Comorbid Chronic Diseases and Health-related Quality of Life in the Obese Population: a Longitudinal Analysis of a Nationally Representative Household Survey in Australia

Objective This study aims to examine the relationship between nine comorbid chronic conditions and HRQoL separately, along with the number of chronic diseases among the Australian obese population. Methods Data for this study were sourced from three waves (waves 9, 13 and 17) of the Household, Income and Labour Dynamics in Australia (HILDA) survey. The paper studies 9,444 person-year observations from 5,524 individuals over the years 2009, 2013, and 2017. The outcome variable of HRQoL was measured through the 36-Item Short Form Health Survey (SF-36), and the main variables of interest were nine chronic conditions and the number of chronic diseases. Generalized estimating equations (GEE) were used to test the association between comorbid chronic diseases and HRQoL. This study found a negative relationship between the number of comorbid chronic conditions and sub-scale, summary measures, and health utility index of the SF-36. Obese adults with 1, 2, 3, and 3+ comorbid chronic diseases scored lower points on the SF-36 physical component summary (b = -2.83, b = -7.37, b = -11.15, b = -14.29, respectively), mental component summary (b = -1.46, b = -2.34, b = -3.66, and b = -6.34, respectively), and in the short-form six-dimension utility index (SF-6D) scale (b = -0.030, b = -0.063, b = -0.099, and b = -0.138, respectively) compared to obese peers without comorbid chronic diseases. The number of chronic conditions was associated with reductions in the score of all eight dimensions of the SF-36. Obese people with any of the nine studied comorbid chronic diseases (heart disease, circulatory disease, hypertension, type 1 diabetes, type 2 diabetes, asthma, bronchitis, arthritis, and cancer) were with lower HRQoL compared to peers without that particular comorbid chronic disease. measures of the SF-36. The ndings, therefore, call for improved holistic management of obesity and interventions to reduce obesity-related comorbidities to improve HRQoL of obese Australian. 11 (b = -11.15), and 14 (b = -14.29) points/units lower on the PCS indicator, and 1 (b = -1.46), 2 (b = -2.34), 4 (b = -3.66), and 6 (b = -6.34) units lower on the MCS indicator, respectively, compared with obese people without comorbid chronic diseases. Models 1 and 2 also report the effects of individual chronic diseases on both PCS and MCS indicators. The result showed that obese people with any of the nine chronic diseases had signicantly lower scores on both PCS and MCS indicators. For example, the effect of having cancer in obese people on both PCS (b = -4.08) and MCS (b = -2.27) were lower than counterparts without cancer.


Introduction
Overweight and obesity are rapidly growing public health problems affecting many countries worldwide. Among the adult population, in 2016, more than 1.9 billion adults were overweight globally, of which 650 million were obese [1]. In Australia, obesity has increased from 18.5-27.9% between 1995 and 2015 [2]. Two-thirds (67%, 12.5 million) of Australian adults were either overweight or obese in 2017-18 [3], and adult obesity prevalence was projected to increase from 19% in 1995 to 35% by 2025 [2].
Quality of life broadly refers to the extent to which an individual can function successfully in daily life and their perceived well-being across physical, emotional, and social structures [4,5]. Obesity is associated with increased comorbidity, mortality and reduced health-related quality of life (HRQoL) [6][7][8][9][10]. The health burden among individuals with raised body mass index (BMI) is becoming concerning, especially in those with co-occurring chronic conditions [11,12]. The relationship between BMI and HRQoL has been investigated in several population-based studies and have con rmed a negative association between BMI and self-perceived quality of life, with a higher risk of poorer HRQoL in overweight and obese persons [6,[13][14][15][16][17]. Further, obese persons report pain which has been considered the most signi cant impairment to HRQoL [13,15,18]. Moreover, overweight and obese people experience higher psychological distress, which is another considerable impairment in their HRQoL [15,19]. Several studies across diverse geographical locations have reported that comorbid chronic diseases are associated with poor quality of life. For instance, earlier studies found that overweight or obese individuals often report physical, mental, and social relationship problems [20][21][22][23][24]. Of those reporting poor HRQoL, the highest burden was found among those with multiple comorbid chronic conditions [25]. Other empirical studies have reported poor HRQoL among persons with comorbid or multimorbid diseases [26][27][28].
HRQoL among obese individuals is understudied, with only a few empirical studies focused on establishing the association between comorbidities and quality of life. Two studies have reported that overweight and obesity were associated with low or poor HRQoL [16,29]. However, a recent study has not found a statistically signi cant association between HRQoL (measured by SF-6D) and comorbid chronic diseases in the Australian general population [7]. The discrepancy in the relationship between comorbid conditions and HRQoL in the existing literature warrants further investigation to draw robust conclusions on the longitudinal relationship between comorbidity and HRQoL in the obese population. Therefore, this paper aims to examine the associations between comorbid chronic diseases and HRQoL among the Australian obese population. This is a novel study that provides a signi cant opportunity to advance the understanding of the relationship between nine comorbid chronic diseases and HRQoL in the obese population separately, along with the number of comorbidities. The study will provide insights into the need for measures to prevent overweight and obesity, manage those with comorbid conditions, and prevent further development of comorbidities among overweight and obesity with a view to improving HRQoL.

Methodology Data Source and Sample selection
This study's data were sourced from the Household, Income and Labour Dynamics in Australia (HILDA) survey, a nationally representative longitudinal study of the Australian population. The survey collects information annually on many aspects of life, such as wealth, labour market outcomes, household and family relationships, fertility, health and education. The survey was started in 2001, and a multistage sampling approach was used to select an initial sample of households. At rst, 488 Census Collection Districts (CD) were sampled with a probability proportional to size sampling technique; each consists of 200-250 households approximately across Australia. Secondly, from each of the CDs, a sample of 22-34 dwellings was selected randomly. Finally, up to three households from each dwelling were selected that results in the selection of a total of 12,252 households. Individuals aged 15 years or older residing in each household were included in the sample. The sample was expanded over time by including any child born or adopted by groups of respondents or by any new household member resulting from adjustments of the originating households' composition. Therefore, the survey follows the lives of more than 17,000 Australian adults annually.
This study utilized three waves of data: wave 9 (2009), 13 (2013) and 17 (2017) from the HILDA survey, spanning a period of nine years. The main reason for selecting these three waves is that data on comorbid chronic conditions were available only in these waves. This study restricted the sample to only obese adults aged 15 years or over. Missing observations on the outcome (dimensions of HRQoL) and main variables of interest (chronic diseases) were excluded from subsample analyses. After adjusting the inclusion and exclusion criteria, the nal analytic sample consists of 9,444 person-year observations from 5,524 unique respondents.

Outcome variable
The outcome of interest in the present analysis is the health-related quality of life (HRQoL). HRQoL was measured through the RAND 36-Item Short Form Survey Instrument (SF-36). The SF-36 health survey is made up of 36 questions that cover eight dimensions: physical functioning (PF), role physical (RP); bodily pain (BP), general health (GH), vitality (VT), social functioning (SF); role emotional (RE); and mental health (MH). For example, the physical functioning dimension was assessed by ten questions, and each question has three levels (Yes, limited a lot; Yes, limited a little; and No, not limited at all). These levels were scaled as 1, 2, and 3 and thus, summed values lie between 10 to 30. This computed value was further transformed into a 0-100 scale. Similarly, each of the eight dimensions' score scale ranged from 0 to 100, wherein 0 represents the worst and 100 represents the best health status. It is important to note that SF- respectively, with higher scores indicating better QoL [5].
Another instrument that is widely used in the economic evaluation as a measure of HRQoL is SF-6D. The SF-6D utility index can be derived from the SF-36 score and places health states in a scale that ranges from 0 to 1. The value 1 indicates full health (all the eight dimensions at the best level), and 0 shows the worst health (equivalent to death).

Exposure variables
In the present analyses, comorbid chronic diseases are considered as the main exposure variables. This study assessed nine self-reported chronic diseases: heart disease, circulatory disease, hypertension, type 1 diabetes, type 2 diabetes, asthma, bronchitis, arthritis, and cancer. The HILDA survey collects information on an individual's chronic disease status by asking the question: 'have you ever been told by a medical practitioner that you have been diagnosed with a serious illness or medical conditions'. The responses were taken in binary form: an answer of zero means no, and one means yes. The variable number of comorbid chronic disease is constructed by summing up the nine studied chronic diseases. The variable was categorized into ve: 0 (having no chronic condition), 1 (having only one chronic condition), 2 (having two of the studied chronic diseases), 3 (having three of the studied chronic diseases) and 3+ (having more than three of the studied chronic diseases).

Other Covariates
A set of socio-demographic and behavioural characteristics were included in the study as potential confounders. All the explanatory variables were categorized using dummies. Socio-demographic factors include age (15-25, 26- Islander); and location (major city, regional city [inner and outer regional], remote areas [remote or very remote areas]).
Behavioural characteristics include smoking status (never smoked, ex-smoker, current smoker); alcohol consumption (never drink, ex-drinker, only rarely to 3 days per week, 3+ days per week); and physical activity that lasts at least 30 minutes (not at all to <1 per week, 1-3 times per week, ≥4 times per week).

Statistical analysis
The authors constructed an unbalanced longitudinal data set consisting of 9,444 person-year observations of 5,524 unique participants by linking de-identi ed individuals' records wherein respondent information appeared more than once (up to three times). The current analyses report the pooled descriptive statistics as mean (SD) for continuous variables and percentages with 95% con dence intervals (CIs) for categorical variables.
This study tted multivariate regression models to explore the relationship between comorbid chronic diseases and HRQoL. The regression models take the following form: In equation 1, represents the summary measures, health utility index, and a particular dimension of SF-36 representing respondents' QoL. CD is the key variables of interests that capture the presence of comorbid chronic diseases in the respondents, X is a vector of control variables, is the error term, and subscripts i refer to individual and t indicates periods.
This study constructed ten different models, de ned by the primary variables of interest: number of comorbid chronic diseases, solely heart disease, circulatory disease exclusively, solely hypertension, solely type 1 diabetes, solely type 2 diabetes, solely asthma, solely bronchitis, solely arthritis, and solely cancer. The reference category was always the absence of the comorbid chronic diseases. All models were adjusted for age, gender, civil status, education, equivalized household income, labour force status, race, place of living, smoking status, alcohol consumption, and physical activity.
This study deployed the Generalized Estimating Equation (GEE) to estimate the effects of comorbid chronic diseases on HRQoL. A signi cant advantage of using the GEE technique is that it provides unbiased estimates of populationaveraged regression coe cients when the data's correlation structure is misspeci ed. A p-value of less than 0.05 was considered statistically signi cant, and the regression results were reported for three levels of P <0.001, <0.01, and <0.05. All analyses were conducted using STATA version 16. Table 1 shows the pooled summary statistics for the 9,444 Australian adults. The mean score for the eight domains of the SF-36 were 76. 35  The mean component summary measures (PCS and MCS) and health utility index (SF-6D) derived from the SF-36 were 45.78 ± 11.38, 47.72 ± 11.26, and 0.73 ± 0.13 (mean ± SD), respectively. Among the study sample, over one quarter (27%) have at least one chronic condition, followed by two (15%), three (7%), and more than three (4%) comorbid chronic diseases. The most common comorbid chronic disease among the obese adults was hypertension (29%), followed by arthritis (24%), asthma (14%), type 2 diabetes (10%), and heart disease (6%).

Results
The results also reveal that almost one-fourth of the participants were over sixty years (25%), over half were female (55%), and married (53%). Of the total, 18% had university quali cations, 61% were employed, 96% were nonindigenous, 61% lived in major cities, 19% were current smoker, 22% drunk over three days a week, and over one third (36%) do not perform physical activities.   On the SF-6D scale, obese adults with 1, 2, 3 and more than 3 comorbid chronic diseases scored 3 (b = -0.03), 6 (b = -0.063), 10 (b = -0.099), and 14 (b = -0.138) percentage points lower, respectively, compared with obese peers who do not have any chronic disease (model 3). Similarly, the results also showed that obese people having any type of the studied chronic diseases (heart disease, circulatory disease, hypertension, type 1 diabetes, type 2 diabetes, asthma, bronchitis, arthritis, and cancer) scored lower on the SF-6D scale compared with obese people without that particular chronic disease. For example, obese adults with heart disease scored 6 (b = -0.058) percentage points lower on the SF-6D scale than their counterparts without heart disease.

Discussion
This study is the rst to assess the relationships between comorbid chronic diseases and HRQoL among the obese population in Australia. The current study further highlighted the interplay of nine chronic diseases in the previously found association between obesity and HRQoL [6,14].
The study results showed that the PCS, MCS, and SF-6D scores in obese people reduced sharply with an increasing number of chronic diseases. The negative association between the rising number of comorbid chronic conditions and overall HRQoL is similar to previous studies that reported a signi cant reduction in HRQoL among persons having multimorbidities [9,26,[30][31][32][33][34][35]. The results showed that obese individuals having any of the nine studied chronic diseases were associated with reduced PCS, MCS, and SF-6D scores. Results from previous empirical studies showed that an increase of the number of comorbidities in an individual or patient was associated with lower HRQoL [23,25,27,28], which is consistent with the current study ndings. Also, earlier studies have reported a statistically signi cant negative association between a higher number of comorbid chronic conditions and worse scores on PCS and MCS in obese people [30,36]. Further, this current study revealed that a higher number of chronic diseases was associated with a reduction in scores in all eight dimensions of the SF-36. Similar ndings have been highlighted elsewhere that studied the association between comorbid diseases and HRQoL [37][38][39].
Although consistent ndings were revealed, some of the earlier studies used a different survey instrument other than the SF-36 to measure HRQoL [31,32,34,35]. Therefore, there is a need for careful interpretation of the current study ndings compared with the previous literature. The current results indicate that the burden posed by comorbid chronic diseases in an individual irrespective of the underlying condition, and the association could be attributed to several plausible factors. First, the observed lower HRQoL could be due to the synergistic effects that coexist among chronic diseases, resulting from one condition hampering a patient's ability to adhere to treatment for another [40]. An additional reason could be that obese individuals are at a greater risk of developing several chronic cardiovascular, muscular-skeletal, and metabolic comorbid conditions [41,42]. As a result, these conditions in the obese population could have negated their quality of life due to the increasing deteriorating effects of multiple chronic diseases [9].
Besides, comorbidities may profoundly impact patients' ability to manage their self-care and may pose signi cant barriers to lifestyle changes and regimen adherence [38]. Further, the present study results could have been in uenced by comorbid mental health disorders that are most prevalent among persons suffering from chronic diseases.
The present study has several strengths. Firstly, this is the rst longitudinal study that reports the relationships between This study did not require ethical approval as the analysis used only de-identi ed existing unit record data from the HILDA survey. However, the authors completed and signed the Con dentiality Deed Poll and sent it to NCLD (ncldresearch@dss.gov.au) and ADA (ada@anu.edu.au) before the data applications' approval. Therefore, the datasets analysed and/or generated during the current study are subject to the signed con dentiality deed.

Con ict of interest
The authors declare that they have no con icts of interest.

Availability of data and materials
The data used for the study was collected from the Melbourne Institute of Applied Economic and Social Research.
There are some restrictions on this data and it is not available to the public. Those interested in accessing this data should contact the Melbourne Institute of Applied Economic and Social Research, The University of Melbourne, VIC 3010, Australia.

Funding
This research did not receive any speci c grant from funding agencies in the public, commercial or not-for-pro t sectors. Mean SF-6D score by age and gender