.01 0.854 (0.818–0.886) p < 0.01 0.942 (0.934–0.948) 0.925 (0.917–0.932)
Disutility days 77.5 (53.5, 105.2) 54.5 (41.0, 69.7) 57.9 (43.2, 74.6) 49.3 (21.4, 100.0) 31.3 (22.0, 42.1) 93.1 (49.4, 148.2) 43.6 (36.6, 51.5) 53.1 (41.5, 66.3) 21.3 (18.8, 24.0) 27.3 (24.7, 30.1)
Very remote Mean utility score 0.797 (0.687–0.892) p < 0.01 0.864 (0.818–0.904) p < 0.01 0.864 (0.808–0.909) p < 0.01 0.877 (0.803–0.937) p = 0.01 0.913 (0.873–0.944) p = 0.04 0.807 (0.658–0.906) p = 0.07 0.888 (0.862–0.911) p < 0.01 0.888 (0.852–0.917) p < 0.01 0.948 (0.938–0.956) 0.931 (0.921–0.940)
Disutility days 74.2 (39.5, 114.4) 49.7 (35.2, 66.2) 49.8 (33.2, 70.0) 45.0 (23.2, 71.7) 31.6 (20.6, 46.3) 70.4 (34.2, 125.0) 40.8 (32.4, 50.3) 41.0 (30.4, 53.9) 19.1 (16.1, 22.5) 25.3 (21.9, 28.9)
Open in a new tab

Excludes records with invalid or unknown response

COPD, Chronic obstructive pulmonary disease; CHD, Coronary heart disease; CCI: Charlson comorbidity index

Age was not used for CCI calculation

Reference group for Wilcoxon test was the ‘None of the listed chronic disease’ group

Across different sociodemographic groups, those living with chronic diseases reported lower mean utility scores, and correspondingly more disutility days, than those without any listed chronic diseases. For example, those 18 to 49 years with COPD had a much lower mean utility score (0.805) than peers with none of the listed chronic diseases (0.941; p < 0.01).

Generally, adults who were older had lower mean utility scores than younger adults. However, adults who were 80 years and older in the None of the listed chronic diseases group had a mean value of 0.906, which was higher than most of the mean utility score reported for younger adults living with one of the listed chronic diseases (Table 2).

Average treatment effect on treated and associated disutility days

From model diagnostics using standardised mean difference (SMD), CCI values bounded at 1 were the most desirable balance, and hence were converted into a binary variable indicating whether a person had any of the CCI comorbidities or not. Effective sample sizes for each model are in Supplementary Table S6, with balance diagnostics plots in Supplementary Figs. S5 and S6. SMD for all the covariates included in the models were below the 0.1 threshold [43], although SMD for PSs exceeded 0.1 in some instances (Supplementary Fig. S5).

Substantial disutility day reductions were observed if adults did not have chronic diseases. Based on the calculated ATTs and after adjusting by age, smoking status and CCI, the greatest reduction in mean utility score was for adults living with COPD, which would be approximately 0.100 higher (saving 36.5 disutility days) without COPD. Results for other listed conditions ranged from 14.8 to 28.4 disutility days (Table 3). Sensitivity analysis was also conducted, by including covariates that only had paths directly to the outcome, but not to the disease, in the regression model estimating ATT coefficients, together with the variables included in the PS (see Supplementary Table S7). Regression model results were still significant with marginal differences in ATTs (Supplementary Table S7).

Table 3.

Average treatment effect on treated (ATT) and disutility days per year by listed chronic diseases

Chronic disease ATT (95% CI) Disutility days (95% CI) E-value (upper limit)
COPD  − 0.100 (− 0.122, − 0.078) 36.5 days (28.6, 44.5) 2.35 (2.08)
CHD  − 0.041 (− 0.052, − 0.029) 14.8 days (10.6, 19.0) 1.72 (1.56)
Diabetes  − 0.059 (− 0.073, − 0.045) 21.4 days (16.3, 26.6) 1.93 (1.75)
Stroke  − 0.054 (− 0.086, − 0.022) 19.7 days (7.9, 31.4) 1.82 (1.42)
Renal diseases  − 0.078 (− 0.108, − 0.048) 28.4 days (17.6, 39.3) 2.08 (1.72)
Open in a new tab

Discussion

One study aim was to use linked administrative health records to estimate chronic disease and comorbidity prevalence in the Queensland adult general population. Observed prevalences were 2.9% (COPD), 9.2% (CHD), 5.9% (diabetes), 1.2% (stroke), 5.7% (cancer) and 1.6% (renal disease). Except for cancer, results compared relatively well with estimates based on self-report from the 2022 Australian National Health Survey (NHS; Supplementary Table S4) [44].

Conversely, comorbidity between listed diseases was considerably higher, approximately double in some case, than NHS estimates. For example, the proportion of adults 65 years and older living with COPD who also had diabetes was 19.8% while in the current study it was 24.5%. While methodological differences make comparisons challenging, variability in comorbidity estimated from self-report and administrative health records is not uncommon. Differences have been reported in chronic disease comorbidity prevalence [45] and incidence estimates [46], and across panel survey waves [47]. Agreement between sources tended to be higher for more severe conditions, however, there was considerable variability in the difference between estimates collected by self-reported compared to other methods. While some studies showed higher sensitivity for conditions treated outside of tertiary care by self-report [48], specificity tended to be more comparable [46]. The potential reduced sensitivity of linked administrative records was mitigated in the current study by using a long look-back period and focusing on chronic conditions important for delivery of tertiary care.

The other aims were to report health utility scores/disutility days for a subset of chronic conditions. To our knowledge this is the first study to evaluate EQ-5D-5L utility scores among adults with chronic diseases in an Australian context, and fills some gaps in an earlier systematic review [17]. This study demonstrated that HRQoL was lower among adults living with chronic diseases. Differences in utility scores ranged from 0.883 (42.7 disutility days) for cancer to 0.780 (80.4 disutility days) for COPD compared to 0.936 (23.5 disutility days) for adults without any of the listed chronic diseases. Adults in the older cohort without any of the listed chronic diseases also had higher average utility score than younger adults living with listed chronic diseases in several cases such as COPD. While survivorship bias and the exclusion of institutionalised adults likely contributed to relatively high utility scores in older cohorts, results demonstrate the considerable burden of chronic diseases at any age.

The current study reported a higher mean adult general population EQ-5D-5L utility score than in the national sample reported by Redwood et al. (0.917 vs 0.86) [49], likely due to sampling framework differences. For example, the proportions of unemployed and retirees were much lower in our study cohort (6.0% vs 11.2% and 19.1% vs 25.7%, respectively; Supplementary Table S5) [49], as was the proportion of current smokers (approximately 15% vs 23.2%, respectively) [22]. Our overall utility score was comparable to the mean utility score reported for South Australian population based on United Kingdom value set for those 15 years and older (0.91) [50]. When compared to the results from an international systematic review, our disease estimates compared fairly well (Supplementary Table S8) [17], supporting the validity of our findings.

Although diseases are not all preventable, this study showed that delaying disease onset translated to large annual HRQoL gains. After adjustment, the average disutility days averted by preventing chronic diseases equalled approximately a month annually for some conditions, ranging from 14.8 days for coronary heart disease to 36.5 days for COPD. Societal and economic impacts of such gains are expected to be substantial, and increasing efforts to prevent or delay chronic diseases is warranted. Although economically population ageing is often viewed negatively due to factors such as reduced productivity and higher healthcare cost, some argue that if people are living longer in a healthy state, longevity should be positive for the economy overall [51, 52]. The critical aspect, therefore, is to extend the average years lived in good health [53].

Relationships between modifiable risk factors such as smoking, high alcohol intake, obesity and unhealthy diet to chronic diseases onset [54, 55] and reduced quality-adjusted life expectancy [20] are well established. Reducing the population-level prevalence of these behaviours would contribute to decreasing current and future health burden. PS modelling showed that the reduction of HRQoL utility score remained after adjusting for measured confounders such as age, smoking and obesity.

Strengths of this study include a large, general population survey with a probability-based sampling framework, allowing robust health utility score estimates. Presence of chronic diseases was based on diagnosis information from linked hospital systems data rather than by participants’ self-report, providing more objective participant categorisation. A long look-back period and the high linkage consent rate provided rich patient information on chronic disease experience. Results and methodology are generalisable to other population health surveys, and the list of chronic diseases could be extended pending data availability.

This study also had a number of limitations. First, while using hospital records to identify chronic diseases is a strength, it required a hospital admission or public hospital emergency department presentation and to have the condition recorded within those events. Thus, prevalence results may still be underestimated, especially for conditions primarily treated in other settings. This may explain the lower diabetes prevalence compared to the NHS (Supplementary Table S2). Changes in the Australian Coding Standards over time may influence the likelihood of diagnoses being recorded, especially in admitted patient care settings.

HRQoL may be biased because, as a general population survey, QPHS eligibility excludes adults in institutional settings such as hospitals and aged care. This may have overestimated HRQoL in older age groups, especially those who are close to their end of life when hospital utilisation is typically highest [56]. People with lived experience, such as people living with diseases and their carers were not involved in study design. Carers could not be identified from survey questions or linked data, therefore, impacts of chronic disease to carers could not be assessed. Adults with chronic conditions were identified by data linkage and there were no survey items about specific diseases on the QPHS survey, preventing further contextualisation of lived experiences.

Diseases severity, which was unmeasured, may influence participants’ HRQoL. For some diseases, participants may have recovered and regained their pre-diagnoses health status. Diseases with rapid mortality introduce survivorship bias, potentially a greater consideration as the study was conducted during the COVID-19 pandemic. This may explain why utility scores for adults with some conditions, such as cancer, were higher than others. Similarly, the length of time someone lived with the disease could have influenced the HRQoL, which was not controlled for in the analyses.

A limitation of ATT analyses is that PSs could not be adjusted by unmeasured confounders. Including important disease onset predictors, such as genetic factors and unmeasured modifiable risk factors, could improve PS assignment accuracy and weighting. Despite age and smoking status being important disease predictors, neither age of smoking initiation or disease onset were available. Further, while model diagnostics indicated that dichotomisation of the CCI was the most appropriate form, this could have impacted PS assignment granularity. E-values calculated to understand the effect of unmeasured confounders, however, showed that unmeasured confounders with risk ratio ranging from 1.72 (CHD) to 2.35 (COPD), associated with both the disease and HRQoL, would be required to nullify the ATT. Also, while measured risk factors, such as BMI and socioeconomic status would have been useful covariates to include in PSs, they were not suitable as they were collected during the survey (post disease onset). Sensitivity analyses, however, showed marginal differences in ATTs.

Slight demographic differences were observed for those who consented to linkage (Supplementary Table S2) with males and younger age groups more likely to consent. Non-consent bias, however, is anticipated to be modest, as the overall EQ-5D-5L utility scores were comparable to previous entire-sample results [20]. Data linkage quality may impact results and while record linkage is a powerful tool used increasingly in research, systematic bias may result from incomplete or poor linkage [57]. Although a majority of participant records (> 95%) linked to at least one index file record, and quality assurance exercise are routinely conducted on the DLQ’s Master Linkage File [27, 58], the extent of false matches, both positive and negative, could not been quantified.

Conclusion

The study investigated the differences in HRQoL among Queensland adults with listed chronic diseases using linked hospital administrative records. The study showed that HRQoL was substantially lower for those with chronic diseases, resulting in more annual disutility days. While ageing populations are likely to see increases in chronic diseases prevalence, preventing or delaying the onset of those diseases by supporting healthy behaviours would likely lead to improved population-level HRQoL, and positively impact individuals, society and the economy.

Supplementary Information

Below is the link to the electronic supplementary material.

Supplementary Material 1 (5MB, pdf)

Acknowledgements

The authors would like to thank the Queensland residents who participated in the survey for their time and cooperation. The authors also would like to thank Associate Professor Nicole White (Queensland University of Technology) for statistical review, Professor Ross Andrews (Public Health Intelligence Branch, Queensland Health) for providing feedback on the manuscript, Douglas Lincoln and the Surveillance team (Public Health Intelligence Branch, Queensland Health) for implementing and processing the Queensland Preventive Health Survey data, Queensland Government Statistician’s Office and CATI team for conducting the telephone interviews, Data Linkage Queensland, Queensland Health, for conducting the record linkage and providing the ICD-10-AM codes for CCI calculation, and various data custodians for approving the access and provision of the data.

Author contributions

T.E. conceptualised the study, conducted statistical analyses, and prepared the initial draft. T.E. and S.C. designed the study and conducted literature reviews. S.C. obtained the funding, provided critical input and review for intellectual content of the manuscript and supervised the work. All authors revised the manuscript and approved the final version. The corresponding author attest that all listed authors meet authorship criteria and that no others meeting the criteria have been omitted.

Funding

The study was government funded through Queensland Health as part of ongoing population health surveillance activities. The funder had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript content; and decision to submit the manuscript for publication.

Data availability

The data used for this study, the Queensland Preventive Health Survey, are the property of Queensland Health. Data may be shared and released through an application process which requires Queensland Health data custodian approval in accordance with the Queensland Public Health Act 2005 and the Statistical Returns Act 1896.

Declarations

Ethics approval and consent to participate

The study was approved by a Human Research Ethics Committee (AM/2021/QWMS/13623/AM14, AM/2022/QWMS/13623/AM16). All the participating patients provided verbal informed consent.

Consent for publication

Not applicable.

Competing interests

The authors declare that they have no competing interests.

Footnotes

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

References

  • 1.Australian Institute of Health and Welfare. Health system spending on disease and injury in Australia 2022–23. (2024). https://www.aihw.gov.au/reports/health-welfare-expenditure/health-system-spending-on-disease-and-injury-aus/contents/about. Accessed 29 Nov 2024.
  • 2.Australian Institute of Health and Welfare. Australian Burden of Disease Study 2024. (2024). https://www.aihw.gov.au/reports/burden-of-disease/australian-burden-of-disease-study-2024/contents/about. Accessed 12 Dec 2024.
  • 3.Schofield, D., Shrestha, R. N., Cunich, M. M., et al. (2016). Economic costs of chronic disease through lost productive life years (PLYs) among Australians aged 45–64 years from 2015 to 2030: Results from a microsimulation model. British Medical Journal Open,6, e011151. 10.1136/bmjopen-2016-011151 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 4.Frølich, A., Ghith, N., Schiøtz, M., et al. (2019). Multimorbidity, healthcare utilization and socioeconomic status: A register-based study in Denmark. PLoS ONE,14, e0214183. 10.1371/journal.pone.0214183 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5.McPhail, S. (2016). Multimorbidity in chronic disease: Impact on health care resources and costs. Rocky Mountain Health Reports,9, 143–156. 10.2147/rmhp.s97248 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 6.Soley-Bori, M., Ashworth, M., Bisquera, A., et al. (2021). Impact of multimorbidity on healthcare costs and utilisation: A systematic review of the UK literature. British Journal of General Practice,71, e39-46. 10.3399/bjgp20x713897 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 7.Wang, L., Si, L., Cocker, F., et al. (2018). A systematic review of cost-of-illness studies of multimorbidity. Applied Health Economics and Health Policy,16, 15–29. 10.1007/s40258-017-0346-6 [DOI] [PubMed] [Google Scholar]
  • 8.The EuroQol Group. (1990). EuroQol—A new facility for the measurement of health-related quality of life. Health Policy,16, 199–208. 10.1016/0168-8510(90)90421-9 [DOI] [PubMed] [Google Scholar]
  • 9.National Institute for Health and Care Excellence. NICE technology appraisal and highly specialised technologies guidance: the manual. Manchester: National Institute for Health and Care Excellence 2025.
  • 10.Schron, E., Friedmann, E., & Thomas, S. A. (2014). Does health‐related quality of life predict hospitalization or mortality in patients with atrial fibrillation? Cardiovascular Electrophysiology,25, 23–28. 10.1111/jce.12266 [DOI] [PubMed] [Google Scholar]
  • 11.Theunissen, L. J. H. J., Van De Pol, J. A. A., Van Steenbergen, G. J., et al. (2023). The prognostic value of quality of life in atrial fibrillation on patient value. Health and Quality of Life Outcomes. 10.1186/s12955-023-02112-2 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 12.Baroutidou, A., Kartas, A., Papazoglou, A. S., et al. (2022). Association of health status metrics with clinical outcomes in patients with adult congenital heart disease and atrial arrhythmias. Journal of Clinical Medicine,11, 6181. 10.3390/jcm11206181 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13.Melis, F., Kvamme, T. L., Tjagvad, C., et al. (2025). Health-related quality of life measures as predictors for recurrent hospitalization and mortality among patients in heroin-assisted treatment. Quality of Life Research. 10.1007/s11136-025-04019-5 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 14.Infectious Disease Clinical Research Program HIV Working Group, Emuren, L., Welles, S., et al. (2018). Lower health-related quality of life predicts all-cause hospitalization among HIV-infected individuals. Health and Quality of Life Outcomes. 10.1186/s12955-018-0931-x [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15.Blok, I. M., Van Riel, A. C. M. J., Schuuring, M. J., et al. (2015). Decrease in quality of life predicts mortality in adult patients with pulmonary arterial hypertension due to congenital heart disease. Netherlands Heart Journal,23, 278–284. 10.1007/s12471-015-0666-9 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 16.Singh, J., Nelson, D., Fink, H., et al. (2005). Health-related quality of life predicts future health care utilization and mortality in veterans with self-reported physician-diagnosed arthritis: The veterans arthritis quality of life study. Seminars in Arthritis and Rheumatism,34, 755–765. 10.1016/j.semarthrit.2004.08.001 [DOI] [PubMed] [Google Scholar]
  • 17.Van Wilder, L., Rammant, E., Clays, E., et al. (2019). A comprehensive catalogue of EQ-5D scores in chronic disease: Results of a systematic review. Quality of Life Research,28, 3153–3161. 10.1007/s11136-019-02300-y [DOI] [PubMed] [Google Scholar]
  • 18.Australian Commission on Safety and Quality in Health Care. PROMs lists. (2025). https://www.safetyandquality.gov.au/our-work/indicators-measurement-and-reporting/patient-reported-outcome-measures/proms-lists#generic-proms-list. Accessed 20 July 2025.
  • 19.Clemens, S., Begum, N., Harper, C., et al. (2014). A comparison of EQ-5D-3L population norms in Queensland, Australia, estimated using utility value sets from Australia, the UK and USA. Quality of Life Research,23, 2375–2381. 10.1007/s11136-014-0676-x [DOI] [PubMed] [Google Scholar]
  • 20.Endo, T., Lee, X. J., & Clemens, S. L. (2024). EQ-5D-5L population norms and quality-adjusted life expectancy by sociodemographic characteristics and modifiable risk factors for adults in Queensland, Australia. Value in Health,27, 633–641. 10.1016/j.jval.2024.02.007 [DOI] [PubMed] [Google Scholar]
  • 21.Oppe, M., Devlin, N. J., Van Hout, B., et al. (2014). A program of methodological research to arrive at the new international EQ-5D-5L valuation protocol. Value in Health,17, 445–453. 10.1016/j.jval.2014.04.002 [DOI] [PubMed] [Google Scholar]
  • 22.Queensland Health. Queensland preventive health survey. 2024. www.health.qld.gov.au/phsurvey. Accessed 7 Nov 2024.
  • 23.Norman, R., Mulhern, B., Lancsar, E., et al. (2023). The use of a discrete choice experiment including both duration and dead for the development of an EQ-5D-5L value set for Australia. PharmacoEconomics,41, 427–438. 10.1007/s40273-023-01243-0 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 24.Australian Bureau of Statistics. Socio-Economic Indexes for Areas (SEIFA), Australia. Socio-Economic Indexes for Areas (SEIFA), Australia. (2023). https://www.abs.gov.au/statistics/people/people-and-communities/socio-economic-indexes-areas-seifa-australia/latest-release. Accessed 8 July 2023.
  • 25.Australian Bureau of Statistics. Remoteness Areas. ASGS Edition 3 Remoteness Areas, Reference period: July 2021-June 2026. (2023). https://www.abs.gov.au/statistics/standards/australian-statistical-geography-standard-asgs-edition-3/jul2021-jun2026/remoteness-structure/remoteness-areas. Accessed 8 July 2023.
  • 26.Flack, F., & Smith, M. (2019). The population health research network—Population data centre profile. International Journal of Population Data Science. 10.23889/ijpds.v4i2.1130 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 27.Bunting, D., Endo, T., Watt, K., et al. (2023). Mastering linked datasets: The future of emergency health care research. Prehospital Emergency Care,27, 1031–1040. 10.1080/10903127.2022.2108179 [DOI] [PubMed] [Google Scholar]
  • 28.Charlson, M. E., Pompei, P., Ales, K. L., et al. (1987). A new method of classifying prognostic comorbidity in longitudinal studies: Development and validation. Journal of Chronic Diseases,40, 373–383. 10.1016/0021-9681(87)90171-8 [DOI] [PubMed] [Google Scholar]
  • 29.Nahhas, R. W. (2025). Introduction to regression methods for public health using R. Chapman & Hall CRC. [Google Scholar]
  • 30.Lumley, T. (2004). Analysis of complex survey samples. Journal of Statistical Software. 10.18637/jss.v009.i08 [Google Scholar]
  • 31.Mujica-Mota, R. E., Roberts, M., Abel, G., et al. (2015). Common patterns of morbidity and multi-morbidity and their impact on health-related quality of life: Evidence from a national survey. Quality of Life Research,24, 909–918. 10.1007/s11136-014-0820-7 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 32.Zhao, Q.-Y., Luo, J.-C., Su, Y., et al. (2021). Propensity score matching with R: Conventional methods and new features. Annals of Translational Medicine,9, 812–822. 10.21037/atm-20-3998 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 33.Franklin, M., Peasgood, T., & Tennant, P. W. G. (2025). Depicting patient-reported outcome measures within directed acyclic graphs: Practice and implications for causal reasoning. Quality of Life Research. 10.1007/s11136-025-04007-9 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 34.Wild, C., Weiderpass, E., Stewart, B. W., et al. (Eds.). (2020). World cancer report: Cancer research for cancer prevention. International Agency for Research on Cancer. [PubMed] [Google Scholar]
  • 35.Barrett, M. (2024). ggdag: Analyze and create elegant directed acyclic graphs. The R Foundation. [Google Scholar]
  • 36.Textor, J., Van Der Zander, B., Gilthorpe, M. S., et al. (2017). Robust causal inference using directed acyclic graphs: The R package ‘dagitty.’ International Journal of Epidemiology. 10.1093/ije/dyw341 [DOI] [PubMed] [Google Scholar]
  • 37.Brookhart, M. A., Schneeweiss, S., Rothman, K. J., et al. (2006). Variable selection for propensity score models. American Journal of Epidemiology,163, 1149–1156. 10.1093/aje/kwj149 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 38.Strandberg, A. Y., Strandberg, T. E., Pitkälä, K., et al. (2008). The effect of smoking in midlife on health-related quality of life in old age: A 26-year prospective study. Archives of Internal Medicine,168, 1968. 10.1001/archinte.168.18.1968 [DOI] [PubMed] [Google Scholar]
  • 39.Greifer, N. (2017). WeightIt: Weighting for covariate balance in observational studies. The R Foundation. [Google Scholar]
  • 40.VanderWeele, T. J., & Ding, P. (2017). Sensitivity analysis in observational research: Introducing the E-Value. Annals of Internal Medicine,167, 268–274. 10.7326/m16-2607 [DOI] [PubMed] [Google Scholar]
  • 41.Mathur, M. B., Smith, L. H., Ding, P., & Van der Weele, T. J. EValue: Sensitivity analyses for unmeasured confounding and other biases in observational studies and meta-analyses.
  • 42.R Core Team. (2025). R: A language and environment for statistical computing. R Foundation for Statistical Computing. [Google Scholar]
  • 43.Austin, P. C. (2009). Balance diagnostics for comparing the distribution of baseline covariates between treatment groups in propensity‐score matched samples. Statistics in Medicine,28, 3083–3107. 10.1002/sim.3697 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 44.Australian Bureau of Statistics. Comparing ABS long-term health conditions data sources. (2022). https://www.abs.gov.au/statistics/detailed-methodology-information/information-papers/comparing-abs-long-term-health-conditions-data-sources. Accessed 13 June 2023.
  • 45.Wu, C.-S., Hsu, L.-Y., Shen, C.-Y., et al. (2025). Validation of self-reported medical condition in the Taiwan Biobank. Journal of Epidemiology,35, 141–146. 10.2188/jea.JE20240110 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 46.Oksanen, T., Kivimäki, M., Pentti, J., et al. (2010). Self-report as an indicator of incident disease. Annals of Epidemiology,20, 547–554. 10.1016/j.annepidem.2010.03.017 [DOI] [PubMed] [Google Scholar]
  • 47.Cigolle, C. T., Nagel, C. L., Blaum, C. S., et al. (2016). Inconsistency in the self-report of chronic diseases in panel surveys: Developing an adjudication method for the health and retirement study. GERONB.10.1093/geronb/gbw063 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 48.Singh, J. A. (2009). Accuracy of Veterans Affairs databases for diagnoses of chronic diseases. Preventing Chronic Disease,6, A126. [PMC free article] [PubMed] [Google Scholar]
  • 49.Redwood, L., Currow, D., Kochovska, S., et al. (2023). Australian population norms for health-related quality of life measured using the EQ-5D–5L, and relationships with sociodemographic characteristics. Quality of Life Research. 10.1007/s11136-023-03558-z [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 50.McCaffrey, N., Kaambwa, B., Currow, D. C., et al. (2016). Health-related quality of life measured using the EQ-5D–5L: South Australian population norms. Health and Quality of Life Outcomes,14, Article 133. 10.1186/s12955-016-0537-0 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 51.Scott, A. J. (2021). The longevity economy. The Lancet Healthy Longevity,2, e828–e835. 10.1016/S2666-7568(21)00250-6 [DOI] [PubMed] [Google Scholar]
  • 52.Cylus, J., & Al Tayara, L. (2021). Health, an ageing labour force, and the economy: Does health moderate the relationship between population age-structure and economic growth? Social Science & Medicine,287, 114353. 10.1016/j.socscimed.2021.114353 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 53.European Observatory on Health Systems and Policies. (2020). How will population ageing affect health expenditure trends in Australia and what are the implications if people age in good health?: New evidence for the Western Pacific Region: Australia. World Health Organization. [Google Scholar]
  • 54.Brauer, M., Roth, G. A., Aravkin, A. Y., et al. (2024). Global burden and strength of evidence for 88 risk factors in 204 countries and 811 subnational locations, 1990–2021: A systematic analysis for the Global Burden of Disease Study 2021. Lancet,403, 2162–2203. 10.1016/S0140-6736(24)00933-4 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 55.Crosland, P., Ananthapavan, J., Davison, J., et al. (2019). The health burden of preventable disease in Australia: A systematic review. Australian and New Zealand Journal of Public Health,43, 163–170. 10.1111/1753-6405.12882 [DOI] [PubMed] [Google Scholar]
  • 56.Rosenwax, L. K., McNamara, B. A., Murray, K., et al. (2011). Hospital and emergency department use in the last year of life: A baseline for future modifications to end‐of‐life care. Medical Journal of Australia,194, 570–573. 10.5694/j.1326-5377.2011.tb03106.x [DOI] [PubMed] [Google Scholar]
  • 57.Bohensky, M. A., Jolley, D., Sundararajan, V., et al. (2010). Data linkage: A powerful research tool with potential problems. BMC Health Services Research. 10.1186/1472-6963-10-346 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 58.Queensland Health. Queensland Data Linkage Framework. (2021).

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplementary Material 1 (5MB, pdf)

Data Availability Statement

The data used for this study, the Queensland Preventive Health Survey, are the property of Queensland Health. Data may be shared and released through an application process which requires Queensland Health data custodian approval in accordance with the Queensland Public Health Act 2005 and the Statistical Returns Act 1896.

Articles from Quality of Life Research are provided here courtesy of Springer

RESOURCES